Jev-Style-2B-Decision-v3: release

.gitattributes

@@ -33,3 +33,8 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
 *.zip filter=lfs diff=lfs merge=lfs -text
 *.zst filter=lfs diff=lfs merge=lfs -text
 *tfevents* filter=lfs diff=lfs merge=lfs -text
+8bit/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+bf16/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+figures/banner.png filter=lfs diff=lfs merge=lfs -text
+figures/jevbench.png filter=lfs diff=lfs merge=lfs -text
+figures/zeroshot.png filter=lfs diff=lfs merge=lfs -text

8bit/chat_template.jinja

@@ -0,0 +1,154 @@
+{%- set image_count = namespace(value=0) %}
+{%- set video_count = namespace(value=0) %}
+{%- macro render_content(content, do_vision_count, is_system_content=false) %}
+    {%- if content is string %}
+        {{- content }}
+    {%- elif content is iterable and content is not mapping %}
+        {%- for item in content %}
+            {%- if 'image' in item or 'image_url' in item or item.type == 'image' %}
+                {%- if is_system_content %}
+                    {{- raise_exception('System message cannot contain images.') }}
+                {%- endif %}
+                {%- if do_vision_count %}
+                    {%- set image_count.value = image_count.value + 1 %}
+                {%- endif %}
+                {%- if add_vision_id %}
+                    {{- 'Picture ' ~ image_count.value ~ ': ' }}
+                {%- endif %}
+                {{- '<|vision_start|><|image_pad|><|vision_end|>' }}
+            {%- elif 'video' in item or item.type == 'video' %}
+                {%- if is_system_content %}
+                    {{- raise_exception('System message cannot contain videos.') }}
+                {%- endif %}
+                {%- if do_vision_count %}
+                    {%- set video_count.value = video_count.value + 1 %}
+                {%- endif %}
+                {%- if add_vision_id %}
+                    {{- 'Video ' ~ video_count.value ~ ': ' }}
+                {%- endif %}
+                {{- '<|vision_start|><|video_pad|><|vision_end|>' }}
+            {%- elif 'text' in item %}
+                {{- item.text }}
+            {%- else %}
+                {{- raise_exception('Unexpected item type in content.') }}
+            {%- endif %}
+        {%- endfor %}
+    {%- elif content is none or content is undefined %}
+        {{- '' }}
+    {%- else %}
+        {{- raise_exception('Unexpected content type.') }}
+    {%- endif %}
+{%- endmacro %}
+{%- if not messages %}
+    {{- raise_exception('No messages provided.') }}
+{%- endif %}
+{%- if tools and tools is iterable and tools is not mapping %}
+    {{- '<|im_start|>system\n' }}
+    {{- "# Tools\n\nYou have access to the following functions:\n\n<tools>" }}
+    {%- for tool in tools %}
+        {{- "\n" }}
+        {{- tool | tojson }}
+    {%- endfor %}
+    {{- "\n</tools>" }}
+    {{- '\n\nIf you choose to call a function ONLY reply in the following format with NO suffix:\n\n<tool_call>\n<function=example_function_name>\n<parameter=example_parameter_1>\nvalue_1\n</parameter>\n<parameter=example_parameter_2>\nThis is the value for the second parameter\nthat can span\nmultiple lines\n</parameter>\n</function>\n</tool_call>\n\n<IMPORTANT>\nReminder:\n- Function calls MUST follow the specified format: an inner <function=...></function> block must be nested within <tool_call></tool_call> XML tags\n- Required parameters MUST be specified\n- You may provide optional reasoning for your function call in natural language BEFORE the function call, but NOT after\n- If there is no function call available, answer the question like normal with your current knowledge and do not tell the user about function calls\n</IMPORTANT>' }}
+    {%- if messages[0].role == 'system' %}
+        {%- set content = render_content(messages[0].content, false, true)|trim %}
+        {%- if content %}
+            {{- '\n\n' + content }}
+        {%- endif %}
+    {%- endif %}
+    {{- '<|im_end|>\n' }}
+{%- else %}
+    {%- if messages[0].role == 'system' %}
+        {%- set content = render_content(messages[0].content, false, true)|trim %}
+        {{- '<|im_start|>system\n' + content + '<|im_end|>\n' }}
+    {%- endif %}
+{%- endif %}
+{%- set ns = namespace(multi_step_tool=true, last_query_index=messages|length - 1) %}
+{%- for message in messages[::-1] %}
+    {%- set index = (messages|length - 1) - loop.index0 %}
+    {%- if ns.multi_step_tool and message.role == "user" %}
+        {%- set content = render_content(message.content, false)|trim %}
+        {%- if not(content.startswith('<tool_response>') and content.endswith('</tool_response>')) %}
+            {%- set ns.multi_step_tool = false %}
+            {%- set ns.last_query_index = index %}
+        {%- endif %}
+    {%- endif %}
+{%- endfor %}
+{%- if ns.multi_step_tool %}
+    {{- raise_exception('No user query found in messages.') }}
+{%- endif %}
+{%- for message in messages %}
+    {%- set content = render_content(message.content, true)|trim %}
+    {%- if message.role == "system" %}
+        {%- if not loop.first %}
+            {{- raise_exception('System message must be at the beginning.') }}
+        {%- endif %}
+    {%- elif message.role == "user" %}
+        {{- '<|im_start|>' + message.role + '\n' + content + '<|im_end|>' + '\n' }}
+    {%- elif message.role == "assistant" %}
+        {%- set reasoning_content = '' %}
+        {%- if message.reasoning_content is string %}
+            {%- set reasoning_content = message.reasoning_content %}
+        {%- else %}
+            {%- if '</think>' in content %}
+                {%- set reasoning_content = content.split('</think>')[0].rstrip('\n').split('<think>')[-1].lstrip('\n') %}
+                {%- set content = content.split('</think>')[-1].lstrip('\n') %}
+            {%- endif %}
+        {%- endif %}
+        {%- set reasoning_content = reasoning_content|trim %}
+        {%- if loop.index0 > ns.last_query_index %}
+            {{- '<|im_start|>' + message.role + '\n<think>\n' + reasoning_content + '\n</think>\n\n' + content }}
+        {%- else %}
+            {{- '<|im_start|>' + message.role + '\n' + content }}
+        {%- endif %}
+        {%- if message.tool_calls and message.tool_calls is iterable and message.tool_calls is not mapping %}
+            {%- for tool_call in message.tool_calls %}
+                {%- if tool_call.function is defined %}
+                    {%- set tool_call = tool_call.function %}
+                {%- endif %}
+                {%- if loop.first %}
+                    {%- if content|trim %}
+                        {{- '\n\n<tool_call>\n<function=' + tool_call.name + '>\n' }}
+                    {%- else %}
+                        {{- '<tool_call>\n<function=' + tool_call.name + '>\n' }}
+                    {%- endif %}
+                {%- else %}
+                    {{- '\n<tool_call>\n<function=' + tool_call.name + '>\n' }}
+                {%- endif %}
+                {%- if tool_call.arguments is defined %}
+                    {%- for args_name, args_value in tool_call.arguments|items %}
+                        {{- '<parameter=' + args_name + '>\n' }}
+                        {%- set args_value = args_value | tojson | safe if args_value is mapping or (args_value is sequence and args_value is not string) else args_value | string %}
+                        {{- args_value }}
+                        {{- '\n</parameter>\n' }}
+                    {%- endfor %}
+                {%- endif %}
+                {{- '</function>\n</tool_call>' }}
+            {%- endfor %}
+        {%- endif %}
+        {{- '<|im_end|>\n' }}
+    {%- elif message.role == "tool" %}
+        {%- if loop.previtem and loop.previtem.role != "tool" %}
+            {{- '<|im_start|>user' }}
+        {%- endif %}
+        {{- '\n<tool_response>\n' }}
+        {{- content }}
+        {{- '\n</tool_response>' }}
+        {%- if not loop.last and loop.nextitem.role != "tool" %}
+            {{- '<|im_end|>\n' }}
+        {%- elif loop.last %}
+            {{- '<|im_end|>\n' }}
+        {%- endif %}
+    {%- else %}
+        {{- raise_exception('Unexpected message role.') }}
+    {%- endif %}
+{%- endfor %}
+{%- if add_generation_prompt %}
+    {{- '<|im_start|>assistant\n' }}
+    {%- if enable_thinking is defined and enable_thinking is true %}
+        {{- '<think>\n' }}
+    {%- else %}
+        {{- '<think>\n\n</think>\n\n' }}
+    {%- endif %}
+{%- endif %}

8bit/config.json

@@ -0,0 +1,85 @@
+{
+    "architectures": [
+        "Qwen3_5ForCausalLM"
+    ],
+    "attention_bias": false,
+    "attention_dropout": 0.0,
+    "attn_output_gate": true,
+    "bos_token_id": null,
+    "dtype": "bfloat16",
+    "eos_token_id": 248044,
+    "full_attention_interval": 4,
+    "head_dim": 256,
+    "hidden_act": "silu",
+    "hidden_size": 2048,
+    "initializer_range": 0.02,
+    "intermediate_size": 6144,
+    "layer_types": [
+        "linear_attention",
+        "linear_attention",
+        "linear_attention",
+        "full_attention",
+        "linear_attention",
+        "linear_attention",
+        "linear_attention",
+        "full_attention",
+        "linear_attention",
+        "linear_attention",
+        "linear_attention",
+        "full_attention",
+        "linear_attention",
+        "linear_attention",
+        "linear_attention",
+        "full_attention",
+        "linear_attention",
+        "linear_attention",
+        "linear_attention",
+        "full_attention",
+        "linear_attention",
+        "linear_attention",
+        "linear_attention",
+        "full_attention"
+    ],
+    "linear_conv_kernel_dim": 4,
+    "linear_key_head_dim": 128,
+    "linear_num_key_heads": 16,
+    "linear_num_value_heads": 16,
+    "linear_value_head_dim": 128,
+    "mamba_ssm_dtype": "float32",
+    "max_position_embeddings": 262144,
+    "mlp_only_layers": [],
+    "model_type": "qwen3_5",
+    "mtp_num_hidden_layers": 1,
+    "mtp_use_dedicated_embeddings": false,
+    "num_attention_heads": 8,
+    "num_hidden_layers": 24,
+    "num_key_value_heads": 2,
+    "pad_token_id": null,
+    "partial_rotary_factor": 0.25,
+    "quantization": {
+        "group_size": 64,
+        "bits": 8,
+        "mode": "affine"
+    },
+    "quantization_config": {
+        "group_size": 64,
+        "bits": 8,
+        "mode": "affine"
+    },
+    "rms_norm_eps": 1e-06,
+    "rope_parameters": {
+        "mrope_interleaved": true,
+        "mrope_section": [
+            11,
+            11,
+            10
+        ],
+        "partial_rotary_factor": 0.25,
+        "rope_theta": 10000000,
+        "type": "default"
+    },
+    "tie_word_embeddings": true,
+    "transformers_version": "5.17.0",
+    "use_cache": false,
+    "vocab_size": 248320
+}

8bit/generation_config.json

@@ -0,0 +1,6 @@
+{
+  "_from_model_config": true,
+  "eos_token_id": 248044,
+  "transformers_version": "5.17.0",
+  "use_cache": true
+}

8bit/macjev_norms_fp32.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:02f08c422228929c44dd84856678ea95d41401138aa56b1b771a654376d3e51c
+size 420112

8bit/model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:184b4dda2f1285fe95a00a6271b364c861d370fbf913030f0259b2423f88a740
+size 2000043057

8bit/model.safetensors.index.json

@@ -0,0 +1,702 @@
+{
+    "metadata": {
+        "total_size": 1999953536,
+        "total_parameters": 1881824512
+    },
+    "weight_map": {
+        "language_model.model.embed_tokens.biases": "model.safetensors",
+        "language_model.model.embed_tokens.scales": "model.safetensors",
+        "language_model.model.embed_tokens.weight": "model.safetensors",
+        "language_model.model.layers.0.input_layernorm.weight": "model.safetensors",
+        "language_model.model.layers.0.linear_attn.A_log": "model.safetensors",
+        "language_model.model.layers.0.linear_attn.conv1d.weight": "model.safetensors",
+        "language_model.model.layers.0.linear_attn.dt_bias": "model.safetensors",
+        "language_model.model.layers.0.linear_attn.in_proj_a.biases": "model.safetensors",
+        "language_model.model.layers.0.linear_attn.in_proj_a.scales": "model.safetensors",
+        "language_model.model.layers.0.linear_attn.in_proj_a.weight": "model.safetensors",
+        "language_model.model.layers.0.linear_attn.in_proj_b.biases": "model.safetensors",
+        "language_model.model.layers.0.linear_attn.in_proj_b.scales": "model.safetensors",
+        "language_model.model.layers.0.linear_attn.in_proj_b.weight": "model.safetensors",
+        "language_model.model.layers.0.linear_attn.in_proj_qkv.biases": "model.safetensors",
+        "language_model.model.layers.0.linear_attn.in_proj_qkv.scales": "model.safetensors",
+        "language_model.model.layers.0.linear_attn.in_proj_qkv.weight": "model.safetensors",
+        "language_model.model.layers.0.linear_attn.in_proj_z.biases": "model.safetensors",
+        "language_model.model.layers.0.linear_attn.in_proj_z.scales": "model.safetensors",
+        "language_model.model.layers.0.linear_attn.in_proj_z.weight": "model.safetensors",
+        "language_model.model.layers.0.linear_attn.norm.weight": "model.safetensors",
+        "language_model.model.layers.0.linear_attn.out_proj.biases": "model.safetensors",
+        "language_model.model.layers.0.linear_attn.out_proj.scales": "model.safetensors",
+        "language_model.model.layers.0.linear_attn.out_proj.weight": "model.safetensors",
+        "language_model.model.layers.0.mlp.down_proj.biases": "model.safetensors",
+        "language_model.model.layers.0.mlp.down_proj.scales": "model.safetensors",
+        "language_model.model.layers.0.mlp.down_proj.weight": "model.safetensors",
+        "language_model.model.layers.0.mlp.gate_proj.biases": "model.safetensors",
+        "language_model.model.layers.0.mlp.gate_proj.scales": "model.safetensors",
+        "language_model.model.layers.0.mlp.gate_proj.weight": "model.safetensors",
+        "language_model.model.layers.0.mlp.up_proj.biases": "model.safetensors",
+        "language_model.model.layers.0.mlp.up_proj.scales": "model.safetensors",
+        "language_model.model.layers.0.mlp.up_proj.weight": "model.safetensors",
+        "language_model.model.layers.0.post_attention_layernorm.weight": "model.safetensors",
+        "language_model.model.layers.1.input_layernorm.weight": "model.safetensors",
+        "language_model.model.layers.1.linear_attn.A_log": "model.safetensors",
+        "language_model.model.layers.1.linear_attn.conv1d.weight": "model.safetensors",
+        "language_model.model.layers.1.linear_attn.dt_bias": "model.safetensors",
+        "language_model.model.layers.1.linear_attn.in_proj_a.biases": "model.safetensors",
+        "language_model.model.layers.1.linear_attn.in_proj_a.scales": "model.safetensors",
+        "language_model.model.layers.1.linear_attn.in_proj_a.weight": "model.safetensors",
+        "language_model.model.layers.1.linear_attn.in_proj_b.biases": "model.safetensors",
+        "language_model.model.layers.1.linear_attn.in_proj_b.scales": "model.safetensors",
+        "language_model.model.layers.1.linear_attn.in_proj_b.weight": "model.safetensors",
+        "language_model.model.layers.1.linear_attn.in_proj_qkv.biases": "model.safetensors",
+        "language_model.model.layers.1.linear_attn.in_proj_qkv.scales": "model.safetensors",
+        "language_model.model.layers.1.linear_attn.in_proj_qkv.weight": "model.safetensors",
+        "language_model.model.layers.1.linear_attn.in_proj_z.biases": "model.safetensors",
+        "language_model.model.layers.1.linear_attn.in_proj_z.scales": "model.safetensors",
+        "language_model.model.layers.1.linear_attn.in_proj_z.weight": "model.safetensors",
+        "language_model.model.layers.1.linear_attn.norm.weight": "model.safetensors",
+        "language_model.model.layers.1.linear_attn.out_proj.biases": "model.safetensors",
+        "language_model.model.layers.1.linear_attn.out_proj.scales": "model.safetensors",
+        "language_model.model.layers.1.linear_attn.out_proj.weight": "model.safetensors",
+        "language_model.model.layers.1.mlp.down_proj.biases": "model.safetensors",
+        "language_model.model.layers.1.mlp.down_proj.scales": "model.safetensors",
+        "language_model.model.layers.1.mlp.down_proj.weight": "model.safetensors",
+        "language_model.model.layers.1.mlp.gate_proj.biases": "model.safetensors",
+        "language_model.model.layers.1.mlp.gate_proj.scales": "model.safetensors",
+        "language_model.model.layers.1.mlp.gate_proj.weight": "model.safetensors",
+        "language_model.model.layers.1.mlp.up_proj.biases": "model.safetensors",
+        "language_model.model.layers.1.mlp.up_proj.scales": "model.safetensors",
+        "language_model.model.layers.1.mlp.up_proj.weight": "model.safetensors",
+        "language_model.model.layers.1.post_attention_layernorm.weight": "model.safetensors",
+        "language_model.model.layers.10.input_layernorm.weight": "model.safetensors",
+        "language_model.model.layers.10.linear_attn.A_log": "model.safetensors",
+        "language_model.model.layers.10.linear_attn.conv1d.weight": "model.safetensors",
+        "language_model.model.layers.10.linear_attn.dt_bias": "model.safetensors",
+        "language_model.model.layers.10.linear_attn.in_proj_a.biases": "model.safetensors",
+        "language_model.model.layers.10.linear_attn.in_proj_a.scales": "model.safetensors",
+        "language_model.model.layers.10.linear_attn.in_proj_a.weight": "model.safetensors",
+        "language_model.model.layers.10.linear_attn.in_proj_b.biases": "model.safetensors",
+        "language_model.model.layers.10.linear_attn.in_proj_b.scales": "model.safetensors",
+        "language_model.model.layers.10.linear_attn.in_proj_b.weight": "model.safetensors",
+        "language_model.model.layers.10.linear_attn.in_proj_qkv.biases": "model.safetensors",
+        "language_model.model.layers.10.linear_attn.in_proj_qkv.scales": "model.safetensors",
+        "language_model.model.layers.10.linear_attn.in_proj_qkv.weight": "model.safetensors",
+        "language_model.model.layers.10.linear_attn.in_proj_z.biases": "model.safetensors",
+        "language_model.model.layers.10.linear_attn.in_proj_z.scales": "model.safetensors",
+        "language_model.model.layers.10.linear_attn.in_proj_z.weight": "model.safetensors",
+        "language_model.model.layers.10.linear_attn.norm.weight": "model.safetensors",
+        "language_model.model.layers.10.linear_attn.out_proj.biases": "model.safetensors",
+        "language_model.model.layers.10.linear_attn.out_proj.scales": "model.safetensors",
+        "language_model.model.layers.10.linear_attn.out_proj.weight": "model.safetensors",
+        "language_model.model.layers.10.mlp.down_proj.biases": "model.safetensors",
+        "language_model.model.layers.10.mlp.down_proj.scales": "model.safetensors",
+        "language_model.model.layers.10.mlp.down_proj.weight": "model.safetensors",
+        "language_model.model.layers.10.mlp.gate_proj.biases": "model.safetensors",
+        "language_model.model.layers.10.mlp.gate_proj.scales": "model.safetensors",
+        "language_model.model.layers.10.mlp.gate_proj.weight": "model.safetensors",
+        "language_model.model.layers.10.mlp.up_proj.biases": "model.safetensors",
+        "language_model.model.layers.10.mlp.up_proj.scales": "model.safetensors",
+        "language_model.model.layers.10.mlp.up_proj.weight": "model.safetensors",
+        "language_model.model.layers.10.post_attention_layernorm.weight": "model.safetensors",
+        "language_model.model.layers.11.input_layernorm.weight": "model.safetensors",
+        "language_model.model.layers.11.mlp.down_proj.biases": "model.safetensors",
+        "language_model.model.layers.11.mlp.down_proj.scales": "model.safetensors",
+        "language_model.model.layers.11.mlp.down_proj.weight": "model.safetensors",
+        "language_model.model.layers.11.mlp.gate_proj.biases": "model.safetensors",
+        "language_model.model.layers.11.mlp.gate_proj.scales": "model.safetensors",
+        "language_model.model.layers.11.mlp.gate_proj.weight": "model.safetensors",
+        "language_model.model.layers.11.mlp.up_proj.biases": "model.safetensors",
+        "language_model.model.layers.11.mlp.up_proj.scales": "model.safetensors",
+        "language_model.model.layers.11.mlp.up_proj.weight": "model.safetensors",
+        "language_model.model.layers.11.post_attention_layernorm.weight": "model.safetensors",
+        "language_model.model.layers.11.self_attn.k_norm.weight": "model.safetensors",
+        "language_model.model.layers.11.self_attn.k_proj.biases": "model.safetensors",
+        "language_model.model.layers.11.self_attn.k_proj.scales": "model.safetensors",
+        "language_model.model.layers.11.self_attn.k_proj.weight": "model.safetensors",
+        "language_model.model.layers.11.self_attn.o_proj.biases": "model.safetensors",
+        "language_model.model.layers.11.self_attn.o_proj.scales": "model.safetensors",
+        "language_model.model.layers.11.self_attn.o_proj.weight": "model.safetensors",
+        "language_model.model.layers.11.self_attn.q_norm.weight": "model.safetensors",
+        "language_model.model.layers.11.self_attn.q_proj.biases": "model.safetensors",
+        "language_model.model.layers.11.self_attn.q_proj.scales": "model.safetensors",
+        "language_model.model.layers.11.self_attn.q_proj.weight": "model.safetensors",
+        "language_model.model.layers.11.self_attn.v_proj.biases": "model.safetensors",
+        "language_model.model.layers.11.self_attn.v_proj.scales": "model.safetensors",
+        "language_model.model.layers.11.self_attn.v_proj.weight": "model.safetensors",
+        "language_model.model.layers.12.input_layernorm.weight": "model.safetensors",
+        "language_model.model.layers.12.linear_attn.A_log": "model.safetensors",
+        "language_model.model.layers.12.linear_attn.conv1d.weight": "model.safetensors",
+        "language_model.model.layers.12.linear_attn.dt_bias": "model.safetensors",
+        "language_model.model.layers.12.linear_attn.in_proj_a.biases": "model.safetensors",
+        "language_model.model.layers.12.linear_attn.in_proj_a.scales": "model.safetensors",
+        "language_model.model.layers.12.linear_attn.in_proj_a.weight": "model.safetensors",
+        "language_model.model.layers.12.linear_attn.in_proj_b.biases": "model.safetensors",
+        "language_model.model.layers.12.linear_attn.in_proj_b.scales": "model.safetensors",
+        "language_model.model.layers.12.linear_attn.in_proj_b.weight": "model.safetensors",
+        "language_model.model.layers.12.linear_attn.in_proj_qkv.biases": "model.safetensors",
+        "language_model.model.layers.12.linear_attn.in_proj_qkv.scales": "model.safetensors",
+        "language_model.model.layers.12.linear_attn.in_proj_qkv.weight": "model.safetensors",
+        "language_model.model.layers.12.linear_attn.in_proj_z.biases": "model.safetensors",
+        "language_model.model.layers.12.linear_attn.in_proj_z.scales": "model.safetensors",
+        "language_model.model.layers.12.linear_attn.in_proj_z.weight": "model.safetensors",
+        "language_model.model.layers.12.linear_attn.norm.weight": "model.safetensors",
+        "language_model.model.layers.12.linear_attn.out_proj.biases": "model.safetensors",
+        "language_model.model.layers.12.linear_attn.out_proj.scales": "model.safetensors",
+        "language_model.model.layers.12.linear_attn.out_proj.weight": "model.safetensors",
+        "language_model.model.layers.12.mlp.down_proj.biases": "model.safetensors",
+        "language_model.model.layers.12.mlp.down_proj.scales": "model.safetensors",
+        "language_model.model.layers.12.mlp.down_proj.weight": "model.safetensors",
+        "language_model.model.layers.12.mlp.gate_proj.biases": "model.safetensors",
+        "language_model.model.layers.12.mlp.gate_proj.scales": "model.safetensors",
+        "language_model.model.layers.12.mlp.gate_proj.weight": "model.safetensors",
+        "language_model.model.layers.12.mlp.up_proj.biases": "model.safetensors",
+        "language_model.model.layers.12.mlp.up_proj.scales": "model.safetensors",
+        "language_model.model.layers.12.mlp.up_proj.weight": "model.safetensors",
+        "language_model.model.layers.12.post_attention_layernorm.weight": "model.safetensors",
+        "language_model.model.layers.13.input_layernorm.weight": "model.safetensors",
+        "language_model.model.layers.13.linear_attn.A_log": "model.safetensors",
+        "language_model.model.layers.13.linear_attn.conv1d.weight": "model.safetensors",
+        "language_model.model.layers.13.linear_attn.dt_bias": "model.safetensors",
+        "language_model.model.layers.13.linear_attn.in_proj_a.biases": "model.safetensors",
+        "language_model.model.layers.13.linear_attn.in_proj_a.scales": "model.safetensors",
+        "language_model.model.layers.13.linear_attn.in_proj_a.weight": "model.safetensors",
+        "language_model.model.layers.13.linear_attn.in_proj_b.biases": "model.safetensors",
+        "language_model.model.layers.13.linear_attn.in_proj_b.scales": "model.safetensors",
+        "language_model.model.layers.13.linear_attn.in_proj_b.weight": "model.safetensors",
+        "language_model.model.layers.13.linear_attn.in_proj_qkv.biases": "model.safetensors",
+        "language_model.model.layers.13.linear_attn.in_proj_qkv.scales": "model.safetensors",
+        "language_model.model.layers.13.linear_attn.in_proj_qkv.weight": "model.safetensors",
+        "language_model.model.layers.13.linear_attn.in_proj_z.biases": "model.safetensors",
+        "language_model.model.layers.13.linear_attn.in_proj_z.scales": "model.safetensors",
+        "language_model.model.layers.13.linear_attn.in_proj_z.weight": "model.safetensors",
+        "language_model.model.layers.13.linear_attn.norm.weight": "model.safetensors",
+        "language_model.model.layers.13.linear_attn.out_proj.biases": "model.safetensors",
+        "language_model.model.layers.13.linear_attn.out_proj.scales": "model.safetensors",
+        "language_model.model.layers.13.linear_attn.out_proj.weight": "model.safetensors",
+        "language_model.model.layers.13.mlp.down_proj.biases": "model.safetensors",
+        "language_model.model.layers.13.mlp.down_proj.scales": "model.safetensors",
+        "language_model.model.layers.13.mlp.down_proj.weight": "model.safetensors",
+        "language_model.model.layers.13.mlp.gate_proj.biases": "model.safetensors",
+        "language_model.model.layers.13.mlp.gate_proj.scales": "model.safetensors",
+        "language_model.model.layers.13.mlp.gate_proj.weight": "model.safetensors",
+        "language_model.model.layers.13.mlp.up_proj.biases": "model.safetensors",
+        "language_model.model.layers.13.mlp.up_proj.scales": "model.safetensors",
+        "language_model.model.layers.13.mlp.up_proj.weight": "model.safetensors",
+        "language_model.model.layers.13.post_attention_layernorm.weight": "model.safetensors",
+        "language_model.model.layers.14.input_layernorm.weight": "model.safetensors",
+        "language_model.model.layers.14.linear_attn.A_log": "model.safetensors",
+        "language_model.model.layers.14.linear_attn.conv1d.weight": "model.safetensors",
+        "language_model.model.layers.14.linear_attn.dt_bias": "model.safetensors",
+        "language_model.model.layers.14.linear_attn.in_proj_a.biases": "model.safetensors",
+        "language_model.model.layers.14.linear_attn.in_proj_a.scales": "model.safetensors",
+        "language_model.model.layers.14.linear_attn.in_proj_a.weight": "model.safetensors",
+        "language_model.model.layers.14.linear_attn.in_proj_b.biases": "model.safetensors",
+        "language_model.model.layers.14.linear_attn.in_proj_b.scales": "model.safetensors",
+        "language_model.model.layers.14.linear_attn.in_proj_b.weight": "model.safetensors",
+        "language_model.model.layers.14.linear_attn.in_proj_qkv.biases": "model.safetensors",
+        "language_model.model.layers.14.linear_attn.in_proj_qkv.scales": "model.safetensors",
+        "language_model.model.layers.14.linear_attn.in_proj_qkv.weight": "model.safetensors",
+        "language_model.model.layers.14.linear_attn.in_proj_z.biases": "model.safetensors",
+        "language_model.model.layers.14.linear_attn.in_proj_z.scales": "model.safetensors",
+        "language_model.model.layers.14.linear_attn.in_proj_z.weight": "model.safetensors",
+        "language_model.model.layers.14.linear_attn.norm.weight": "model.safetensors",
+        "language_model.model.layers.14.linear_attn.out_proj.biases": "model.safetensors",
+        "language_model.model.layers.14.linear_attn.out_proj.scales": "model.safetensors",
+        "language_model.model.layers.14.linear_attn.out_proj.weight": "model.safetensors",
+        "language_model.model.layers.14.mlp.down_proj.biases": "model.safetensors",
+        "language_model.model.layers.14.mlp.down_proj.scales": "model.safetensors",
+        "language_model.model.layers.14.mlp.down_proj.weight": "model.safetensors",
+        "language_model.model.layers.14.mlp.gate_proj.biases": "model.safetensors",
+        "language_model.model.layers.14.mlp.gate_proj.scales": "model.safetensors",
+        "language_model.model.layers.14.mlp.gate_proj.weight": "model.safetensors",
+        "language_model.model.layers.14.mlp.up_proj.biases": "model.safetensors",
+        "language_model.model.layers.14.mlp.up_proj.scales": "model.safetensors",
+        "language_model.model.layers.14.mlp.up_proj.weight": "model.safetensors",
+        "language_model.model.layers.14.post_attention_layernorm.weight": "model.safetensors",
+        "language_model.model.layers.15.input_layernorm.weight": "model.safetensors",
+        "language_model.model.layers.15.mlp.down_proj.biases": "model.safetensors",
+        "language_model.model.layers.15.mlp.down_proj.scales": "model.safetensors",
+        "language_model.model.layers.15.mlp.down_proj.weight": "model.safetensors",
+        "language_model.model.layers.15.mlp.gate_proj.biases": "model.safetensors",
+        "language_model.model.layers.15.mlp.gate_proj.scales": "model.safetensors",
+        "language_model.model.layers.15.mlp.gate_proj.weight": "model.safetensors",
+        "language_model.model.layers.15.mlp.up_proj.biases": "model.safetensors",
+        "language_model.model.layers.15.mlp.up_proj.scales": "model.safetensors",
+        "language_model.model.layers.15.mlp.up_proj.weight": "model.safetensors",
+        "language_model.model.layers.15.post_attention_layernorm.weight": "model.safetensors",
+        "language_model.model.layers.15.self_attn.k_norm.weight": "model.safetensors",
+        "language_model.model.layers.15.self_attn.k_proj.biases": "model.safetensors",
+        "language_model.model.layers.15.self_attn.k_proj.scales": "model.safetensors",
+        "language_model.model.layers.15.self_attn.k_proj.weight": "model.safetensors",
+        "language_model.model.layers.15.self_attn.o_proj.biases": "model.safetensors",
+        "language_model.model.layers.15.self_attn.o_proj.scales": "model.safetensors",
+        "language_model.model.layers.15.self_attn.o_proj.weight": "model.safetensors",
+        "language_model.model.layers.15.self_attn.q_norm.weight": "model.safetensors",
+        "language_model.model.layers.15.self_attn.q_proj.biases": "model.safetensors",
+        "language_model.model.layers.15.self_attn.q_proj.scales": "model.safetensors",
+        "language_model.model.layers.15.self_attn.q_proj.weight": "model.safetensors",
+        "language_model.model.layers.15.self_attn.v_proj.biases": "model.safetensors",
+        "language_model.model.layers.15.self_attn.v_proj.scales": "model.safetensors",
+        "language_model.model.layers.15.self_attn.v_proj.weight": "model.safetensors",
+        "language_model.model.layers.16.input_layernorm.weight": "model.safetensors",
+        "language_model.model.layers.16.linear_attn.A_log": "model.safetensors",
+        "language_model.model.layers.16.linear_attn.conv1d.weight": "model.safetensors",
+        "language_model.model.layers.16.linear_attn.dt_bias": "model.safetensors",
+        "language_model.model.layers.16.linear_attn.in_proj_a.biases": "model.safetensors",
+        "language_model.model.layers.16.linear_attn.in_proj_a.scales": "model.safetensors",
+        "language_model.model.layers.16.linear_attn.in_proj_a.weight": "model.safetensors",
+        "language_model.model.layers.16.linear_attn.in_proj_b.biases": "model.safetensors",
+        "language_model.model.layers.16.linear_attn.in_proj_b.scales": "model.safetensors",
+        "language_model.model.layers.16.linear_attn.in_proj_b.weight": "model.safetensors",
+        "language_model.model.layers.16.linear_attn.in_proj_qkv.biases": "model.safetensors",
+        "language_model.model.layers.16.linear_attn.in_proj_qkv.scales": "model.safetensors",
+        "language_model.model.layers.16.linear_attn.in_proj_qkv.weight": "model.safetensors",
+        "language_model.model.layers.16.linear_attn.in_proj_z.biases": "model.safetensors",
+        "language_model.model.layers.16.linear_attn.in_proj_z.scales": "model.safetensors",
+        "language_model.model.layers.16.linear_attn.in_proj_z.weight": "model.safetensors",
+        "language_model.model.layers.16.linear_attn.norm.weight": "model.safetensors",
+        "language_model.model.layers.16.linear_attn.out_proj.biases": "model.safetensors",
+        "language_model.model.layers.16.linear_attn.out_proj.scales": "model.safetensors",
+        "language_model.model.layers.16.linear_attn.out_proj.weight": "model.safetensors",
+        "language_model.model.layers.16.mlp.down_proj.biases": "model.safetensors",
+        "language_model.model.layers.16.mlp.down_proj.scales": "model.safetensors",
+        "language_model.model.layers.16.mlp.down_proj.weight": "model.safetensors",
+        "language_model.model.layers.16.mlp.gate_proj.biases": "model.safetensors",
+        "language_model.model.layers.16.mlp.gate_proj.scales": "model.safetensors",
+        "language_model.model.layers.16.mlp.gate_proj.weight": "model.safetensors",
+        "language_model.model.layers.16.mlp.up_proj.biases": "model.safetensors",
+        "language_model.model.layers.16.mlp.up_proj.scales": "model.safetensors",
+        "language_model.model.layers.16.mlp.up_proj.weight": "model.safetensors",
+        "language_model.model.layers.16.post_attention_layernorm.weight": "model.safetensors",
+        "language_model.model.layers.17.input_layernorm.weight": "model.safetensors",
+        "language_model.model.layers.17.linear_attn.A_log": "model.safetensors",
+        "language_model.model.layers.17.linear_attn.conv1d.weight": "model.safetensors",
+        "language_model.model.layers.17.linear_attn.dt_bias": "model.safetensors",
+        "language_model.model.layers.17.linear_attn.in_proj_a.biases": "model.safetensors",
+        "language_model.model.layers.17.linear_attn.in_proj_a.scales": "model.safetensors",
+        "language_model.model.layers.17.linear_attn.in_proj_a.weight": "model.safetensors",
+        "language_model.model.layers.17.linear_attn.in_proj_b.biases": "model.safetensors",
+        "language_model.model.layers.17.linear_attn.in_proj_b.scales": "model.safetensors",
+        "language_model.model.layers.17.linear_attn.in_proj_b.weight": "model.safetensors",
+        "language_model.model.layers.17.linear_attn.in_proj_qkv.biases": "model.safetensors",
+        "language_model.model.layers.17.linear_attn.in_proj_qkv.scales": "model.safetensors",
+        "language_model.model.layers.17.linear_attn.in_proj_qkv.weight": "model.safetensors",
+        "language_model.model.layers.17.linear_attn.in_proj_z.biases": "model.safetensors",
+        "language_model.model.layers.17.linear_attn.in_proj_z.scales": "model.safetensors",
+        "language_model.model.layers.17.linear_attn.in_proj_z.weight": "model.safetensors",
+        "language_model.model.layers.17.linear_attn.norm.weight": "model.safetensors",
+        "language_model.model.layers.17.linear_attn.out_proj.biases": "model.safetensors",
+        "language_model.model.layers.17.linear_attn.out_proj.scales": "model.safetensors",
+        "language_model.model.layers.17.linear_attn.out_proj.weight": "model.safetensors",
+        "language_model.model.layers.17.mlp.down_proj.biases": "model.safetensors",
+        "language_model.model.layers.17.mlp.down_proj.scales": "model.safetensors",
+        "language_model.model.layers.17.mlp.down_proj.weight": "model.safetensors",
+        "language_model.model.layers.17.mlp.gate_proj.biases": "model.safetensors",
+        "language_model.model.layers.17.mlp.gate_proj.scales": "model.safetensors",
+        "language_model.model.layers.17.mlp.gate_proj.weight": "model.safetensors",
+        "language_model.model.layers.17.mlp.up_proj.biases": "model.safetensors",
+        "language_model.model.layers.17.mlp.up_proj.scales": "model.safetensors",
+        "language_model.model.layers.17.mlp.up_proj.weight": "model.safetensors",
+        "language_model.model.layers.17.post_attention_layernorm.weight": "model.safetensors",
+        "language_model.model.layers.18.input_layernorm.weight": "model.safetensors",
+        "language_model.model.layers.18.linear_attn.A_log": "model.safetensors",
+        "language_model.model.layers.18.linear_attn.conv1d.weight": "model.safetensors",
+        "language_model.model.layers.18.linear_attn.dt_bias": "model.safetensors",
+        "language_model.model.layers.18.linear_attn.in_proj_a.biases": "model.safetensors",
+        "language_model.model.layers.18.linear_attn.in_proj_a.scales": "model.safetensors",
+        "language_model.model.layers.18.linear_attn.in_proj_a.weight": "model.safetensors",
+        "language_model.model.layers.18.linear_attn.in_proj_b.biases": "model.safetensors",
+        "language_model.model.layers.18.linear_attn.in_proj_b.scales": "model.safetensors",
+        "language_model.model.layers.18.linear_attn.in_proj_b.weight": "model.safetensors",
+        "language_model.model.layers.18.linear_attn.in_proj_qkv.biases": "model.safetensors",
+        "language_model.model.layers.18.linear_attn.in_proj_qkv.scales": "model.safetensors",
+        "language_model.model.layers.18.linear_attn.in_proj_qkv.weight": "model.safetensors",
+        "language_model.model.layers.18.linear_attn.in_proj_z.biases": "model.safetensors",
+        "language_model.model.layers.18.linear_attn.in_proj_z.scales": "model.safetensors",
+        "language_model.model.layers.18.linear_attn.in_proj_z.weight": "model.safetensors",
+        "language_model.model.layers.18.linear_attn.norm.weight": "model.safetensors",
+        "language_model.model.layers.18.linear_attn.out_proj.biases": "model.safetensors",
+        "language_model.model.layers.18.linear_attn.out_proj.scales": "model.safetensors",
+        "language_model.model.layers.18.linear_attn.out_proj.weight": "model.safetensors",
+        "language_model.model.layers.18.mlp.down_proj.biases": "model.safetensors",
+        "language_model.model.layers.18.mlp.down_proj.scales": "model.safetensors",
+        "language_model.model.layers.18.mlp.down_proj.weight": "model.safetensors",
+        "language_model.model.layers.18.mlp.gate_proj.biases": "model.safetensors",
+        "language_model.model.layers.18.mlp.gate_proj.scales": "model.safetensors",
+        "language_model.model.layers.18.mlp.gate_proj.weight": "model.safetensors",
+        "language_model.model.layers.18.mlp.up_proj.biases": "model.safetensors",
+        "language_model.model.layers.18.mlp.up_proj.scales": "model.safetensors",
+        "language_model.model.layers.18.mlp.up_proj.weight": "model.safetensors",
+        "language_model.model.layers.18.post_attention_layernorm.weight": "model.safetensors",
+        "language_model.model.layers.19.input_layernorm.weight": "model.safetensors",
+        "language_model.model.layers.19.mlp.down_proj.biases": "model.safetensors",
+        "language_model.model.layers.19.mlp.down_proj.scales": "model.safetensors",
+        "language_model.model.layers.19.mlp.down_proj.weight": "model.safetensors",
+        "language_model.model.layers.19.mlp.gate_proj.biases": "model.safetensors",
+        "language_model.model.layers.19.mlp.gate_proj.scales": "model.safetensors",
+        "language_model.model.layers.19.mlp.gate_proj.weight": "model.safetensors",
+        "language_model.model.layers.19.mlp.up_proj.biases": "model.safetensors",
+        "language_model.model.layers.19.mlp.up_proj.scales": "model.safetensors",
+        "language_model.model.layers.19.mlp.up_proj.weight": "model.safetensors",
+        "language_model.model.layers.19.post_attention_layernorm.weight": "model.safetensors",
+        "language_model.model.layers.19.self_attn.k_norm.weight": "model.safetensors",
+        "language_model.model.layers.19.self_attn.k_proj.biases": "model.safetensors",
+        "language_model.model.layers.19.self_attn.k_proj.scales": "model.safetensors",
+        "language_model.model.layers.19.self_attn.k_proj.weight": "model.safetensors",
+        "language_model.model.layers.19.self_attn.o_proj.biases": "model.safetensors",
+        "language_model.model.layers.19.self_attn.o_proj.scales": "model.safetensors",
+        "language_model.model.layers.19.self_attn.o_proj.weight": "model.safetensors",
+        "language_model.model.layers.19.self_attn.q_norm.weight": "model.safetensors",
+        "language_model.model.layers.19.self_attn.q_proj.biases": "model.safetensors",
+        "language_model.model.layers.19.self_attn.q_proj.scales": "model.safetensors",
+        "language_model.model.layers.19.self_attn.q_proj.weight": "model.safetensors",
+        "language_model.model.layers.19.self_attn.v_proj.biases": "model.safetensors",
+        "language_model.model.layers.19.self_attn.v_proj.scales": "model.safetensors",
+        "language_model.model.layers.19.self_attn.v_proj.weight": "model.safetensors",
+        "language_model.model.layers.2.input_layernorm.weight": "model.safetensors",
+        "language_model.model.layers.2.linear_attn.A_log": "model.safetensors",
+        "language_model.model.layers.2.linear_attn.conv1d.weight": "model.safetensors",
+        "language_model.model.layers.2.linear_attn.dt_bias": "model.safetensors",
+        "language_model.model.layers.2.linear_attn.in_proj_a.biases": "model.safetensors",
+        "language_model.model.layers.2.linear_attn.in_proj_a.scales": "model.safetensors",
+        "language_model.model.layers.2.linear_attn.in_proj_a.weight": "model.safetensors",
+        "language_model.model.layers.2.linear_attn.in_proj_b.biases": "model.safetensors",
+        "language_model.model.layers.2.linear_attn.in_proj_b.scales": "model.safetensors",
+        "language_model.model.layers.2.linear_attn.in_proj_b.weight": "model.safetensors",
+        "language_model.model.layers.2.linear_attn.in_proj_qkv.biases": "model.safetensors",
+        "language_model.model.layers.2.linear_attn.in_proj_qkv.scales": "model.safetensors",
+        "language_model.model.layers.2.linear_attn.in_proj_qkv.weight": "model.safetensors",
+        "language_model.model.layers.2.linear_attn.in_proj_z.biases": "model.safetensors",
+        "language_model.model.layers.2.linear_attn.in_proj_z.scales": "model.safetensors",
+        "language_model.model.layers.2.linear_attn.in_proj_z.weight": "model.safetensors",
+        "language_model.model.layers.2.linear_attn.norm.weight": "model.safetensors",
+        "language_model.model.layers.2.linear_attn.out_proj.biases": "model.safetensors",
+        "language_model.model.layers.2.linear_attn.out_proj.scales": "model.safetensors",
+        "language_model.model.layers.2.linear_attn.out_proj.weight": "model.safetensors",
+        "language_model.model.layers.2.mlp.down_proj.biases": "model.safetensors",
+        "language_model.model.layers.2.mlp.down_proj.scales": "model.safetensors",
+        "language_model.model.layers.2.mlp.down_proj.weight": "model.safetensors",
+        "language_model.model.layers.2.mlp.gate_proj.biases": "model.safetensors",
+        "language_model.model.layers.2.mlp.gate_proj.scales": "model.safetensors",
+        "language_model.model.layers.2.mlp.gate_proj.weight": "model.safetensors",
+        "language_model.model.layers.2.mlp.up_proj.biases": "model.safetensors",
+        "language_model.model.layers.2.mlp.up_proj.scales": "model.safetensors",
+        "language_model.model.layers.2.mlp.up_proj.weight": "model.safetensors",
+        "language_model.model.layers.2.post_attention_layernorm.weight": "model.safetensors",
+        "language_model.model.layers.20.input_layernorm.weight": "model.safetensors",
+        "language_model.model.layers.20.linear_attn.A_log": "model.safetensors",
+        "language_model.model.layers.20.linear_attn.conv1d.weight": "model.safetensors",
+        "language_model.model.layers.20.linear_attn.dt_bias": "model.safetensors",
+        "language_model.model.layers.20.linear_attn.in_proj_a.biases": "model.safetensors",
+        "language_model.model.layers.20.linear_attn.in_proj_a.scales": "model.safetensors",
+        "language_model.model.layers.20.linear_attn.in_proj_a.weight": "model.safetensors",
+        "language_model.model.layers.20.linear_attn.in_proj_b.biases": "model.safetensors",
+        "language_model.model.layers.20.linear_attn.in_proj_b.scales": "model.safetensors",
+        "language_model.model.layers.20.linear_attn.in_proj_b.weight": "model.safetensors",
+        "language_model.model.layers.20.linear_attn.in_proj_qkv.biases": "model.safetensors",
+        "language_model.model.layers.20.linear_attn.in_proj_qkv.scales": "model.safetensors",
+        "language_model.model.layers.20.linear_attn.in_proj_qkv.weight": "model.safetensors",
+        "language_model.model.layers.20.linear_attn.in_proj_z.biases": "model.safetensors",
+        "language_model.model.layers.20.linear_attn.in_proj_z.scales": "model.safetensors",
+        "language_model.model.layers.20.linear_attn.in_proj_z.weight": "model.safetensors",
+        "language_model.model.layers.20.linear_attn.norm.weight": "model.safetensors",
+        "language_model.model.layers.20.linear_attn.out_proj.biases": "model.safetensors",
+        "language_model.model.layers.20.linear_attn.out_proj.scales": "model.safetensors",
+        "language_model.model.layers.20.linear_attn.out_proj.weight": "model.safetensors",
+        "language_model.model.layers.20.mlp.down_proj.biases": "model.safetensors",
+        "language_model.model.layers.20.mlp.down_proj.scales": "model.safetensors",
+        "language_model.model.layers.20.mlp.down_proj.weight": "model.safetensors",
+        "language_model.model.layers.20.mlp.gate_proj.biases": "model.safetensors",
+        "language_model.model.layers.20.mlp.gate_proj.scales": "model.safetensors",
+        "language_model.model.layers.20.mlp.gate_proj.weight": "model.safetensors",
+        "language_model.model.layers.20.mlp.up_proj.biases": "model.safetensors",
+        "language_model.model.layers.20.mlp.up_proj.scales": "model.safetensors",
+        "language_model.model.layers.20.mlp.up_proj.weight": "model.safetensors",
+        "language_model.model.layers.20.post_attention_layernorm.weight": "model.safetensors",
+        "language_model.model.layers.21.input_layernorm.weight": "model.safetensors",
+        "language_model.model.layers.21.linear_attn.A_log": "model.safetensors",
+        "language_model.model.layers.21.linear_attn.conv1d.weight": "model.safetensors",
+        "language_model.model.layers.21.linear_attn.dt_bias": "model.safetensors",
+        "language_model.model.layers.21.linear_attn.in_proj_a.biases": "model.safetensors",
+        "language_model.model.layers.21.linear_attn.in_proj_a.scales": "model.safetensors",
+        "language_model.model.layers.21.linear_attn.in_proj_a.weight": "model.safetensors",
+        "language_model.model.layers.21.linear_attn.in_proj_b.biases": "model.safetensors",
+        "language_model.model.layers.21.linear_attn.in_proj_b.scales": "model.safetensors",
+        "language_model.model.layers.21.linear_attn.in_proj_b.weight": "model.safetensors",
+        "language_model.model.layers.21.linear_attn.in_proj_qkv.biases": "model.safetensors",
+        "language_model.model.layers.21.linear_attn.in_proj_qkv.scales": "model.safetensors",
+        "language_model.model.layers.21.linear_attn.in_proj_qkv.weight": "model.safetensors",
+        "language_model.model.layers.21.linear_attn.in_proj_z.biases": "model.safetensors",
+        "language_model.model.layers.21.linear_attn.in_proj_z.scales": "model.safetensors",
+        "language_model.model.layers.21.linear_attn.in_proj_z.weight": "model.safetensors",
+        "language_model.model.layers.21.linear_attn.norm.weight": "model.safetensors",
+        "language_model.model.layers.21.linear_attn.out_proj.biases": "model.safetensors",
+        "language_model.model.layers.21.linear_attn.out_proj.scales": "model.safetensors",
+        "language_model.model.layers.21.linear_attn.out_proj.weight": "model.safetensors",
+        "language_model.model.layers.21.mlp.down_proj.biases": "model.safetensors",
+        "language_model.model.layers.21.mlp.down_proj.scales": "model.safetensors",
+        "language_model.model.layers.21.mlp.down_proj.weight": "model.safetensors",
+        "language_model.model.layers.21.mlp.gate_proj.biases": "model.safetensors",
+        "language_model.model.layers.21.mlp.gate_proj.scales": "model.safetensors",
+        "language_model.model.layers.21.mlp.gate_proj.weight": "model.safetensors",
+        "language_model.model.layers.21.mlp.up_proj.biases": "model.safetensors",
+        "language_model.model.layers.21.mlp.up_proj.scales": "model.safetensors",
+        "language_model.model.layers.21.mlp.up_proj.weight": "model.safetensors",
+        "language_model.model.layers.21.post_attention_layernorm.weight": "model.safetensors",
+        "language_model.model.layers.22.input_layernorm.weight": "model.safetensors",
+        "language_model.model.layers.22.linear_attn.A_log": "model.safetensors",
+        "language_model.model.layers.22.linear_attn.conv1d.weight": "model.safetensors",
+        "language_model.model.layers.22.linear_attn.dt_bias": "model.safetensors",
+        "language_model.model.layers.22.linear_attn.in_proj_a.biases": "model.safetensors",
+        "language_model.model.layers.22.linear_attn.in_proj_a.scales": "model.safetensors",
+        "language_model.model.layers.22.linear_attn.in_proj_a.weight": "model.safetensors",
+        "language_model.model.layers.22.linear_attn.in_proj_b.biases": "model.safetensors",
+        "language_model.model.layers.22.linear_attn.in_proj_b.scales": "model.safetensors",
+        "language_model.model.layers.22.linear_attn.in_proj_b.weight": "model.safetensors",
+        "language_model.model.layers.22.linear_attn.in_proj_qkv.biases": "model.safetensors",
+        "language_model.model.layers.22.linear_attn.in_proj_qkv.scales": "model.safetensors",
+        "language_model.model.layers.22.linear_attn.in_proj_qkv.weight": "model.safetensors",
+        "language_model.model.layers.22.linear_attn.in_proj_z.biases": "model.safetensors",
+        "language_model.model.layers.22.linear_attn.in_proj_z.scales": "model.safetensors",
+        "language_model.model.layers.22.linear_attn.in_proj_z.weight": "model.safetensors",
+        "language_model.model.layers.22.linear_attn.norm.weight": "model.safetensors",
+        "language_model.model.layers.22.linear_attn.out_proj.biases": "model.safetensors",
+        "language_model.model.layers.22.linear_attn.out_proj.scales": "model.safetensors",
+        "language_model.model.layers.22.linear_attn.out_proj.weight": "model.safetensors",
+        "language_model.model.layers.22.mlp.down_proj.biases": "model.safetensors",
+        "language_model.model.layers.22.mlp.down_proj.scales": "model.safetensors",
+        "language_model.model.layers.22.mlp.down_proj.weight": "model.safetensors",
+        "language_model.model.layers.22.mlp.gate_proj.biases": "model.safetensors",
+        "language_model.model.layers.22.mlp.gate_proj.scales": "model.safetensors",
+        "language_model.model.layers.22.mlp.gate_proj.weight": "model.safetensors",
+        "language_model.model.layers.22.mlp.up_proj.biases": "model.safetensors",
+        "language_model.model.layers.22.mlp.up_proj.scales": "model.safetensors",
+        "language_model.model.layers.22.mlp.up_proj.weight": "model.safetensors",
+        "language_model.model.layers.22.post_attention_layernorm.weight": "model.safetensors",
+        "language_model.model.layers.23.input_layernorm.weight": "model.safetensors",
+        "language_model.model.layers.23.mlp.down_proj.biases": "model.safetensors",
+        "language_model.model.layers.23.mlp.down_proj.scales": "model.safetensors",
+        "language_model.model.layers.23.mlp.down_proj.weight": "model.safetensors",
+        "language_model.model.layers.23.mlp.gate_proj.biases": "model.safetensors",
+        "language_model.model.layers.23.mlp.gate_proj.scales": "model.safetensors",
+        "language_model.model.layers.23.mlp.gate_proj.weight": "model.safetensors",
+        "language_model.model.layers.23.mlp.up_proj.biases": "model.safetensors",
+        "language_model.model.layers.23.mlp.up_proj.scales": "model.safetensors",
+        "language_model.model.layers.23.mlp.up_proj.weight": "model.safetensors",
+        "language_model.model.layers.23.post_attention_layernorm.weight": "model.safetensors",
+        "language_model.model.layers.23.self_attn.k_norm.weight": "model.safetensors",
+        "language_model.model.layers.23.self_attn.k_proj.biases": "model.safetensors",
+        "language_model.model.layers.23.self_attn.k_proj.scales": "model.safetensors",
+        "language_model.model.layers.23.self_attn.k_proj.weight": "model.safetensors",
+        "language_model.model.layers.23.self_attn.o_proj.biases": "model.safetensors",
+        "language_model.model.layers.23.self_attn.o_proj.scales": "model.safetensors",
+        "language_model.model.layers.23.self_attn.o_proj.weight": "model.safetensors",
+        "language_model.model.layers.23.self_attn.q_norm.weight": "model.safetensors",
+        "language_model.model.layers.23.self_attn.q_proj.biases": "model.safetensors",
+        "language_model.model.layers.23.self_attn.q_proj.scales": "model.safetensors",
+        "language_model.model.layers.23.self_attn.q_proj.weight": "model.safetensors",
+        "language_model.model.layers.23.self_attn.v_proj.biases": "model.safetensors",
+        "language_model.model.layers.23.self_attn.v_proj.scales": "model.safetensors",
+        "language_model.model.layers.23.self_attn.v_proj.weight": "model.safetensors",
+        "language_model.model.layers.3.input_layernorm.weight": "model.safetensors",
+        "language_model.model.layers.3.mlp.down_proj.biases": "model.safetensors",
+        "language_model.model.layers.3.mlp.down_proj.scales": "model.safetensors",
+        "language_model.model.layers.3.mlp.down_proj.weight": "model.safetensors",
+        "language_model.model.layers.3.mlp.gate_proj.biases": "model.safetensors",
+        "language_model.model.layers.3.mlp.gate_proj.scales": "model.safetensors",
+        "language_model.model.layers.3.mlp.gate_proj.weight": "model.safetensors",
+        "language_model.model.layers.3.mlp.up_proj.biases": "model.safetensors",
+        "language_model.model.layers.3.mlp.up_proj.scales": "model.safetensors",
+        "language_model.model.layers.3.mlp.up_proj.weight": "model.safetensors",
+        "language_model.model.layers.3.post_attention_layernorm.weight": "model.safetensors",
+        "language_model.model.layers.3.self_attn.k_norm.weight": "model.safetensors",
+        "language_model.model.layers.3.self_attn.k_proj.biases": "model.safetensors",
+        "language_model.model.layers.3.self_attn.k_proj.scales": "model.safetensors",
+        "language_model.model.layers.3.self_attn.k_proj.weight": "model.safetensors",
+        "language_model.model.layers.3.self_attn.o_proj.biases": "model.safetensors",
+        "language_model.model.layers.3.self_attn.o_proj.scales": "model.safetensors",
+        "language_model.model.layers.3.self_attn.o_proj.weight": "model.safetensors",
+        "language_model.model.layers.3.self_attn.q_norm.weight": "model.safetensors",
+        "language_model.model.layers.3.self_attn.q_proj.biases": "model.safetensors",
+        "language_model.model.layers.3.self_attn.q_proj.scales": "model.safetensors",
+        "language_model.model.layers.3.self_attn.q_proj.weight": "model.safetensors",
+        "language_model.model.layers.3.self_attn.v_proj.biases": "model.safetensors",
+        "language_model.model.layers.3.self_attn.v_proj.scales": "model.safetensors",
+        "language_model.model.layers.3.self_attn.v_proj.weight": "model.safetensors",
+        "language_model.model.layers.4.input_layernorm.weight": "model.safetensors",
+        "language_model.model.layers.4.linear_attn.A_log": "model.safetensors",
+        "language_model.model.layers.4.linear_attn.conv1d.weight": "model.safetensors",
+        "language_model.model.layers.4.linear_attn.dt_bias": "model.safetensors",
+        "language_model.model.layers.4.linear_attn.in_proj_a.biases": "model.safetensors",
+        "language_model.model.layers.4.linear_attn.in_proj_a.scales": "model.safetensors",
+        "language_model.model.layers.4.linear_attn.in_proj_a.weight": "model.safetensors",
+        "language_model.model.layers.4.linear_attn.in_proj_b.biases": "model.safetensors",
+        "language_model.model.layers.4.linear_attn.in_proj_b.scales": "model.safetensors",
+        "language_model.model.layers.4.linear_attn.in_proj_b.weight": "model.safetensors",
+        "language_model.model.layers.4.linear_attn.in_proj_qkv.biases": "model.safetensors",
+        "language_model.model.layers.4.linear_attn.in_proj_qkv.scales": "model.safetensors",
+        "language_model.model.layers.4.linear_attn.in_proj_qkv.weight": "model.safetensors",
+        "language_model.model.layers.4.linear_attn.in_proj_z.biases": "model.safetensors",
+        "language_model.model.layers.4.linear_attn.in_proj_z.scales": "model.safetensors",
+        "language_model.model.layers.4.linear_attn.in_proj_z.weight": "model.safetensors",
+        "language_model.model.layers.4.linear_attn.norm.weight": "model.safetensors",
+        "language_model.model.layers.4.linear_attn.out_proj.biases": "model.safetensors",
+        "language_model.model.layers.4.linear_attn.out_proj.scales": "model.safetensors",
+        "language_model.model.layers.4.linear_attn.out_proj.weight": "model.safetensors",
+        "language_model.model.layers.4.mlp.down_proj.biases": "model.safetensors",
+        "language_model.model.layers.4.mlp.down_proj.scales": "model.safetensors",
+        "language_model.model.layers.4.mlp.down_proj.weight": "model.safetensors",
+        "language_model.model.layers.4.mlp.gate_proj.biases": "model.safetensors",
+        "language_model.model.layers.4.mlp.gate_proj.scales": "model.safetensors",
+        "language_model.model.layers.4.mlp.gate_proj.weight": "model.safetensors",
+        "language_model.model.layers.4.mlp.up_proj.biases": "model.safetensors",
+        "language_model.model.layers.4.mlp.up_proj.scales": "model.safetensors",
+        "language_model.model.layers.4.mlp.up_proj.weight": "model.safetensors",
+        "language_model.model.layers.4.post_attention_layernorm.weight": "model.safetensors",
+        "language_model.model.layers.5.input_layernorm.weight": "model.safetensors",
+        "language_model.model.layers.5.linear_attn.A_log": "model.safetensors",
+        "language_model.model.layers.5.linear_attn.conv1d.weight": "model.safetensors",
+        "language_model.model.layers.5.linear_attn.dt_bias": "model.safetensors",
+        "language_model.model.layers.5.linear_attn.in_proj_a.biases": "model.safetensors",
+        "language_model.model.layers.5.linear_attn.in_proj_a.scales": "model.safetensors",
+        "language_model.model.layers.5.linear_attn.in_proj_a.weight": "model.safetensors",
+        "language_model.model.layers.5.linear_attn.in_proj_b.biases": "model.safetensors",
+        "language_model.model.layers.5.linear_attn.in_proj_b.scales": "model.safetensors",
+        "language_model.model.layers.5.linear_attn.in_proj_b.weight": "model.safetensors",
+        "language_model.model.layers.5.linear_attn.in_proj_qkv.biases": "model.safetensors",
+        "language_model.model.layers.5.linear_attn.in_proj_qkv.scales": "model.safetensors",
+        "language_model.model.layers.5.linear_attn.in_proj_qkv.weight": "model.safetensors",
+        "language_model.model.layers.5.linear_attn.in_proj_z.biases": "model.safetensors",
+        "language_model.model.layers.5.linear_attn.in_proj_z.scales": "model.safetensors",
+        "language_model.model.layers.5.linear_attn.in_proj_z.weight": "model.safetensors",
+        "language_model.model.layers.5.linear_attn.norm.weight": "model.safetensors",
+        "language_model.model.layers.5.linear_attn.out_proj.biases": "model.safetensors",
+        "language_model.model.layers.5.linear_attn.out_proj.scales": "model.safetensors",
+        "language_model.model.layers.5.linear_attn.out_proj.weight": "model.safetensors",
+        "language_model.model.layers.5.mlp.down_proj.biases": "model.safetensors",
+        "language_model.model.layers.5.mlp.down_proj.scales": "model.safetensors",
+        "language_model.model.layers.5.mlp.down_proj.weight": "model.safetensors",
+        "language_model.model.layers.5.mlp.gate_proj.biases": "model.safetensors",
+        "language_model.model.layers.5.mlp.gate_proj.scales": "model.safetensors",
+        "language_model.model.layers.5.mlp.gate_proj.weight": "model.safetensors",
+        "language_model.model.layers.5.mlp.up_proj.biases": "model.safetensors",
+        "language_model.model.layers.5.mlp.up_proj.scales": "model.safetensors",
+        "language_model.model.layers.5.mlp.up_proj.weight": "model.safetensors",
+        "language_model.model.layers.5.post_attention_layernorm.weight": "model.safetensors",
+        "language_model.model.layers.6.input_layernorm.weight": "model.safetensors",
+        "language_model.model.layers.6.linear_attn.A_log": "model.safetensors",
+        "language_model.model.layers.6.linear_attn.conv1d.weight": "model.safetensors",
+        "language_model.model.layers.6.linear_attn.dt_bias": "model.safetensors",
+        "language_model.model.layers.6.linear_attn.in_proj_a.biases": "model.safetensors",
+        "language_model.model.layers.6.linear_attn.in_proj_a.scales": "model.safetensors",
+        "language_model.model.layers.6.linear_attn.in_proj_a.weight": "model.safetensors",
+        "language_model.model.layers.6.linear_attn.in_proj_b.biases": "model.safetensors",
+        "language_model.model.layers.6.linear_attn.in_proj_b.scales": "model.safetensors",
+        "language_model.model.layers.6.linear_attn.in_proj_b.weight": "model.safetensors",
+        "language_model.model.layers.6.linear_attn.in_proj_qkv.biases": "model.safetensors",
+        "language_model.model.layers.6.linear_attn.in_proj_qkv.scales": "model.safetensors",
+        "language_model.model.layers.6.linear_attn.in_proj_qkv.weight": "model.safetensors",
+        "language_model.model.layers.6.linear_attn.in_proj_z.biases": "model.safetensors",
+        "language_model.model.layers.6.linear_attn.in_proj_z.scales": "model.safetensors",
+        "language_model.model.layers.6.linear_attn.in_proj_z.weight": "model.safetensors",
+        "language_model.model.layers.6.linear_attn.norm.weight": "model.safetensors",
+        "language_model.model.layers.6.linear_attn.out_proj.biases": "model.safetensors",
+        "language_model.model.layers.6.linear_attn.out_proj.scales": "model.safetensors",
+        "language_model.model.layers.6.linear_attn.out_proj.weight": "model.safetensors",
+        "language_model.model.layers.6.mlp.down_proj.biases": "model.safetensors",
+        "language_model.model.layers.6.mlp.down_proj.scales": "model.safetensors",
+        "language_model.model.layers.6.mlp.down_proj.weight": "model.safetensors",
+        "language_model.model.layers.6.mlp.gate_proj.biases": "model.safetensors",
+        "language_model.model.layers.6.mlp.gate_proj.scales": "model.safetensors",
+        "language_model.model.layers.6.mlp.gate_proj.weight": "model.safetensors",
+        "language_model.model.layers.6.mlp.up_proj.biases": "model.safetensors",
+        "language_model.model.layers.6.mlp.up_proj.scales": "model.safetensors",
+        "language_model.model.layers.6.mlp.up_proj.weight": "model.safetensors",
+        "language_model.model.layers.6.post_attention_layernorm.weight": "model.safetensors",
+        "language_model.model.layers.7.input_layernorm.weight": "model.safetensors",
+        "language_model.model.layers.7.mlp.down_proj.biases": "model.safetensors",
+        "language_model.model.layers.7.mlp.down_proj.scales": "model.safetensors",
+        "language_model.model.layers.7.mlp.down_proj.weight": "model.safetensors",
+        "language_model.model.layers.7.mlp.gate_proj.biases": "model.safetensors",
+        "language_model.model.layers.7.mlp.gate_proj.scales": "model.safetensors",
+        "language_model.model.layers.7.mlp.gate_proj.weight": "model.safetensors",
+        "language_model.model.layers.7.mlp.up_proj.biases": "model.safetensors",
+        "language_model.model.layers.7.mlp.up_proj.scales": "model.safetensors",
+        "language_model.model.layers.7.mlp.up_proj.weight": "model.safetensors",
+        "language_model.model.layers.7.post_attention_layernorm.weight": "model.safetensors",
+        "language_model.model.layers.7.self_attn.k_norm.weight": "model.safetensors",
+        "language_model.model.layers.7.self_attn.k_proj.biases": "model.safetensors",
+        "language_model.model.layers.7.self_attn.k_proj.scales": "model.safetensors",
+        "language_model.model.layers.7.self_attn.k_proj.weight": "model.safetensors",
+        "language_model.model.layers.7.self_attn.o_proj.biases": "model.safetensors",
+        "language_model.model.layers.7.self_attn.o_proj.scales": "model.safetensors",
+        "language_model.model.layers.7.self_attn.o_proj.weight": "model.safetensors",
+        "language_model.model.layers.7.self_attn.q_norm.weight": "model.safetensors",
+        "language_model.model.layers.7.self_attn.q_proj.biases": "model.safetensors",
+        "language_model.model.layers.7.self_attn.q_proj.scales": "model.safetensors",
+        "language_model.model.layers.7.self_attn.q_proj.weight": "model.safetensors",
+        "language_model.model.layers.7.self_attn.v_proj.biases": "model.safetensors",
+        "language_model.model.layers.7.self_attn.v_proj.scales": "model.safetensors",
+        "language_model.model.layers.7.self_attn.v_proj.weight": "model.safetensors",
+        "language_model.model.layers.8.input_layernorm.weight": "model.safetensors",
+        "language_model.model.layers.8.linear_attn.A_log": "model.safetensors",
+        "language_model.model.layers.8.linear_attn.conv1d.weight": "model.safetensors",
+        "language_model.model.layers.8.linear_attn.dt_bias": "model.safetensors",
+        "language_model.model.layers.8.linear_attn.in_proj_a.biases": "model.safetensors",
+        "language_model.model.layers.8.linear_attn.in_proj_a.scales": "model.safetensors",
+        "language_model.model.layers.8.linear_attn.in_proj_a.weight": "model.safetensors",
+        "language_model.model.layers.8.linear_attn.in_proj_b.biases": "model.safetensors",
+        "language_model.model.layers.8.linear_attn.in_proj_b.scales": "model.safetensors",
+        "language_model.model.layers.8.linear_attn.in_proj_b.weight": "model.safetensors",
+        "language_model.model.layers.8.linear_attn.in_proj_qkv.biases": "model.safetensors",
+        "language_model.model.layers.8.linear_attn.in_proj_qkv.scales": "model.safetensors",
+        "language_model.model.layers.8.linear_attn.in_proj_qkv.weight": "model.safetensors",
+        "language_model.model.layers.8.linear_attn.in_proj_z.biases": "model.safetensors",
+        "language_model.model.layers.8.linear_attn.in_proj_z.scales": "model.safetensors",
+        "language_model.model.layers.8.linear_attn.in_proj_z.weight": "model.safetensors",
+        "language_model.model.layers.8.linear_attn.norm.weight": "model.safetensors",
+        "language_model.model.layers.8.linear_attn.out_proj.biases": "model.safetensors",
+        "language_model.model.layers.8.linear_attn.out_proj.scales": "model.safetensors",
+        "language_model.model.layers.8.linear_attn.out_proj.weight": "model.safetensors",
+        "language_model.model.layers.8.mlp.down_proj.biases": "model.safetensors",
+        "language_model.model.layers.8.mlp.down_proj.scales": "model.safetensors",
+        "language_model.model.layers.8.mlp.down_proj.weight": "model.safetensors",
+        "language_model.model.layers.8.mlp.gate_proj.biases": "model.safetensors",
+        "language_model.model.layers.8.mlp.gate_proj.scales": "model.safetensors",
+        "language_model.model.layers.8.mlp.gate_proj.weight": "model.safetensors",
+        "language_model.model.layers.8.mlp.up_proj.biases": "model.safetensors",
+        "language_model.model.layers.8.mlp.up_proj.scales": "model.safetensors",
+        "language_model.model.layers.8.mlp.up_proj.weight": "model.safetensors",
+        "language_model.model.layers.8.post_attention_layernorm.weight": "model.safetensors",
+        "language_model.model.layers.9.input_layernorm.weight": "model.safetensors",
+        "language_model.model.layers.9.linear_attn.A_log": "model.safetensors",
+        "language_model.model.layers.9.linear_attn.conv1d.weight": "model.safetensors",
+        "language_model.model.layers.9.linear_attn.dt_bias": "model.safetensors",
+        "language_model.model.layers.9.linear_attn.in_proj_a.biases": "model.safetensors",
+        "language_model.model.layers.9.linear_attn.in_proj_a.scales": "model.safetensors",
+        "language_model.model.layers.9.linear_attn.in_proj_a.weight": "model.safetensors",
+        "language_model.model.layers.9.linear_attn.in_proj_b.biases": "model.safetensors",
+        "language_model.model.layers.9.linear_attn.in_proj_b.scales": "model.safetensors",
+        "language_model.model.layers.9.linear_attn.in_proj_b.weight": "model.safetensors",
+        "language_model.model.layers.9.linear_attn.in_proj_qkv.biases": "model.safetensors",
+        "language_model.model.layers.9.linear_attn.in_proj_qkv.scales": "model.safetensors",
+        "language_model.model.layers.9.linear_attn.in_proj_qkv.weight": "model.safetensors",
+        "language_model.model.layers.9.linear_attn.in_proj_z.biases": "model.safetensors",
+        "language_model.model.layers.9.linear_attn.in_proj_z.scales": "model.safetensors",
+        "language_model.model.layers.9.linear_attn.in_proj_z.weight": "model.safetensors",
+        "language_model.model.layers.9.linear_attn.norm.weight": "model.safetensors",
+        "language_model.model.layers.9.linear_attn.out_proj.biases": "model.safetensors",
+        "language_model.model.layers.9.linear_attn.out_proj.scales": "model.safetensors",
+        "language_model.model.layers.9.linear_attn.out_proj.weight": "model.safetensors",
+        "language_model.model.layers.9.mlp.down_proj.biases": "model.safetensors",
+        "language_model.model.layers.9.mlp.down_proj.scales": "model.safetensors",
+        "language_model.model.layers.9.mlp.down_proj.weight": "model.safetensors",
+        "language_model.model.layers.9.mlp.gate_proj.biases": "model.safetensors",
+        "language_model.model.layers.9.mlp.gate_proj.scales": "model.safetensors",
+        "language_model.model.layers.9.mlp.gate_proj.weight": "model.safetensors",
+        "language_model.model.layers.9.mlp.up_proj.biases": "model.safetensors",
+        "language_model.model.layers.9.mlp.up_proj.scales": "model.safetensors",
+        "language_model.model.layers.9.mlp.up_proj.weight": "model.safetensors",
+        "language_model.model.layers.9.post_attention_layernorm.weight": "model.safetensors",
+        "language_model.model.norm.weight": "model.safetensors"
+    }
+}

8bit/tokenizer.json

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:06b9509352d2af50381ab2247e083b80d32d5c0aba91c272ca9ff729b6a0e523
+size 19989325

8bit/tokenizer_config.json

@@ -0,0 +1,33 @@
+{
+  "add_prefix_space": false,
+  "audio_bos_token": "<|audio_start|>",
+  "audio_eos_token": "<|audio_end|>",
+  "audio_token": "<|audio_pad|>",
+  "backend": "tokenizers",
+  "bos_token": null,
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "<|im_end|>",
+  "errors": "replace",
+  "image_token": "<|image_pad|>",
+  "is_local": true,
+  "local_files_only": false,
+  "model_max_length": 262144,
+  "model_specific_special_tokens": {
+    "audio_bos_token": "<|audio_start|>",
+    "audio_eos_token": "<|audio_end|>",
+    "audio_token": "<|audio_pad|>",
+    "image_token": "<|image_pad|>",
+    "video_token": "<|video_pad|>",
+    "vision_bos_token": "<|vision_start|>",
+    "vision_eos_token": "<|vision_end|>"
+  },
+  "pad_token": "<|endoftext|>",
+  "pretokenize_regex": "(?i:'s|'t|'re|'ve|'m|'ll|'d)|[^\\r\\n\\p{L}\\p{N}]?[\\p{L}\\p{M}]+|\\p{N}| ?[^\\s\\p{L}\\p{M}\\p{N}]+[\\r\\n]*|\\s*[\\r\\n]+|\\s+(?!\\S)|\\s+",
+  "split_special_tokens": false,
+  "tokenizer_class": "Qwen2Tokenizer",
+  "tool_parser_type": "qwen3_coder",
+  "unk_token": null,
+  "video_token": "<|video_pad|>",
+  "vision_bos_token": "<|vision_start|>",
+  "vision_eos_token": "<|vision_end|>"
+}

LICENSE

@@ -0,0 +1,202 @@
+
+                                 Apache License
+                           Version 2.0, January 2004
+                        http://www.apache.org/licenses/
+
+   TERMS AND CONDITIONS FOR USE, REPRODUCTION, AND DISTRIBUTION
+
+   1. Definitions.
+
+      "License" shall mean the terms and conditions for use, reproduction,
+      and distribution as defined by Sections 1 through 9 of this document.
+
+      "Licensor" shall mean the copyright owner or entity authorized by
+      the copyright owner that is granting the License.
+
+      "Legal Entity" shall mean the union of the acting entity and all
+      other entities that control, are controlled by, or are under common
+      control with that entity. For the purposes of this definition,
+      "control" means (i) the power, direct or indirect, to cause the
+      direction or management of such entity, whether by contract or
+      otherwise, or (ii) ownership of fifty percent (50%) or more of the
+      outstanding shares, or (iii) beneficial ownership of such entity.
+
+      "You" (or "Your") shall mean an individual or Legal Entity
+      exercising permissions granted by this License.
+
+      "Source" form shall mean the preferred form for making modifications,
+      including but not limited to software source code, documentation
+      source, and configuration files.
+
+      "Object" form shall mean any form resulting from mechanical
+      transformation or translation of a Source form, including but
+      not limited to compiled object code, generated documentation,
+      and conversions to other media types.
+
+      "Work" shall mean the work of authorship, whether in Source or
+      Object form, made available under the License, as indicated by a
+      copyright notice that is included in or attached to the work
+      (an example is provided in the Appendix below).
+
+      "Derivative Works" shall mean any work, whether in Source or Object
+      form, that is based on (or derived from) the Work and for which the
+      editorial revisions, annotations, elaborations, or other modifications
+      represent, as a whole, an original work of authorship. For the purposes
+      of this License, Derivative Works shall not include works that remain
+      separable from, or merely link (or bind by name) to the interfaces of,
+      the Work and Derivative Works thereof.
+
+      "Contribution" shall mean any work of authorship, including
+      the original version of the Work and any modifications or additions
+      to that Work or Derivative Works thereof, that is intentionally
+      submitted to Licensor for inclusion in the Work by the copyright owner
+      or by an individual or Legal Entity authorized to submit on behalf of
+      the copyright owner. For the purposes of this definition, "submitted"
+      means any form of electronic, verbal, or written communication sent
+      to the Licensor or its representatives, including but not limited to
+      communication on electronic mailing lists, source code control systems,
+      and issue tracking systems that are managed by, or on behalf of, the
+      Licensor for the purpose of discussing and improving the Work, but
+      excluding communication that is conspicuously marked or otherwise
+      designated in writing by the copyright owner as "Not a Contribution."
+
+      "Contributor" shall mean Licensor and any individual or Legal Entity
+      on behalf of whom a Contribution has been received by Licensor and
+      subsequently incorporated within the Work.
+
+   2. Grant of Copyright License. Subject to the terms and conditions of
+      this License, each Contributor hereby grants to You a perpetual,
+      worldwide, non-exclusive, no-charge, royalty-free, irrevocable
+      copyright license to reproduce, prepare Derivative Works of,
+      publicly display, publicly perform, sublicense, and distribute the
+      Work and such Derivative Works in Source or Object form.
+
+   3. Grant of Patent License. Subject to the terms and conditions of
+      this License, each Contributor hereby grants to You a perpetual,
+      worldwide, non-exclusive, no-charge, royalty-free, irrevocable
+      (except as stated in this section) patent license to make, have made,
+      use, offer to sell, sell, import, and otherwise transfer the Work,
+      where such license applies only to those patent claims licensable
+      by such Contributor that are necessarily infringed by their
+      Contribution(s) alone or by combination of their Contribution(s)
+      with the Work to which such Contribution(s) was submitted. If You
+      institute patent litigation against any entity (including a
+      cross-claim or counterclaim in a lawsuit) alleging that the Work
+      or a Contribution incorporated within the Work constitutes direct
+      or contributory patent infringement, then any patent licenses
+      granted to You under this License for that Work shall terminate
+      as of the date such litigation is filed.
+
+   4. Redistribution. You may reproduce and distribute copies of the
+      Work or Derivative Works thereof in any medium, with or without
+      modifications, and in Source or Object form, provided that You
+      meet the following conditions:
+
+      (a) You must give any other recipients of the Work or
+          Derivative Works a copy of this License; and
+
+      (b) You must cause any modified files to carry prominent notices
+          stating that You changed the files; and
+
+      (c) You must retain, in the Source form of any Derivative Works
+          that You distribute, all copyright, patent, trademark, and
+          attribution notices from the Source form of the Work,
+          excluding those notices that do not pertain to any part of
+          the Derivative Works; and
+
+      (d) If the Work includes a "NOTICE" text file as part of its
+          distribution, then any Derivative Works that You distribute must
+          include a readable copy of the attribution notices contained
+          within such NOTICE file, excluding those notices that do not
+          pertain to any part of the Derivative Works, in at least one
+          of the following places: within a NOTICE text file distributed
+          as part of the Derivative Works; within the Source form or
+          documentation, if provided along with the Derivative Works; or,
+          within a display generated by the Derivative Works, if and
+          wherever such third-party notices normally appear. The contents
+          of the NOTICE file are for informational purposes only and
+          do not modify the License. You may add Your own attribution
+          notices within Derivative Works that You distribute, alongside
+          or as an addendum to the NOTICE text from the Work, provided
+          that such additional attribution notices cannot be construed
+          as modifying the License.
+
+      You may add Your own copyright statement to Your modifications and
+      may provide additional or different license terms and conditions
+      for use, reproduction, or distribution of Your modifications, or
+      for any such Derivative Works as a whole, provided Your use,
+      reproduction, and distribution of the Work otherwise complies with
+      the conditions stated in this License.
+
+   5. Submission of Contributions. Unless You explicitly state otherwise,
+      any Contribution intentionally submitted for inclusion in the Work
+      by You to the Licensor shall be under the terms and conditions of
+      this License, without any additional terms or conditions.
+      Notwithstanding the above, nothing herein shall supersede or modify
+      the terms of any separate license agreement you may have executed
+      with Licensor regarding such Contributions.
+
+   6. Trademarks. This License does not grant permission to use the trade
+      names, trademarks, service marks, or product names of the Licensor,
+      except as required for reasonable and customary use in describing the
+      origin of the Work and reproducing the content of the NOTICE file.
+
+   7. Disclaimer of Warranty. Unless required by applicable law or
+      agreed to in writing, Licensor provides the Work (and each
+      Contributor provides its Contributions) on an "AS IS" BASIS,
+      WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or
+      implied, including, without limitation, any warranties or conditions
+      of TITLE, NON-INFRINGEMENT, MERCHANTABILITY, or FITNESS FOR A
+      PARTICULAR PURPOSE. You are solely responsible for determining the
+      appropriateness of using or redistributing the Work and assume any
+      risks associated with Your exercise of permissions under this License.
+
+   8. Limitation of Liability. In no event and under no legal theory,
+      whether in tort (including negligence), contract, or otherwise,
+      unless required by applicable law (such as deliberate and grossly
+      negligent acts) or agreed to in writing, shall any Contributor be
+      liable to You for damages, including any direct, indirect, special,
+      incidental, or consequential damages of any character arising as a
+      result of this License or out of the use or inability to use the
+      Work (including but not limited to damages for loss of goodwill,
+      work stoppage, computer failure or malfunction, or any and all
+      other commercial damages or losses), even if such Contributor
+      has been advised of the possibility of such damages.
+
+   9. Accepting Warranty or Additional Liability. While redistributing
+      the Work or Derivative Works thereof, You may choose to offer,
+      and charge a fee for, acceptance of support, warranty, indemnity,
+      or other liability obligations and/or rights consistent with this
+      License. However, in accepting such obligations, You may act only
+      on Your own behalf and on Your sole responsibility, not on behalf
+      of any other Contributor, and only if You agree to indemnify,
+      defend, and hold each Contributor harmless for any liability
+      incurred by, or claims asserted against, such Contributor by reason
+      of your accepting any such warranty or additional liability.
+
+   END OF TERMS AND CONDITIONS
+
+   APPENDIX: How to apply the Apache License to your work.
+
+      To apply the Apache License to your work, attach the following
+      boilerplate notice, with the fields enclosed by brackets "[]"
+      replaced with your own identifying information. (Don't include
+      the brackets!)  The text should be enclosed in the appropriate
+      comment syntax for the file format. We also recommend that a
+      file or class name and description of purpose be included on the
+      same "printed page" as the copyright notice for easier
+      identification within third-party archives.
+
+   Copyright 2026 Alibaba Cloud
+
+   Licensed under the Apache License, Version 2.0 (the "License");
+   you may not use this file except in compliance with the License.
+   You may obtain a copy of the License at
+
+       http://www.apache.org/licenses/LICENSE-2.0
+
+   Unless required by applicable law or agreed to in writing, software
+   distributed under the License is distributed on an "AS IS" BASIS,
+   WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+   See the License for the specific language governing permissions and
+   limitations under the License.

NOTICE

@@ -0,0 +1,35 @@
+chaoliangUNSW/Jev-Style-2B-Decision-v3-MLX
+Copyright 2026 chaoliangUNSW. Licensed under the Apache License, Version 2.0 (see LICENSE).
+
+This model is a fine-tuned derivative of Qwen3.5-2B (https://huggingface.co/Qwen/Qwen3.5-2B,
+revision 15852e8c16360a2fea060d615a32b45270f8a8fc), Copyright 2026 Alibaba Cloud, licensed under the Apache
+License, Version 2.0. The LICENSE file in this repository is the license file distributed with Qwen3.5-2B.
+
+Modifications relative to Qwen3.5-2B:
+- all text-model weights were fine-tuned (full fine-tuning) to score typed decision questions
+  (choice / score / true-false) with a verdict readout: logit(" yes") - logit(" no") at one " ->" slot per
+  option, under a block-causal attention protocol (2,048-token blocks); one global calibration temperature was
+  fitted afterwards (readout_config.json);
+- the vision tower (model.visual.*) and the multi-token-prediction head (mtp.*) were removed; the checkpoint
+  is a text-only Qwen3_5ForCausalLM with tied input/output embeddings;
+- converted to MLX format with mlx-lm 0.31.3 / mlx 0.32.2 (https://github.com/ml-explore/mlx-lm,
+  MIT License, Copyright Apple Inc.): bf16/ = bfloat16, 8bit/ = affine 8-bit quantisation, group size 64;
+  each folder also holds macjev_norms_fp32.safetensors (the exact FP32 (1 + w) RMSNorm weights);
+- the runtime script contains a modified copy of mlx-lm 0.31.3's qwen3_5 GatedDeltaNet.__call__ (MIT License,
+  Copyright Apple Inc.; the q/k normalisation epsilon changed and the sharding branches removed); the MIT
+  copyright and permission notice is reproduced in THIRD_PARTY_NOTICES.md;
+- added the runtime script, readout/release configuration files, the integrity manifest and this NOTICE.
+
+The question types (choice / score / noul) follow the typed-decision convention of Laya
+(https://github.com/NandhaKishorM/laya, Apache-2.0) so both models can be evaluated on the same
+inputs. No Laya code or weights are included.
+
+Part of the third generation (v3) of the Jev-Style decision series (v3 also includes
+chaoliangUNSW/Jev-Style-0.8B-Decision-v3). Earlier generations: v1 = chaoliangUNSW/Jev-Style-Qwen3.5-2B-Decision
+(public GGUF release: chaoliangUNSW/Jev-Style-Qwen3.5-2B-Decision-GGUF) and v2 =
+chaoliangUNSW/Jev-Style-Qwen3.5-2B-Decision-v2, both built on Qwen3.5-2B-Base. Jev-Style-2B-Decision-v3 is
+a full fine-tune of Qwen/Qwen3.5-2B.
+
+Not affiliated with, endorsed by or connected to TypeSafe or Jev. "Jev-Style" only describes the kind
+of model (a small typed-decision model in a similar style); no Jev weights, code or outputs are included.
+Not affiliated with or endorsed by Alibaba Cloud / the Qwen team or the Laya authors.

README.md

@@ -0,0 +1,194 @@
+---
+license: apache-2.0
+base_model: chaoliangUNSW/Jev-Style-2B-Decision-v3
+base_model_relation: quantized
+library_name: mlx
+pipeline_tag: text-classification
+tags:
+  - decision-model
+  - decision-making
+  - jev-style
+  - system-one
+  - calibration
+  - long-context
+  - qwen3.5
+  - mlx
+  - apple-silicon
+  - on-device
+  - llm-routing
+  - guardrails
+---
+
+# Jev-Style-2B-Decision-v3-MLX
+
+**[Try it in your browser →](https://huggingface.co/spaces/chaoliangUNSW/jev-style-2b)**
+
+**Website:** [jevstyle.com](https://jevstyle.com/#v3-2b) · **GitHub:** [jev-style](https://github.com/lawrence3699/jev-style) · **Collection:** [all v3 builds and demos](https://huggingface.co/collections/chaoliangUNSW/jev-style-decision-v3-08b-2b-6ab87f32380cbd8c03b608b9)
+
+**Jev-Style decision series:** [v1 · 2B](https://huggingface.co/chaoliangUNSW/Jev-Style-Qwen3.5-2B-Decision-MLX-bf16) → [v2 · 2B](https://huggingface.co/chaoliangUNSW/Jev-Style-Qwen3.5-2B-Decision-v2-MLX-bf16) → [v3 · 0.8B](https://huggingface.co/chaoliangUNSW/Jev-Style-0.8B-Decision-v3-MLX) → **v3 · 2B**
+
+<!-- PIP_SNIPPET_AFTER_0.3.0 -->
+
+**Jev-style decisions, now at 2B, native on Apple silicon.** These are the MLX builds of
+[Jev-Style-2B-Decision-v3](https://huggingface.co/chaoliangUNSW/Jev-Style-2B-Decision-v3): bf16 (3.76 GB) and
+8-bit (2.00 GB) in one repository, with one runtime (`jev_style_decision_mlx.py`; choose the folder with `--model-dir bf16` or `--model-dir 8bit`). Full
+results, protocols, training data and licences are on the
+[main model card](https://huggingface.co/chaoliangUNSW/Jev-Style-2B-Decision-v3).
+
+![Jev-Style 2B Decision v3: 73.6% on JevBench v1.4.1 public items, the highest among the Qwen3.5-2B-family systems on the board; Jev is ahead at 86.6%; shown are the Qwen3.5-2B-family systems, our 0.8B v3, Laya and Jev, and 42 of the 82 board systems score higher; 25,600 tokens per call with no option cap](figures/banner.png)
+
+| Public benchmark | **Jev-Style v3 · 2B** | Jev-Style v3 · 0.8B | Jev 1.13 (API) |
+|---|:---:|:---:|:---:|
+| JevBench v1.4.1, 231 public items ↑ | 73.6% | 64.1% | 86.6% |
+| tweet_topic, zero-shot, accuracy ↑ | 82.2% | 75.5% | 79.3%¹ |
+| fin_topic, zero-shot, accuracy ↑ | 61.1% | 46.7% | 67.0%¹ |
+| Longest input per call | 25,600 tokens, no option cap | 25,600 tokens | |
+
+<sub>Benchmark numbers of the model, measured with its GGUF F16 build (the pre-declared engine; one global temperature; each benchmark run once). The MLX builds were checked against the same FP32 reference (table below). JevBench: self-run with the official harness, not an official board entry; 95% CI 67.6–78.9% (Wilson). 73.6% is the highest JevBench public accuracy among the Qwen3.5-2B-family systems on the v1.4.1 board (decider-2b 71.0%, open-jev-zefan-2b 64.5%); decider-2b lies inside the CI. Jev is well ahead on JevBench and ahead on fin_topic. ¹ Jev numbers from the elcronos study (raw API), not re-run by us; on tweet_topic the 2B's macro-F1 (67.8%) is below Jev's (69.4%). Details: [main card](https://huggingface.co/chaoliangUNSW/Jev-Style-2B-Decision-v3#results).</sub>
+
+**25,600 tokens, no option cap.** State, questions and every option share one 25,600-token budget; question and
+options over 2,048 tokens use a numbered-option catalogue. Nothing is truncated.
+
+## Precisions and parity
+
+| Precision | Folder | Weights | Same top-1 as PyTorch FP32 (1,000 rows) | Max abs Δp | Accuracy (FP32: 80.8%) | Long fixture (43 questions) | Gate |
+|---|---|---:|---:|---:|---:|---:|:---:|
+| **bf16** (default) | `bf16/` | 3.76 GB | **99.7%** | 0.035 | 80.5% | **43 / 43** | PASS |
+| **8-bit** (affine, group size 64) | `8bit/` | 2.00 GB | **99.6%** | 0.162 | 80.8% | **43 / 43** | PASS |
+
+- Each folder holds `model.safetensors`, its `config.json`, the tokenizer and `macjev_norms_fp32.safetensors`
+  (FP32 norm weights, 0.42 MB, required). The top level holds the shared runtime `jev_style_decision_mlx.py`,
+  `readout_config.json`, `release_config.json`, the sha256 `manifest.json`, `LICENSE`, `NOTICE`,
+  `THIRD_PARTY_NOTICES.md` and a copy of `bf16/config.json` (for the Hub's download counter). Converted with mlx
+  0.32.2 and mlx-lm 0.31.3.
+- The 8-bit build makes the same call as full precision on 99.6% of the 1,000 rows at about half the size.
+- **Which folder:** `bf16/` for closeness to the FP32 reference (max abs Δp 0.035 vs 0.162); `8bit/` when memory is
+  tight.
+
+<sub>Reference: HF FP32 on CPU, exact block attention, on the released bf16 checkpoint; gates declared before any format was scored. 1,000 real development rows (≤4,096 tokens) test agreement between formats. Long fixture: 35 requests / 43 questions up to 25,600 tokens, including catalogue-overflow questions and up to 151 options. Sizes are the weight files (GB = 10^9 bytes).</sub>
+
+<!-- LATENCY:BEGIN generated from latency_2b.json (validation/latency_2b.json in the main repository); do not edit by hand -->
+
+## Speed
+
+**Read once, then ask.** On an Apple M1 Max (MLX bf16), the first question about a 24,501-token input took 15.5 s; a further question about the same state took 0.15 s, because the state is computed once and reused (medians). Ten questions about that state in one call took 16.2 s.
+
+| State | Questions per call | MLX bf16 | MLX 8-bit |
+|---|---:|---:|---:|
+| 878 tokens | 1 | 0.57 s | 0.72 s |
+| 878 tokens | 10 | 1.07 s | 1.27 s |
+| 3,950 tokens | 1 | 2.26 s | 2.96 s |
+| 3,950 tokens | 10 | 2.80 s | 3.54 s |
+| 24,436 tokens | 1 | 15.5 s | 19.9 s |
+| 24,436 tokens | 10 | 16.2 s | 20.8 s |
+| 24,436 tokens, already computed | 1 | 0.15 s | 0.16 s |
+
+- Pick a precision folder by size and closeness to FP32 (Precisions table above). bf16 and 8-bit were timed one after another on the same shared machine; in these runs 8-bit was the slower folder in every row (indicative only).
+
+<sub>Apple M1 Max, 64 GB, macOS 15.7.5. Wall time around one `decide` / `score_many` call (tokenisation included), median of 3 calls with the state recomputed each time; the model was loaded beforehand (loading took 3.1–3.7 s here, not included). States: English documentation and source code of 878, 3,950, 24,436 tokens plus the question; 10 questions = 4 choice, 4 true/false and 2 score questions about the same state in one call; with the question and options each input was up to 943, 4,015 and 24,501 tokens. MLX: mlx 0.32.2 / mlx-lm 0.31.3. Results were identical with and without a precomputed state. Other jobs shared the machine during these runs (1-minute load average 5.5–8.6 at the end of each run), so treat the numbers as indicative. All rows here were measured in one session (2026-09-27 03:21–03:28 AEST); an earlier run of the same rows (00:37–00:45 AEST, load average 11.8–27.5) was discarded because other jobs had slowed it (its times were up to 2.1× longer).</sub>
+
+<!-- LATENCY:END -->
+
+## Quick start
+
+```bash
+pip install -U huggingface_hub
+hf download chaoliangUNSW/Jev-Style-2B-Decision-v3-MLX --local-dir Jev-Style-2B-Decision-v3-MLX
+cd Jev-Style-2B-Decision-v3-MLX
+pip install -r requirements.txt        # mlx 0.32.2, mlx-lm 0.31.3 (exactly), transformers 5.17.0, tokenizers, numpy
+```
+
+Only one precision is needed: add `--exclude "8bit/*"` to the download for bf16 only, or `--exclude "bf16/*"` for
+8-bit only. Pass the precision folder to the runtime with `--model-dir bf16` or `--model-dir 8bit` (in Python,
+`JevStyleDecisionMLX("bf16")` or `JevStyleDecisionMLX("8bit")`).
+
+```bash
+# choice question, bf16 weights
+python jev_style_decision_mlx.py --model-dir bf16 --state "Finder is open on ~/Reports. The user asked: email report.pdf to Ana." \
+  --question "What should the agent do next?" \
+  --options '{"attach_report": "attach report.pdf to a new email to Ana", "rename_report": null, "close_finder": "close the Finder window"}'
+
+# true/false question about a JSON state, 8-bit weights
+python jev_style_decision_mlx.py --model-dir 8bit \
+  --state-json '{"app": "Mail", "draft": {"to": "ana@example.com", "attachments": []}}' \
+  --question "Is the report attached?" --qtype noul
+
+# score (ordinal) question, uncalibrated probabilities (T = 1.0)
+python jev_style_decision_mlx.py --model-dir 8bit --state "The deploy failed twice with the same migration error." \
+  --question "How risky is retrying now?" --qtype score --options '["no risk", "some risk", "high risk"]' --temperature 1.0
+
+# full typed question as JSON, first checking the sha256 of the runtime files and the chosen precision folder (manifest.json)
+python jev_style_decision_mlx.py --model-dir bf16 --verify --state "cart: 3 items" \
+  --question '{"t": "choice", "ins": "Next step?", "crit": {"checkout": null, "keep_shopping": null}}'
+```
+
+The output has `answer`, `probabilities` and raw `scores` per option, `temperature` (0.8279, the calibrated global
+temperature), `top_probability`, `entropy_concentration`, `input_tokens`, `state_tokens`, `head_tokens`, `blocks`
+and `catalogue_overflow`.
+
+```python
+from jev_style_decision_mlx import JevStyleDecisionMLX
+m = JevStyleDecisionMLX("8bit")                # the 8-bit precision folder
+state = {"screen": "Settings > Wi-Fi", "goal": "join the network Office-5G"}
+qs = [{"t": "choice", "ins": "Next action?", "crit": {"tap_office_5g": None, "toggle_wifi_off": None, "go_back": None}},
+      {"t": "noul", "ins": "Is Wi-Fi turned on?", "crit": None}]
+for r in m.score_many(state, qs):          # the state is computed once and reused for every question
+    print(r["answer"], round(r["top_probability"], 4))
+m.decide(state, qs[0])                     # same state again: reused (m.last_timing["state_reused"] is True)
+m.close()
+```
+
+- Batch mode: `--jsonl requests.jsonl` (or `-` for stdin), rows `{"id"?, "state", "question", "options"?, "qtype"?,
+  "temperature"?}`; consecutive rows with an identical state share one state computation, and a malformed row gets
+  an `"error"` field while the batch continues.
+- **Long inputs.** The complete input can be up to 25,600 tokens, with no separate question/options cap. In our
+  smoke test a 22,005-token state with 100 described options ran as 25,393 tokens in 14 blocks, using the
+  catalogue layout. A larger input raises `InputBudgetError` (CLI exit status 2); nothing is truncated.
+- **mlx-lm must be exactly 0.31.3.** The runtime corrects two Qwen3.5 numerics inside mlx-lm and checks the source
+  it patches; other versions are refused with the install command. It also refuses to run without
+  `macjev_norms_fp32.safetensors` or `readout_config.json`, and never falls back to stock norms or T = 1.
+- The input format, block attention and the readout are described on the
+  [main card](https://huggingface.co/chaoliangUNSW/Jev-Style-2B-Decision-v3#input-format-and-readout).
+
+## Scope and limits
+
+- **Runtime required.** `mlx_lm.generate` can load the weights in `bf16/` or `8bit/` (the
+  repository root holds no weights) but still cannot produce the decision scores, and it would run causal attention.
+  Use `jev_style_decision_mlx.py`.
+- **25,600 tokens** is the limit for the whole input (state + question + options + readout).
+- **Reduced-data training.** The model was trained on a reduced data pool (60M tokens).
+- **Decision Index.** Not run by us. The training pool includes the train splits of 7 of its benchmarks and
+  format-imitating data for 7 more, so results on these are not zero-shot; the 14 benchmarks that are not zero-shot
+  for this model are named on the [main card](https://huggingface.co/chaoliangUNSW/Jev-Style-2B-Decision-v3#scope-and-limits).
+
+<details>
+<summary><strong>Results charts</strong></summary>
+
+![JevBench v1.4.1 public accuracy: 2B v3 vs the Qwen3.5-2B-family systems, 0.8B v3 and Laya, with Jev as a reference line](figures/jevbench.png)
+
+![Zero-shot tweet_topic and fin_topic accuracy: 2B v3 vs 0.8B v3 and Jev](figures/zeroshot.png)
+
+<sub>Protocol notes are under each chart and on the main card. Plotted values and sources: [jevbench.data.json](figures/jevbench.data.json), [zeroshot.data.json](figures/zeroshot.data.json).</sub>
+
+</details>
+
+## Disclaimers and licence
+
+Apache-2.0. Built on Qwen/Qwen3.5-2B (Apache-2.0); `NOTICE` lists the modifications. Some training data has
+restrictive or unclear terms (for example research-only jailbreak prompts and share-alike CC BY-SA sources), and some
+training rows are outputs of OpenAI GPT and Anthropic Claude models, whose providers' terms of use may restrict how
+models trained on them may be used. See [Training data and
+licences](https://huggingface.co/chaoliangUNSW/Jev-Style-2B-Decision-v3#training-data-and-licences) on the main card.
+The runtime contains a modified copy of one mlx-lm 0.31.3 function (the Qwen3.5 `GatedDeltaNet.__call__`; MIT License,
+Copyright Apple Inc.); its copyright and permission notice is in `THIRD_PARTY_NOTICES.md`. Not affiliated with,
+endorsed by or connected to TypeSafe AI or Jev (no Jev weights, code or outputs are used), the Laya authors or the
+Qwen team.
+
+**AI disclosure:** code written with AI coding assistants (Claude Code) under my direction; I designed the project,
+trained the models and verified the results.
+
+## Contact
+
+I welcome internship, employment, and research collaboration opportunities. Please contact me at [**yanchaoliang369@gmail.com**](mailto:yanchaoliang369@gmail.com).
+
+欢迎提供实习、工作及科研合作机会，请邮件联系：[yanchaoliang369@gmail.com](mailto:yanchaoliang369@gmail.com)。

THIRD_PARTY_NOTICES.md

@@ -0,0 +1,41 @@
+# Third-party notices
+
+This file covers third-party code that is **contained in** this repository. Libraries that are only imported at
+run time (mlx, mlx-lm, tokenizers, numpy) are not redistributed here and keep their own licences.
+
+## mlx-lm 0.31.3 (MIT License)
+
+`jev_style_decision_mlx.py` contains a modified copy of `GatedDeltaNet.__call__` from
+`mlx_lm/models/qwen3_5.py` of mlx-lm 0.31.3 (https://github.com/ml-explore/mlx-lm; that source file carries the
+header "Copyright © 2026 Apple Inc."). In the runtime it is the function `_gdn_hf_call`. Changes made:
+
+- the q/k normalisation epsilon: `mx.fast.rms_norm(x, None, 1e-6)` became `mx.fast.rms_norm(x, None, 1e-6 / Dk)`,
+  so that q/k are l2-normalised with eps 1e-6 as in Hugging Face transformers / llama.cpp;
+- the tensor-parallel (sharding) branches were removed: a sharded layer is refused with an error;
+- rewritten as a module-level function (installed per model instance, never globally) and reformatted.
+
+The mlx-lm licence (the LICENSE file distributed with mlx-lm 0.31.3), reproduced verbatim:
+
+```text
+MIT License
+
+Copyright © 2023 Apple Inc.
+
+Permission is hereby granted, free of charge, to any person obtaining a copy
+of this software and associated documentation files (the "Software"), to deal
+in the Software without restriction, including without limitation the rights
+to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+copies of the Software, and to permit persons to whom the Software is
+furnished to do so, subject to the following conditions:
+
+The above copyright notice and this permission notice shall be included in all
+copies or substantial portions of the Software.
+
+THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+SOFTWARE.
+```

bf16/chat_template.jinja

@@ -0,0 +1,154 @@
+{%- set image_count = namespace(value=0) %}
+{%- set video_count = namespace(value=0) %}
+{%- macro render_content(content, do_vision_count, is_system_content=false) %}
+    {%- if content is string %}
+        {{- content }}
+    {%- elif content is iterable and content is not mapping %}
+        {%- for item in content %}
+            {%- if 'image' in item or 'image_url' in item or item.type == 'image' %}
+                {%- if is_system_content %}
+                    {{- raise_exception('System message cannot contain images.') }}
+                {%- endif %}
+                {%- if do_vision_count %}
+                    {%- set image_count.value = image_count.value + 1 %}
+                {%- endif %}
+                {%- if add_vision_id %}
+                    {{- 'Picture ' ~ image_count.value ~ ': ' }}
+                {%- endif %}
+                {{- '<|vision_start|><|image_pad|><|vision_end|>' }}
+            {%- elif 'video' in item or item.type == 'video' %}
+                {%- if is_system_content %}
+                    {{- raise_exception('System message cannot contain videos.') }}
+                {%- endif %}
+                {%- if do_vision_count %}
+                    {%- set video_count.value = video_count.value + 1 %}
+                {%- endif %}
+                {%- if add_vision_id %}
+                    {{- 'Video ' ~ video_count.value ~ ': ' }}
+                {%- endif %}
+                {{- '<|vision_start|><|video_pad|><|vision_end|>' }}
+            {%- elif 'text' in item %}
+                {{- item.text }}
+            {%- else %}
+                {{- raise_exception('Unexpected item type in content.') }}
+            {%- endif %}
+        {%- endfor %}
+    {%- elif content is none or content is undefined %}
+        {{- '' }}
+    {%- else %}
+        {{- raise_exception('Unexpected content type.') }}
+    {%- endif %}
+{%- endmacro %}
+{%- if not messages %}
+    {{- raise_exception('No messages provided.') }}
+{%- endif %}
+{%- if tools and tools is iterable and tools is not mapping %}
+    {{- '<|im_start|>system\n' }}
+    {{- "# Tools\n\nYou have access to the following functions:\n\n<tools>" }}
+    {%- for tool in tools %}
+        {{- "\n" }}
+        {{- tool | tojson }}
+    {%- endfor %}
+    {{- "\n</tools>" }}
+    {{- '\n\nIf you choose to call a function ONLY reply in the following format with NO suffix:\n\n<tool_call>\n<function=example_function_name>\n<parameter=example_parameter_1>\nvalue_1\n</parameter>\n<parameter=example_parameter_2>\nThis is the value for the second parameter\nthat can span\nmultiple lines\n</parameter>\n</function>\n</tool_call>\n\n<IMPORTANT>\nReminder:\n- Function calls MUST follow the specified format: an inner <function=...></function> block must be nested within <tool_call></tool_call> XML tags\n- Required parameters MUST be specified\n- You may provide optional reasoning for your function call in natural language BEFORE the function call, but NOT after\n- If there is no function call available, answer the question like normal with your current knowledge and do not tell the user about function calls\n</IMPORTANT>' }}
+    {%- if messages[0].role == 'system' %}
+        {%- set content = render_content(messages[0].content, false, true)|trim %}
+        {%- if content %}
+            {{- '\n\n' + content }}
+        {%- endif %}
+    {%- endif %}
+    {{- '<|im_end|>\n' }}
+{%- else %}
+    {%- if messages[0].role == 'system' %}
+        {%- set content = render_content(messages[0].content, false, true)|trim %}
+        {{- '<|im_start|>system\n' + content + '<|im_end|>\n' }}
+    {%- endif %}
+{%- endif %}
+{%- set ns = namespace(multi_step_tool=true, last_query_index=messages|length - 1) %}
+{%- for message in messages[::-1] %}
+    {%- set index = (messages|length - 1) - loop.index0 %}
+    {%- if ns.multi_step_tool and message.role == "user" %}
+        {%- set content = render_content(message.content, false)|trim %}
+        {%- if not(content.startswith('<tool_response>') and content.endswith('</tool_response>')) %}
+            {%- set ns.multi_step_tool = false %}
+            {%- set ns.last_query_index = index %}
+        {%- endif %}
+    {%- endif %}
+{%- endfor %}
+{%- if ns.multi_step_tool %}
+    {{- raise_exception('No user query found in messages.') }}
+{%- endif %}
+{%- for message in messages %}
+    {%- set content = render_content(message.content, true)|trim %}
+    {%- if message.role == "system" %}
+        {%- if not loop.first %}
+            {{- raise_exception('System message must be at the beginning.') }}
+        {%- endif %}
+    {%- elif message.role == "user" %}
+        {{- '<|im_start|>' + message.role + '\n' + content + '<|im_end|>' + '\n' }}
+    {%- elif message.role == "assistant" %}
+        {%- set reasoning_content = '' %}
+        {%- if message.reasoning_content is string %}
+            {%- set reasoning_content = message.reasoning_content %}
+        {%- else %}
+            {%- if '</think>' in content %}
+                {%- set reasoning_content = content.split('</think>')[0].rstrip('\n').split('<think>')[-1].lstrip('\n') %}
+                {%- set content = content.split('</think>')[-1].lstrip('\n') %}
+            {%- endif %}
+        {%- endif %}
+        {%- set reasoning_content = reasoning_content|trim %}
+        {%- if loop.index0 > ns.last_query_index %}
+            {{- '<|im_start|>' + message.role + '\n<think>\n' + reasoning_content + '\n</think>\n\n' + content }}
+        {%- else %}
+            {{- '<|im_start|>' + message.role + '\n' + content }}
+        {%- endif %}
+        {%- if message.tool_calls and message.tool_calls is iterable and message.tool_calls is not mapping %}
+            {%- for tool_call in message.tool_calls %}
+                {%- if tool_call.function is defined %}
+                    {%- set tool_call = tool_call.function %}
+                {%- endif %}
+                {%- if loop.first %}
+                    {%- if content|trim %}
+                        {{- '\n\n<tool_call>\n<function=' + tool_call.name + '>\n' }}
+                    {%- else %}
+                        {{- '<tool_call>\n<function=' + tool_call.name + '>\n' }}
+                    {%- endif %}
+                {%- else %}
+                    {{- '\n<tool_call>\n<function=' + tool_call.name + '>\n' }}
+                {%- endif %}
+                {%- if tool_call.arguments is defined %}
+                    {%- for args_name, args_value in tool_call.arguments|items %}
+                        {{- '<parameter=' + args_name + '>\n' }}
+                        {%- set args_value = args_value | tojson | safe if args_value is mapping or (args_value is sequence and args_value is not string) else args_value | string %}
+                        {{- args_value }}
+                        {{- '\n</parameter>\n' }}
+                    {%- endfor %}
+                {%- endif %}
+                {{- '</function>\n</tool_call>' }}
+            {%- endfor %}
+        {%- endif %}
+        {{- '<|im_end|>\n' }}
+    {%- elif message.role == "tool" %}
+        {%- if loop.previtem and loop.previtem.role != "tool" %}
+            {{- '<|im_start|>user' }}
+        {%- endif %}
+        {{- '\n<tool_response>\n' }}
+        {{- content }}
+        {{- '\n</tool_response>' }}
+        {%- if not loop.last and loop.nextitem.role != "tool" %}
+            {{- '<|im_end|>\n' }}
+        {%- elif loop.last %}
+            {{- '<|im_end|>\n' }}
+        {%- endif %}
+    {%- else %}
+        {{- raise_exception('Unexpected message role.') }}
+    {%- endif %}
+{%- endfor %}
+{%- if add_generation_prompt %}
+    {{- '<|im_start|>assistant\n' }}
+    {%- if enable_thinking is defined and enable_thinking is true %}
+        {{- '<think>\n' }}
+    {%- else %}
+        {{- '<think>\n\n</think>\n\n' }}
+    {%- endif %}
+{%- endif %}

bf16/config.json

@@ -0,0 +1,75 @@
+{
+    "architectures": [
+        "Qwen3_5ForCausalLM"
+    ],
+    "attention_bias": false,
+    "attention_dropout": 0.0,
+    "attn_output_gate": true,
+    "bos_token_id": null,
+    "dtype": "bfloat16",
+    "eos_token_id": 248044,
+    "full_attention_interval": 4,
+    "head_dim": 256,
+    "hidden_act": "silu",
+    "hidden_size": 2048,
+    "initializer_range": 0.02,
+    "intermediate_size": 6144,
+    "layer_types": [
+        "linear_attention",
+        "linear_attention",
+        "linear_attention",
+        "full_attention",
+        "linear_attention",
+        "linear_attention",
+        "linear_attention",
+        "full_attention",
+        "linear_attention",
+        "linear_attention",
+        "linear_attention",
+        "full_attention",
+        "linear_attention",
+        "linear_attention",
+        "linear_attention",
+        "full_attention",
+        "linear_attention",
+        "linear_attention",
+        "linear_attention",
+        "full_attention",
+        "linear_attention",
+        "linear_attention",
+        "linear_attention",
+        "full_attention"
+    ],
+    "linear_conv_kernel_dim": 4,
+    "linear_key_head_dim": 128,
+    "linear_num_key_heads": 16,
+    "linear_num_value_heads": 16,
+    "linear_value_head_dim": 128,
+    "mamba_ssm_dtype": "float32",
+    "max_position_embeddings": 262144,
+    "mlp_only_layers": [],
+    "model_type": "qwen3_5",
+    "mtp_num_hidden_layers": 1,
+    "mtp_use_dedicated_embeddings": false,
+    "num_attention_heads": 8,
+    "num_hidden_layers": 24,
+    "num_key_value_heads": 2,
+    "pad_token_id": null,
+    "partial_rotary_factor": 0.25,
+    "rms_norm_eps": 1e-06,
+    "rope_parameters": {
+        "mrope_interleaved": true,
+        "mrope_section": [
+            11,
+            11,
+            10
+        ],
+        "partial_rotary_factor": 0.25,
+        "rope_theta": 10000000,
+        "type": "default"
+    },
+    "tie_word_embeddings": true,
+    "transformers_version": "5.17.0",
+    "use_cache": false,
+    "vocab_size": 248320
+}

bf16/generation_config.json

@@ -0,0 +1,6 @@
+{
+  "_from_model_config": true,
+  "eos_token_id": 248044,
+  "transformers_version": "5.17.0",
+  "use_cache": true
+}

bf16/macjev_norms_fp32.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a30e1a72d02598b405978c0eff08c3bfca3b523fa0b3d4b4c80d2a185e0b8081
+size 420112

bf16/model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:00629c8d17ea115c21c3fc0d6ddc5d4aa02887343a8ff3b3cf79ab197624d574
+size 3763691755

bf16/model.safetensors.index.json

@@ -0,0 +1,328 @@
+{
+    "metadata": {
+        "total_size": 3763650176,
+        "total_parameters": 1881824512
+    },
+    "weight_map": {
+        "language_model.model.embed_tokens.weight": "model.safetensors",
+        "language_model.model.layers.0.input_layernorm.weight": "model.safetensors",
+        "language_model.model.layers.0.linear_attn.A_log": "model.safetensors",
+        "language_model.model.layers.0.linear_attn.conv1d.weight": "model.safetensors",
+        "language_model.model.layers.0.linear_attn.dt_bias": "model.safetensors",
+        "language_model.model.layers.0.linear_attn.in_proj_a.weight": "model.safetensors",
+        "language_model.model.layers.0.linear_attn.in_proj_b.weight": "model.safetensors",
+        "language_model.model.layers.0.linear_attn.in_proj_qkv.weight": "model.safetensors",
+        "language_model.model.layers.0.linear_attn.in_proj_z.weight": "model.safetensors",
+        "language_model.model.layers.0.linear_attn.norm.weight": "model.safetensors",
+        "language_model.model.layers.0.linear_attn.out_proj.weight": "model.safetensors",
+        "language_model.model.layers.0.mlp.down_proj.weight": "model.safetensors",
+        "language_model.model.layers.0.mlp.gate_proj.weight": "model.safetensors",
+        "language_model.model.layers.0.mlp.up_proj.weight": "model.safetensors",
+        "language_model.model.layers.0.post_attention_layernorm.weight": "model.safetensors",
+        "language_model.model.layers.1.input_layernorm.weight": "model.safetensors",
+        "language_model.model.layers.1.linear_attn.A_log": "model.safetensors",
+        "language_model.model.layers.1.linear_attn.conv1d.weight": "model.safetensors",
+        "language_model.model.layers.1.linear_attn.dt_bias": "model.safetensors",
+        "language_model.model.layers.1.linear_attn.in_proj_a.weight": "model.safetensors",
+        "language_model.model.layers.1.linear_attn.in_proj_b.weight": "model.safetensors",
+        "language_model.model.layers.1.linear_attn.in_proj_qkv.weight": "model.safetensors",
+        "language_model.model.layers.1.linear_attn.in_proj_z.weight": "model.safetensors",
+        "language_model.model.layers.1.linear_attn.norm.weight": "model.safetensors",
+        "language_model.model.layers.1.linear_attn.out_proj.weight": "model.safetensors",
+        "language_model.model.layers.1.mlp.down_proj.weight": "model.safetensors",
+        "language_model.model.layers.1.mlp.gate_proj.weight": "model.safetensors",
+        "language_model.model.layers.1.mlp.up_proj.weight": "model.safetensors",
+        "language_model.model.layers.1.post_attention_layernorm.weight": "model.safetensors",
+        "language_model.model.layers.10.input_layernorm.weight": "model.safetensors",
+        "language_model.model.layers.10.linear_attn.A_log": "model.safetensors",
+        "language_model.model.layers.10.linear_attn.conv1d.weight": "model.safetensors",
+        "language_model.model.layers.10.linear_attn.dt_bias": "model.safetensors",
+        "language_model.model.layers.10.linear_attn.in_proj_a.weight": "model.safetensors",
+        "language_model.model.layers.10.linear_attn.in_proj_b.weight": "model.safetensors",
+        "language_model.model.layers.10.linear_attn.in_proj_qkv.weight": "model.safetensors",
+        "language_model.model.layers.10.linear_attn.in_proj_z.weight": "model.safetensors",
+        "language_model.model.layers.10.linear_attn.norm.weight": "model.safetensors",
+        "language_model.model.layers.10.linear_attn.out_proj.weight": "model.safetensors",
+        "language_model.model.layers.10.mlp.down_proj.weight": "model.safetensors",
+        "language_model.model.layers.10.mlp.gate_proj.weight": "model.safetensors",
+        "language_model.model.layers.10.mlp.up_proj.weight": "model.safetensors",
+        "language_model.model.layers.10.post_attention_layernorm.weight": "model.safetensors",
+        "language_model.model.layers.11.input_layernorm.weight": "model.safetensors",
+        "language_model.model.layers.11.mlp.down_proj.weight": "model.safetensors",
+        "language_model.model.layers.11.mlp.gate_proj.weight": "model.safetensors",
+        "language_model.model.layers.11.mlp.up_proj.weight": "model.safetensors",
+        "language_model.model.layers.11.post_attention_layernorm.weight": "model.safetensors",
+        "language_model.model.layers.11.self_attn.k_norm.weight": "model.safetensors",
+        "language_model.model.layers.11.self_attn.k_proj.weight": "model.safetensors",
+        "language_model.model.layers.11.self_attn.o_proj.weight": "model.safetensors",
+        "language_model.model.layers.11.self_attn.q_norm.weight": "model.safetensors",
+        "language_model.model.layers.11.self_attn.q_proj.weight": "model.safetensors",
+        "language_model.model.layers.11.self_attn.v_proj.weight": "model.safetensors",
+        "language_model.model.layers.12.input_layernorm.weight": "model.safetensors",
+        "language_model.model.layers.12.linear_attn.A_log": "model.safetensors",
+        "language_model.model.layers.12.linear_attn.conv1d.weight": "model.safetensors",
+        "language_model.model.layers.12.linear_attn.dt_bias": "model.safetensors",
+        "language_model.model.layers.12.linear_attn.in_proj_a.weight": "model.safetensors",
+        "language_model.model.layers.12.linear_attn.in_proj_b.weight": "model.safetensors",
+        "language_model.model.layers.12.linear_attn.in_proj_qkv.weight": "model.safetensors",
+        "language_model.model.layers.12.linear_attn.in_proj_z.weight": "model.safetensors",
+        "language_model.model.layers.12.linear_attn.norm.weight": "model.safetensors",
+        "language_model.model.layers.12.linear_attn.out_proj.weight": "model.safetensors",
+        "language_model.model.layers.12.mlp.down_proj.weight": "model.safetensors",
+        "language_model.model.layers.12.mlp.gate_proj.weight": "model.safetensors",
+        "language_model.model.layers.12.mlp.up_proj.weight": "model.safetensors",
+        "language_model.model.layers.12.post_attention_layernorm.weight": "model.safetensors",
+        "language_model.model.layers.13.input_layernorm.weight": "model.safetensors",
+        "language_model.model.layers.13.linear_attn.A_log": "model.safetensors",
+        "language_model.model.layers.13.linear_attn.conv1d.weight": "model.safetensors",
+        "language_model.model.layers.13.linear_attn.dt_bias": "model.safetensors",
+        "language_model.model.layers.13.linear_attn.in_proj_a.weight": "model.safetensors",
+        "language_model.model.layers.13.linear_attn.in_proj_b.weight": "model.safetensors",
+        "language_model.model.layers.13.linear_attn.in_proj_qkv.weight": "model.safetensors",
+        "language_model.model.layers.13.linear_attn.in_proj_z.weight": "model.safetensors",
+        "language_model.model.layers.13.linear_attn.norm.weight": "model.safetensors",
+        "language_model.model.layers.13.linear_attn.out_proj.weight": "model.safetensors",
+        "language_model.model.layers.13.mlp.down_proj.weight": "model.safetensors",
+        "language_model.model.layers.13.mlp.gate_proj.weight": "model.safetensors",
+        "language_model.model.layers.13.mlp.up_proj.weight": "model.safetensors",
+        "language_model.model.layers.13.post_attention_layernorm.weight": "model.safetensors",
+        "language_model.model.layers.14.input_layernorm.weight": "model.safetensors",
+        "language_model.model.layers.14.linear_attn.A_log": "model.safetensors",
+        "language_model.model.layers.14.linear_attn.conv1d.weight": "model.safetensors",
+        "language_model.model.layers.14.linear_attn.dt_bias": "model.safetensors",
+        "language_model.model.layers.14.linear_attn.in_proj_a.weight": "model.safetensors",
+        "language_model.model.layers.14.linear_attn.in_proj_b.weight": "model.safetensors",
+        "language_model.model.layers.14.linear_attn.in_proj_qkv.weight": "model.safetensors",
+        "language_model.model.layers.14.linear_attn.in_proj_z.weight": "model.safetensors",
+        "language_model.model.layers.14.linear_attn.norm.weight": "model.safetensors",
+        "language_model.model.layers.14.linear_attn.out_proj.weight": "model.safetensors",
+        "language_model.model.layers.14.mlp.down_proj.weight": "model.safetensors",
+        "language_model.model.layers.14.mlp.gate_proj.weight": "model.safetensors",
+        "language_model.model.layers.14.mlp.up_proj.weight": "model.safetensors",
+        "language_model.model.layers.14.post_attention_layernorm.weight": "model.safetensors",
+        "language_model.model.layers.15.input_layernorm.weight": "model.safetensors",
+        "language_model.model.layers.15.mlp.down_proj.weight": "model.safetensors",
+        "language_model.model.layers.15.mlp.gate_proj.weight": "model.safetensors",
+        "language_model.model.layers.15.mlp.up_proj.weight": "model.safetensors",
+        "language_model.model.layers.15.post_attention_layernorm.weight": "model.safetensors",
+        "language_model.model.layers.15.self_attn.k_norm.weight": "model.safetensors",
+        "language_model.model.layers.15.self_attn.k_proj.weight": "model.safetensors",
+        "language_model.model.layers.15.self_attn.o_proj.weight": "model.safetensors",
+        "language_model.model.layers.15.self_attn.q_norm.weight": "model.safetensors",
+        "language_model.model.layers.15.self_attn.q_proj.weight": "model.safetensors",
+        "language_model.model.layers.15.self_attn.v_proj.weight": "model.safetensors",
+        "language_model.model.layers.16.input_layernorm.weight": "model.safetensors",
+        "language_model.model.layers.16.linear_attn.A_log": "model.safetensors",
+        "language_model.model.layers.16.linear_attn.conv1d.weight": "model.safetensors",
+        "language_model.model.layers.16.linear_attn.dt_bias": "model.safetensors",
+        "language_model.model.layers.16.linear_attn.in_proj_a.weight": "model.safetensors",
+        "language_model.model.layers.16.linear_attn.in_proj_b.weight": "model.safetensors",
+        "language_model.model.layers.16.linear_attn.in_proj_qkv.weight": "model.safetensors",
+        "language_model.model.layers.16.linear_attn.in_proj_z.weight": "model.safetensors",
+        "language_model.model.layers.16.linear_attn.norm.weight": "model.safetensors",
+        "language_model.model.layers.16.linear_attn.out_proj.weight": "model.safetensors",
+        "language_model.model.layers.16.mlp.down_proj.weight": "model.safetensors",
+        "language_model.model.layers.16.mlp.gate_proj.weight": "model.safetensors",
+        "language_model.model.layers.16.mlp.up_proj.weight": "model.safetensors",
+        "language_model.model.layers.16.post_attention_layernorm.weight": "model.safetensors",
+        "language_model.model.layers.17.input_layernorm.weight": "model.safetensors",
+        "language_model.model.layers.17.linear_attn.A_log": "model.safetensors",
+        "language_model.model.layers.17.linear_attn.conv1d.weight": "model.safetensors",
+        "language_model.model.layers.17.linear_attn.dt_bias": "model.safetensors",
+        "language_model.model.layers.17.linear_attn.in_proj_a.weight": "model.safetensors",
+        "language_model.model.layers.17.linear_attn.in_proj_b.weight": "model.safetensors",
+        "language_model.model.layers.17.linear_attn.in_proj_qkv.weight": "model.safetensors",
+        "language_model.model.layers.17.linear_attn.in_proj_z.weight": "model.safetensors",
+        "language_model.model.layers.17.linear_attn.norm.weight": "model.safetensors",
+        "language_model.model.layers.17.linear_attn.out_proj.weight": "model.safetensors",
+        "language_model.model.layers.17.mlp.down_proj.weight": "model.safetensors",
+        "language_model.model.layers.17.mlp.gate_proj.weight": "model.safetensors",
+        "language_model.model.layers.17.mlp.up_proj.weight": "model.safetensors",
+        "language_model.model.layers.17.post_attention_layernorm.weight": "model.safetensors",
+        "language_model.model.layers.18.input_layernorm.weight": "model.safetensors",
+        "language_model.model.layers.18.linear_attn.A_log": "model.safetensors",
+        "language_model.model.layers.18.linear_attn.conv1d.weight": "model.safetensors",
+        "language_model.model.layers.18.linear_attn.dt_bias": "model.safetensors",
+        "language_model.model.layers.18.linear_attn.in_proj_a.weight": "model.safetensors",
+        "language_model.model.layers.18.linear_attn.in_proj_b.weight": "model.safetensors",
+        "language_model.model.layers.18.linear_attn.in_proj_qkv.weight": "model.safetensors",
+        "language_model.model.layers.18.linear_attn.in_proj_z.weight": "model.safetensors",
+        "language_model.model.layers.18.linear_attn.norm.weight": "model.safetensors",
+        "language_model.model.layers.18.linear_attn.out_proj.weight": "model.safetensors",
+        "language_model.model.layers.18.mlp.down_proj.weight": "model.safetensors",
+        "language_model.model.layers.18.mlp.gate_proj.weight": "model.safetensors",
+        "language_model.model.layers.18.mlp.up_proj.weight": "model.safetensors",
+        "language_model.model.layers.18.post_attention_layernorm.weight": "model.safetensors",
+        "language_model.model.layers.19.input_layernorm.weight": "model.safetensors",
+        "language_model.model.layers.19.mlp.down_proj.weight": "model.safetensors",
+        "language_model.model.layers.19.mlp.gate_proj.weight": "model.safetensors",
+        "language_model.model.layers.19.mlp.up_proj.weight": "model.safetensors",
+        "language_model.model.layers.19.post_attention_layernorm.weight": "model.safetensors",
+        "language_model.model.layers.19.self_attn.k_norm.weight": "model.safetensors",
+        "language_model.model.layers.19.self_attn.k_proj.weight": "model.safetensors",
+        "language_model.model.layers.19.self_attn.o_proj.weight": "model.safetensors",
+        "language_model.model.layers.19.self_attn.q_norm.weight": "model.safetensors",
+        "language_model.model.layers.19.self_attn.q_proj.weight": "model.safetensors",
+        "language_model.model.layers.19.self_attn.v_proj.weight": "model.safetensors",
+        "language_model.model.layers.2.input_layernorm.weight": "model.safetensors",
+        "language_model.model.layers.2.linear_attn.A_log": "model.safetensors",
+        "language_model.model.layers.2.linear_attn.conv1d.weight": "model.safetensors",
+        "language_model.model.layers.2.linear_attn.dt_bias": "model.safetensors",
+        "language_model.model.layers.2.linear_attn.in_proj_a.weight": "model.safetensors",
+        "language_model.model.layers.2.linear_attn.in_proj_b.weight": "model.safetensors",
+        "language_model.model.layers.2.linear_attn.in_proj_qkv.weight": "model.safetensors",
+        "language_model.model.layers.2.linear_attn.in_proj_z.weight": "model.safetensors",
+        "language_model.model.layers.2.linear_attn.norm.weight": "model.safetensors",
+        "language_model.model.layers.2.linear_attn.out_proj.weight": "model.safetensors",
+        "language_model.model.layers.2.mlp.down_proj.weight": "model.safetensors",
+        "language_model.model.layers.2.mlp.gate_proj.weight": "model.safetensors",
+        "language_model.model.layers.2.mlp.up_proj.weight": "model.safetensors",
+        "language_model.model.layers.2.post_attention_layernorm.weight": "model.safetensors",
+        "language_model.model.layers.20.input_layernorm.weight": "model.safetensors",
+        "language_model.model.layers.20.linear_attn.A_log": "model.safetensors",
+        "language_model.model.layers.20.linear_attn.conv1d.weight": "model.safetensors",
+        "language_model.model.layers.20.linear_attn.dt_bias": "model.safetensors",
+        "language_model.model.layers.20.linear_attn.in_proj_a.weight": "model.safetensors",
+        "language_model.model.layers.20.linear_attn.in_proj_b.weight": "model.safetensors",
+        "language_model.model.layers.20.linear_attn.in_proj_qkv.weight": "model.safetensors",
+        "language_model.model.layers.20.linear_attn.in_proj_z.weight": "model.safetensors",
+        "language_model.model.layers.20.linear_attn.norm.weight": "model.safetensors",
+        "language_model.model.layers.20.linear_attn.out_proj.weight": "model.safetensors",
+        "language_model.model.layers.20.mlp.down_proj.weight": "model.safetensors",
+        "language_model.model.layers.20.mlp.gate_proj.weight": "model.safetensors",
+        "language_model.model.layers.20.mlp.up_proj.weight": "model.safetensors",
+        "language_model.model.layers.20.post_attention_layernorm.weight": "model.safetensors",
+        "language_model.model.layers.21.input_layernorm.weight": "model.safetensors",
+        "language_model.model.layers.21.linear_attn.A_log": "model.safetensors",
+        "language_model.model.layers.21.linear_attn.conv1d.weight": "model.safetensors",
+        "language_model.model.layers.21.linear_attn.dt_bias": "model.safetensors",
+        "language_model.model.layers.21.linear_attn.in_proj_a.weight": "model.safetensors",
+        "language_model.model.layers.21.linear_attn.in_proj_b.weight": "model.safetensors",
+        "language_model.model.layers.21.linear_attn.in_proj_qkv.weight": "model.safetensors",
+        "language_model.model.layers.21.linear_attn.in_proj_z.weight": "model.safetensors",
+        "language_model.model.layers.21.linear_attn.norm.weight": "model.safetensors",
+        "language_model.model.layers.21.linear_attn.out_proj.weight": "model.safetensors",
+        "language_model.model.layers.21.mlp.down_proj.weight": "model.safetensors",
+        "language_model.model.layers.21.mlp.gate_proj.weight": "model.safetensors",
+        "language_model.model.layers.21.mlp.up_proj.weight": "model.safetensors",
+        "language_model.model.layers.21.post_attention_layernorm.weight": "model.safetensors",
+        "language_model.model.layers.22.input_layernorm.weight": "model.safetensors",
+        "language_model.model.layers.22.linear_attn.A_log": "model.safetensors",
+        "language_model.model.layers.22.linear_attn.conv1d.weight": "model.safetensors",
+        "language_model.model.layers.22.linear_attn.dt_bias": "model.safetensors",
+        "language_model.model.layers.22.linear_attn.in_proj_a.weight": "model.safetensors",
+        "language_model.model.layers.22.linear_attn.in_proj_b.weight": "model.safetensors",
+        "language_model.model.layers.22.linear_attn.in_proj_qkv.weight": "model.safetensors",
+        "language_model.model.layers.22.linear_attn.in_proj_z.weight": "model.safetensors",
+        "language_model.model.layers.22.linear_attn.norm.weight": "model.safetensors",
+        "language_model.model.layers.22.linear_attn.out_proj.weight": "model.safetensors",
+        "language_model.model.layers.22.mlp.down_proj.weight": "model.safetensors",
+        "language_model.model.layers.22.mlp.gate_proj.weight": "model.safetensors",
+        "language_model.model.layers.22.mlp.up_proj.weight": "model.safetensors",
+        "language_model.model.layers.22.post_attention_layernorm.weight": "model.safetensors",
+        "language_model.model.layers.23.input_layernorm.weight": "model.safetensors",
+        "language_model.model.layers.23.mlp.down_proj.weight": "model.safetensors",
+        "language_model.model.layers.23.mlp.gate_proj.weight": "model.safetensors",
+        "language_model.model.layers.23.mlp.up_proj.weight": "model.safetensors",
+        "language_model.model.layers.23.post_attention_layernorm.weight": "model.safetensors",
+        "language_model.model.layers.23.self_attn.k_norm.weight": "model.safetensors",
+        "language_model.model.layers.23.self_attn.k_proj.weight": "model.safetensors",
+        "language_model.model.layers.23.self_attn.o_proj.weight": "model.safetensors",
+        "language_model.model.layers.23.self_attn.q_norm.weight": "model.safetensors",
+        "language_model.model.layers.23.self_attn.q_proj.weight": "model.safetensors",
+        "language_model.model.layers.23.self_attn.v_proj.weight": "model.safetensors",
+        "language_model.model.layers.3.input_layernorm.weight": "model.safetensors",
+        "language_model.model.layers.3.mlp.down_proj.weight": "model.safetensors",
+        "language_model.model.layers.3.mlp.gate_proj.weight": "model.safetensors",
+        "language_model.model.layers.3.mlp.up_proj.weight": "model.safetensors",
+        "language_model.model.layers.3.post_attention_layernorm.weight": "model.safetensors",
+        "language_model.model.layers.3.self_attn.k_norm.weight": "model.safetensors",
+        "language_model.model.layers.3.self_attn.k_proj.weight": "model.safetensors",
+        "language_model.model.layers.3.self_attn.o_proj.weight": "model.safetensors",
+        "language_model.model.layers.3.self_attn.q_norm.weight": "model.safetensors",
+        "language_model.model.layers.3.self_attn.q_proj.weight": "model.safetensors",
+        "language_model.model.layers.3.self_attn.v_proj.weight": "model.safetensors",
+        "language_model.model.layers.4.input_layernorm.weight": "model.safetensors",
+        "language_model.model.layers.4.linear_attn.A_log": "model.safetensors",
+        "language_model.model.layers.4.linear_attn.conv1d.weight": "model.safetensors",
+        "language_model.model.layers.4.linear_attn.dt_bias": "model.safetensors",
+        "language_model.model.layers.4.linear_attn.in_proj_a.weight": "model.safetensors",
+        "language_model.model.layers.4.linear_attn.in_proj_b.weight": "model.safetensors",
+        "language_model.model.layers.4.linear_attn.in_proj_qkv.weight": "model.safetensors",
+        "language_model.model.layers.4.linear_attn.in_proj_z.weight": "model.safetensors",
+        "language_model.model.layers.4.linear_attn.norm.weight": "model.safetensors",
+        "language_model.model.layers.4.linear_attn.out_proj.weight": "model.safetensors",
+        "language_model.model.layers.4.mlp.down_proj.weight": "model.safetensors",
+        "language_model.model.layers.4.mlp.gate_proj.weight": "model.safetensors",
+        "language_model.model.layers.4.mlp.up_proj.weight": "model.safetensors",
+        "language_model.model.layers.4.post_attention_layernorm.weight": "model.safetensors",
+        "language_model.model.layers.5.input_layernorm.weight": "model.safetensors",
+        "language_model.model.layers.5.linear_attn.A_log": "model.safetensors",
+        "language_model.model.layers.5.linear_attn.conv1d.weight": "model.safetensors",
+        "language_model.model.layers.5.linear_attn.dt_bias": "model.safetensors",
+        "language_model.model.layers.5.linear_attn.in_proj_a.weight": "model.safetensors",
+        "language_model.model.layers.5.linear_attn.in_proj_b.weight": "model.safetensors",
+        "language_model.model.layers.5.linear_attn.in_proj_qkv.weight": "model.safetensors",
+        "language_model.model.layers.5.linear_attn.in_proj_z.weight": "model.safetensors",
+        "language_model.model.layers.5.linear_attn.norm.weight": "model.safetensors",
+        "language_model.model.layers.5.linear_attn.out_proj.weight": "model.safetensors",
+        "language_model.model.layers.5.mlp.down_proj.weight": "model.safetensors",
+        "language_model.model.layers.5.mlp.gate_proj.weight": "model.safetensors",
+        "language_model.model.layers.5.mlp.up_proj.weight": "model.safetensors",
+        "language_model.model.layers.5.post_attention_layernorm.weight": "model.safetensors",
+        "language_model.model.layers.6.input_layernorm.weight": "model.safetensors",
+        "language_model.model.layers.6.linear_attn.A_log": "model.safetensors",
+        "language_model.model.layers.6.linear_attn.conv1d.weight": "model.safetensors",
+        "language_model.model.layers.6.linear_attn.dt_bias": "model.safetensors",
+        "language_model.model.layers.6.linear_attn.in_proj_a.weight": "model.safetensors",
+        "language_model.model.layers.6.linear_attn.in_proj_b.weight": "model.safetensors",
+        "language_model.model.layers.6.linear_attn.in_proj_qkv.weight": "model.safetensors",
+        "language_model.model.layers.6.linear_attn.in_proj_z.weight": "model.safetensors",
+        "language_model.model.layers.6.linear_attn.norm.weight": "model.safetensors",
+        "language_model.model.layers.6.linear_attn.out_proj.weight": "model.safetensors",
+        "language_model.model.layers.6.mlp.down_proj.weight": "model.safetensors",
+        "language_model.model.layers.6.mlp.gate_proj.weight": "model.safetensors",
+        "language_model.model.layers.6.mlp.up_proj.weight": "model.safetensors",
+        "language_model.model.layers.6.post_attention_layernorm.weight": "model.safetensors",
+        "language_model.model.layers.7.input_layernorm.weight": "model.safetensors",
+        "language_model.model.layers.7.mlp.down_proj.weight": "model.safetensors",
+        "language_model.model.layers.7.mlp.gate_proj.weight": "model.safetensors",
+        "language_model.model.layers.7.mlp.up_proj.weight": "model.safetensors",
+        "language_model.model.layers.7.post_attention_layernorm.weight": "model.safetensors",
+        "language_model.model.layers.7.self_attn.k_norm.weight": "model.safetensors",
+        "language_model.model.layers.7.self_attn.k_proj.weight": "model.safetensors",
+        "language_model.model.layers.7.self_attn.o_proj.weight": "model.safetensors",
+        "language_model.model.layers.7.self_attn.q_norm.weight": "model.safetensors",
+        "language_model.model.layers.7.self_attn.q_proj.weight": "model.safetensors",
+        "language_model.model.layers.7.self_attn.v_proj.weight": "model.safetensors",
+        "language_model.model.layers.8.input_layernorm.weight": "model.safetensors",
+        "language_model.model.layers.8.linear_attn.A_log": "model.safetensors",
+        "language_model.model.layers.8.linear_attn.conv1d.weight": "model.safetensors",
+        "language_model.model.layers.8.linear_attn.dt_bias": "model.safetensors",
+        "language_model.model.layers.8.linear_attn.in_proj_a.weight": "model.safetensors",
+        "language_model.model.layers.8.linear_attn.in_proj_b.weight": "model.safetensors",
+        "language_model.model.layers.8.linear_attn.in_proj_qkv.weight": "model.safetensors",
+        "language_model.model.layers.8.linear_attn.in_proj_z.weight": "model.safetensors",
+        "language_model.model.layers.8.linear_attn.norm.weight": "model.safetensors",
+        "language_model.model.layers.8.linear_attn.out_proj.weight": "model.safetensors",
+        "language_model.model.layers.8.mlp.down_proj.weight": "model.safetensors",
+        "language_model.model.layers.8.mlp.gate_proj.weight": "model.safetensors",
+        "language_model.model.layers.8.mlp.up_proj.weight": "model.safetensors",
+        "language_model.model.layers.8.post_attention_layernorm.weight": "model.safetensors",
+        "language_model.model.layers.9.input_layernorm.weight": "model.safetensors",
+        "language_model.model.layers.9.linear_attn.A_log": "model.safetensors",
+        "language_model.model.layers.9.linear_attn.conv1d.weight": "model.safetensors",
+        "language_model.model.layers.9.linear_attn.dt_bias": "model.safetensors",
+        "language_model.model.layers.9.linear_attn.in_proj_a.weight": "model.safetensors",
+        "language_model.model.layers.9.linear_attn.in_proj_b.weight": "model.safetensors",
+        "language_model.model.layers.9.linear_attn.in_proj_qkv.weight": "model.safetensors",
+        "language_model.model.layers.9.linear_attn.in_proj_z.weight": "model.safetensors",
+        "language_model.model.layers.9.linear_attn.norm.weight": "model.safetensors",
+        "language_model.model.layers.9.linear_attn.out_proj.weight": "model.safetensors",
+        "language_model.model.layers.9.mlp.down_proj.weight": "model.safetensors",
+        "language_model.model.layers.9.mlp.gate_proj.weight": "model.safetensors",
+        "language_model.model.layers.9.mlp.up_proj.weight": "model.safetensors",
+        "language_model.model.layers.9.post_attention_layernorm.weight": "model.safetensors",
+        "language_model.model.norm.weight": "model.safetensors"
+    }
+}

bf16/tokenizer.json

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:06b9509352d2af50381ab2247e083b80d32d5c0aba91c272ca9ff729b6a0e523
+size 19989325

bf16/tokenizer_config.json

@@ -0,0 +1,33 @@
+{
+  "add_prefix_space": false,
+  "audio_bos_token": "<|audio_start|>",
+  "audio_eos_token": "<|audio_end|>",
+  "audio_token": "<|audio_pad|>",
+  "backend": "tokenizers",
+  "bos_token": null,
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "<|im_end|>",
+  "errors": "replace",
+  "image_token": "<|image_pad|>",
+  "is_local": true,
+  "local_files_only": false,
+  "model_max_length": 262144,
+  "model_specific_special_tokens": {
+    "audio_bos_token": "<|audio_start|>",
+    "audio_eos_token": "<|audio_end|>",
+    "audio_token": "<|audio_pad|>",
+    "image_token": "<|image_pad|>",
+    "video_token": "<|video_pad|>",
+    "vision_bos_token": "<|vision_start|>",
+    "vision_eos_token": "<|vision_end|>"
+  },
+  "pad_token": "<|endoftext|>",
+  "pretokenize_regex": "(?i:'s|'t|'re|'ve|'m|'ll|'d)|[^\\r\\n\\p{L}\\p{N}]?[\\p{L}\\p{M}]+|\\p{N}| ?[^\\s\\p{L}\\p{M}\\p{N}]+[\\r\\n]*|\\s*[\\r\\n]+|\\s+(?!\\S)|\\s+",
+  "split_special_tokens": false,
+  "tokenizer_class": "Qwen2Tokenizer",
+  "tool_parser_type": "qwen3_coder",
+  "unk_token": null,
+  "video_token": "<|video_pad|>",
+  "vision_bos_token": "<|vision_start|>",
+  "vision_eos_token": "<|vision_end|>"
+}

config.json

@@ -0,0 +1,75 @@
+{
+    "architectures": [
+        "Qwen3_5ForCausalLM"
+    ],
+    "attention_bias": false,
+    "attention_dropout": 0.0,
+    "attn_output_gate": true,
+    "bos_token_id": null,
+    "dtype": "bfloat16",
+    "eos_token_id": 248044,
+    "full_attention_interval": 4,
+    "head_dim": 256,
+    "hidden_act": "silu",
+    "hidden_size": 2048,
+    "initializer_range": 0.02,
+    "intermediate_size": 6144,
+    "layer_types": [
+        "linear_attention",
+        "linear_attention",
+        "linear_attention",
+        "full_attention",
+        "linear_attention",
+        "linear_attention",
+        "linear_attention",
+        "full_attention",
+        "linear_attention",
+        "linear_attention",
+        "linear_attention",
+        "full_attention",
+        "linear_attention",
+        "linear_attention",
+        "linear_attention",
+        "full_attention",
+        "linear_attention",
+        "linear_attention",
+        "linear_attention",
+        "full_attention",
+        "linear_attention",
+        "linear_attention",
+        "linear_attention",
+        "full_attention"
+    ],
+    "linear_conv_kernel_dim": 4,
+    "linear_key_head_dim": 128,
+    "linear_num_key_heads": 16,
+    "linear_num_value_heads": 16,
+    "linear_value_head_dim": 128,
+    "mamba_ssm_dtype": "float32",
+    "max_position_embeddings": 262144,
+    "mlp_only_layers": [],
+    "model_type": "qwen3_5",
+    "mtp_num_hidden_layers": 1,
+    "mtp_use_dedicated_embeddings": false,
+    "num_attention_heads": 8,
+    "num_hidden_layers": 24,
+    "num_key_value_heads": 2,
+    "pad_token_id": null,
+    "partial_rotary_factor": 0.25,
+    "rms_norm_eps": 1e-06,
+    "rope_parameters": {
+        "mrope_interleaved": true,
+        "mrope_section": [
+            11,
+            11,
+            10
+        ],
+        "partial_rotary_factor": 0.25,
+        "rope_theta": 10000000,
+        "type": "default"
+    },
+    "tie_word_embeddings": true,
+    "transformers_version": "5.17.0",
+    "use_cache": false,
+    "vocab_size": 248320
+}

figures/banner.data.json

@@ -0,0 +1,44 @@
+{
+ "left": {
+  "big": "73.6% on JevBench",
+  "sub": "+9.5 points over our 0.8B v3; Jev 86.6%",
+  "tokens": "25,600 tokens per call, no option cap"
+ },
+ "rows": [
+  {
+   "label": "Jev 1.13 (API)",
+   "accuracy": 0.8658008658008658,
+   "pct_1dp": 86.6
+  },
+  {
+   "label": "2B v3 \u00b7 this model",
+   "accuracy": 0.7359307359307359,
+   "pct_1dp": 73.6
+  },
+  {
+   "label": "decider-2b",
+   "accuracy": 0.70995670995671,
+   "pct_1dp": 71.0
+  },
+  {
+   "label": "Open-Jev 2B",
+   "accuracy": 0.645021645021645,
+   "pct_1dp": 64.5
+  },
+  {
+   "label": "0.8B v3",
+   "accuracy": 0.6406926406926406,
+   "pct_1dp": 64.1
+  },
+  {
+   "label": "Laya",
+   "accuracy": 0.5844155844155844,
+   "pct_1dp": 58.4
+  }
+ ],
+ "shown": "Shown: Qwen3.5-2B-family systems, our 0.8B v3, Laya and Jev \u00b7 42 of 82 board systems score higher than 73.6%",
+ "note": "231 public items. 2B v3: self-run with the official harness (GGUF F16), not an official board entry; 95% CI 67.6\u201378.9%, so the lead over decider-2b is inside the CI. Other rows as published on the v1.4.1 board.",
+ "sources": [
+  "figures/jevbench.data.json"
+ ]
+}

figures/jevbench.data.json

@@ -0,0 +1,49 @@
+{
+ "chart": "jevbench",
+ "metric": "JevBench v1.4.1 public accuracy (231 items)",
+ "rows": [
+  {
+   "label": "Jev-Style 2B v3 (this model)",
+   "accuracy": 0.7359307359307359,
+   "accuracy_pct_1dp": 73.6,
+   "source": "https://huggingface.co/chaoliangUNSW/Jev-Style-2B-Decision-v3/blob/main/validation/benchmarks/jevbench_v1.4.1_results.json :: public_accuracy (170/231)"
+  },
+  {
+   "label": "decider-2b (Mapika, Qwen3.5-2B-Base)",
+   "accuracy": 0.70995670995671,
+   "accuracy_pct_1dp": 71.0,
+   "source": "https://github.com/fstandhartinger/jevbench/blob/v1.4.1/results/v1.4.1/jevbench-v1.4.1-results.json :: systems[decider-2b]"
+  },
+  {
+   "label": "Open-Jev 2B (Zefan Cai, Qwen3.5-2B + LoRA)",
+   "accuracy": 0.645021645021645,
+   "accuracy_pct_1dp": 64.5,
+   "source": "https://github.com/fstandhartinger/jevbench/blob/v1.4.1/results/v1.4.1/jevbench-v1.4.1-results.json :: systems[open-jev-zefan-2b]"
+  },
+  {
+   "label": "Jev-Style 0.8B v3",
+   "accuracy": 0.6406926406926406,
+   "accuracy_pct_1dp": 64.1,
+   "source": "https://huggingface.co/chaoliangUNSW/Jev-Style-0.8B-Decision-v3/blob/main/figures/jevbench.data.json (0.8B v3 card; row Jev-Style 0.8B v3, 148/231)"
+  },
+  {
+   "label": "Laya (ModernBERT-large, 421M)",
+   "accuracy": 0.5844155844155844,
+   "accuracy_pct_1dp": 58.4,
+   "source": "https://github.com/fstandhartinger/jevbench/blob/v1.4.1/results/v1.4.1/jevbench-v1.4.1-results.json :: systems[laya]"
+  }
+ ],
+ "reference_line": {
+  "label": "Jev 1.13.0",
+  "accuracy": 0.8658008658008658,
+  "source": "https://github.com/fstandhartinger/jevbench/blob/v1.4.1/results/v1.4.1/jevbench-v1.4.1-results.json :: systems[jev-1.13.0]"
+ },
+ "ci95_wilson_2b": [
+  0.6755578394149232,
+  0.7885850787366094
+ ],
+ "board_systems_higher_than_2b": 42,
+ "board_systems": 82,
+ "board_sha256": "e6754863056503fe2b010410fc7111df884ac1f9ce4449aa369aab61d98092cd",
+ "footnote": "JevBench v1.4.1, 231 public items. 2B v3: self-run once with the official harness (commit 24b9b5c), GGUF F16 engine, one global temperature, not an official board entry; 95% CI 67.6-78.9% (Wilson), so its lead over decider-2b (164/231) is inside the CI; training-pool contamination scan: 0 hits. Other rows: public accuracy as published in the board's v1.4.1 results file. Shown: the Qwen3.5-2B-family systems on the board, our 0.8B v3, Laya and Jev; 42 of the 82 board systems score higher than 73.6%."
+}

figures/zeroshot.data.json

@@ -0,0 +1,40 @@
+{
+ "chart": "zeroshot",
+ "metric": "accuracy over every row of the pinned test files",
+ "values": {
+  "tweet_topic": {
+   "2b": 0.822209096278795,
+   "2b_ci95": [
+    0.8038984051978736,
+    0.8399438865918486
+   ],
+   "2b_macro_f1": 0.677897069437572,
+   "2b_ece15": 0.027919883880813873,
+   "08b": 0.754873006497342,
+   "jev": 0.7932663910218547,
+   "jev_macro_f1": 0.6936,
+   "jev_ece15": 0.0631,
+   "n": 1693
+  },
+  "fin_topic": {
+   "2b": 0.6111246052951178,
+   "2b_ci95": [
+    0.5960650959436483,
+    0.6259412193344669
+   ],
+   "2b_macro_f1": 0.5897964463354406,
+   "2b_ece15": 0.06460657860002232,
+   "08b": 0.4670876852076755,
+   "jev": 0.669905270828273,
+   "jev_macro_f1": 0.6298,
+   "jev_ece15": 0.1664,
+   "n": 4117
+  }
+ },
+ "sources": {
+  "2b": "https://huggingface.co/chaoliangUNSW/Jev-Style-2B-Decision-v3/blob/main/validation/benchmarks/zeroshot_metrics.json",
+  "0.8b": "https://huggingface.co/chaoliangUNSW/Jev-Style-0.8B-Decision-v3/blob/main/figures/zeroshot.json (0.8B v3 card; v3_recomputed: tweet_topic 1278/1693, fin_topic 1923/4117)",
+  "jev": "https://github.com/elcronos/jev-vs-open-decision-models/blob/a1901bc3d520e73936de8d4326545c0cdcf742fb/results/cross_dataset_summary.json (as copied in https://huggingface.co/chaoliangUNSW/Jev-Style-2B-Decision-v3/blob/main/validation/benchmarks/zeroshot_metrics.json :: comparison)"
+ },
+ "footnote": "Zero-shot: none of these test sets is in the 2B or 0.8B training pool; accuracy over every row of the pinned test files (n = 1,693 and 4,117). 2B v3: GGUF F16 engine, one global temperature, run once; tweet_topic 95% CI 80.4-84.0%. Jev (1.13, API): numbers published by the elcronos jev-vs-open-decision-models study (cross_dataset_summary.json @ a1901bc), not re-run by us. Macro-F1 is below Jev on both sets (tweet_topic 67.8% vs 69.4%; fin_topic 59.0% vs 63.0%)."
+}

jev_style_decision_mlx.py

@@ -0,0 +1,1059 @@
+"""Jev-Style-2B-Decision-v3: typed decisions with MLX on Apple silicon (bf16 / 8-bit, --precision).
+
+Self-contained runtime for chaoliangUNSW/Jev-Style-2B-Decision-v3 (Apache-2.0). No dependency on any training
+code: rendering (protocol "macjev-render-v2-long-options"), block-causal attention, the verdict readout and
+calibration are implemented below and reproduce the reference implementation used for evaluation (see
+release_config.json -> "runtime_parity"). Needs mlx, mlx-lm==0.31.3 (pinned, see the MLX backend section),
+tokenizers and numpy.
+
+Calibration: probabilities use the ONE global temperature of readout_config.json (temperatures.global);
+temperature=... (CLI --temperature, JSONL "temperature") overrides it (1.0 = uncalibrated scores). Unlike the
+0.8B v3 runtime there is no --category (no group temperatures) and no --head-max (no question/options cap:
+only the 25,600-token total budget applies).
+
+Python:
+    from jev_style_decision_mlx import JevStyleDecisionMLX
+    m = JevStyleDecisionMLX(".", precision="bf16")          # or "8bit"
+    m.decide("<state>", "Which action?", options={"click": "press it", "wait": None})
+    m.score_many("<state>", [q1, q2, ...])                  # state computed once, reused for every question
+"""
+# ----------------------------------------------------------------------------------------------
+# Shared core (byte-identical in jev_style_decision.py, jev_style_decision_gguf.py and jev_style_decision_mlx.py):
+# input rendering (v2), verdict readout, calibration, budgets, errors, manifest check, CLI/JSONL.
+#
+# Input layout ("macjev-render-v2-long-options", layout "sb"; token segments are encoded separately
+# and concatenated, so slot positions are exact):
+#
+#   state prefix     State:\n<state>\n\n
+#   short form       Question [<type>]: <question>\nOptions:\n
+#   (question +      - <option 1> ->\n ... - <option K> ->\n
+#    options + slots <= 2,048 tokens)
+#   overflow form    Question [<type>]: <question>\nOptions:\n
+#   (otherwise)      Option 1: <option 1>\n ... Option K: <option K>\n            (the catalogue)
+#                    Judge each numbered option in the complete catalogue above:\n
+#                    Option 1 ->\n ... Option K ->\n                               (the rubric)
+#
+# Attention blocks ([start, stop) token ranges): the state is cut into consecutive 2,048-token
+# blocks; the short form is one block; in the overflow form the catalogue and the rubric are each cut
+# into 2,048-token blocks. In the 6 full-attention layers every block attends to all earlier tokens
+# and to itself with NO causal mask inside the block (block-causal); the Gated-DeltaNet layers are
+# ordinary recurrent layers. A backend must compute exactly these blocks (never merged, never re-split).
+#
+# Score of option k = logit(" yes") - logit(" no") at its " ->" slot = h_slot . (W_yes - W_no) in
+# float32 (final normed hidden state, tied embedding rows). Probabilities = softmax(scores / T) in
+# canonical option order. T = the ONE global temperature of readout_config.json
+# (temperatures.global, fitted on 2,000 calibration rows) unless temperature=... overrides it
+# (1.0 = uncalibrated scores). This model has no category / group temperatures and no separate
+# question/options budget, so the 0.8B v3 runtime's --category and --head-max options do not exist here.
+#
+# Budget: the complete input (state + question + options + readout) <= 25,600 tokens. Larger inputs
+# raise InputBudgetError; nothing is ever truncated. Text inside the state, question and options is
+# tokenised with special tokens disabled, so e.g. "<|im_end|>" in user text can never act as a
+# control token. A non-finite score (NaN / inf) raises NonFiniteScoreError: no probabilities are made.
+# ----------------------------------------------------------------------------------------------
+import argparse
+import hashlib
+import json
+import math
+import sys
+from pathlib import Path
+
+import numpy as np
+
+MODEL_NAME = "Jev-Style-2B-Decision-v3"
+TEMPLATE_VERSION = "macjev-render-v2-long-options"
+READOUT_FORMAT = "macjev-readout-v2"
+LAYOUT = "sb"
+BLOCK = 2048                    # attention block size (a processing unit, not a content limit)
+CONTEXT_LIMIT = 25_600          # state + question + options + readout, all included
+QTYPES = ("choice", "score", "noul")
+JSONL_GROUP_MAX = 256           # consecutive JSONL rows with one state scored in one backend call
+HERE = Path(__file__).resolve().parent
+
+
+class InputBudgetError(ValueError):
+    """The rendered input exceeds the token budget. Nothing was truncated."""
+
+
+class QuestionError(ValueError):
+    """The question/options are malformed."""
+
+
+class NonFiniteScoreError(FloatingPointError):
+    """The model produced a non-finite decision score (NaN or inf). No probabilities are returned."""
+
+
+# -- questions ------------------------------------------------------------------------------------
+def option_names(question):
+    """Canonical option identifiers, in the order the probabilities are returned."""
+    if not isinstance(question, dict):
+        raise QuestionError("question must be a dict {'t', 'ins', 'crit'}")
+    t, crit = question.get("t"), question.get("crit")
+    if not isinstance(question.get("ins"), str) or not question["ins"].strip():
+        raise QuestionError("question text ('ins') must be a non-empty string")
+    if t == "choice":
+        if not isinstance(crit, dict) or not crit:
+            raise QuestionError("choice needs a non-empty dict {option name: description or None}")
+        return [str(k) for k in crit]
+    if t == "score":
+        if not isinstance(crit, list) or not 2 <= len(crit) <= 10:
+            raise QuestionError("score needs a list of 2..10 level descriptions")
+        return [str(i) for i in range(len(crit))]
+    if t == "noul":
+        if crit is not None and not isinstance(crit, dict):
+            raise QuestionError("noul criteria must be None or {'false': ..., 'true': ...}")
+        return ["false", "true"]
+    raise QuestionError(f"unknown question type {t!r} (expected one of {QTYPES})")
+
+
+def make_question(question, options=None, qtype=None):
+    """Build a typed question.
+
+    * ``question`` already a dict {"t", "ins", "crit"}: validated and returned.
+    * ``qtype="choice"`` (default when ``options`` is given): ``options`` = {name: description or None}
+      or a list of names.
+    * ``qtype="score"``: ``options`` = list of 2..10 level descriptions (level 0 first).
+    * ``qtype="noul"`` (default when no options): a true/false statement; ``options`` may be
+      {"false": "...", "true": "..."} to describe the two outcomes.
+    """
+    if isinstance(question, dict):
+        q = dict(question)
+    else:
+        if qtype is None:
+            qtype = "choice" if options is not None else "noul"
+        if qtype == "choice":
+            if isinstance(options, (list, tuple)):
+                if len(set(map(str, options))) != len(options):
+                    raise QuestionError("duplicate option names")
+                crit = {str(o): None for o in options}
+            else:
+                crit = options
+        elif qtype == "score":
+            crit = list(options) if options is not None else None
+        else:
+            crit = options
+        q = {"t": qtype, "ins": question, "crit": crit}
+    option_names(q)
+    return q
+
+
+def serialize_state(state):
+    """Strings pass through unchanged; any other JSON value is serialised (ensure_ascii=False)."""
+    if isinstance(state, str):
+        return state
+    return json.dumps(state, ensure_ascii=False)
+
+
+def _criterion(value):
+    if isinstance(value, str):
+        return value
+    return json.dumps(value, ensure_ascii=False, separators=(", ", ": "), default=str)
+
+
+def render_options(question):
+    t, crit = question["t"], question.get("crit")
+    if t == "choice":
+        return [k if v is None or v == "" else f"{k}: {_criterion(v)}" for k, v in crit.items()]
+    if t == "score":
+        return [f"level {i}: {_criterion(c)}" for i, c in enumerate(crit)]
+    crit = crit or {}
+    false_c, true_c = crit.get("false"), crit.get("true")
+    return ["false: " + (_criterion(false_c) if false_c not in (None, "") else "no, the statement does not hold"),
+            "true: " + (_criterion(true_c) if true_c not in (None, "") else "yes, the statement holds")]
+
+
+# -- tokenizer + renderer -----------------------------------------------------------------------
+class TextEncoder:
+    """HF ``tokenizers`` tokenizer.json; no BOS/EOS, special tokens in text are split (never control tokens)."""
+
+    def __init__(self, tokenizer_json):
+        from tokenizers import Tokenizer
+        self.tk = Tokenizer.from_file(str(tokenizer_json))
+        self.tk.encode_special_tokens = True
+
+    def __call__(self, text):
+        return self.tk.encode(text, add_special_tokens=False).ids
+
+    def id_to_token(self, i):
+        return self.tk.id_to_token(int(i))
+
+    def vocab_size(self):
+        return self.tk.get_vocab_size(with_added_tokens=True)
+
+
+def _blocks(start, stop):
+    return [(s, min(s + BLOCK, stop)) for s in range(start, stop, BLOCK)]
+
+
+class Rendered:
+    """One rendered question: token ids, the verdict slots (one per option, canonical order), the
+    attention blocks ([start, stop) pairs tiling [0, len(ids))) and the state prefix length."""
+    __slots__ = ("ids", "prefix_len", "slots", "blocks", "names", "qtype", "catalogue_overflow")
+
+    def __init__(self, ids, prefix_len, slots, blocks, names, qtype, catalogue_overflow):
+        self.ids, self.prefix_len, self.slots, self.blocks = ids, prefix_len, slots, blocks
+        self.names, self.qtype, self.catalogue_overflow = names, qtype, catalogue_overflow
+
+    @property
+    def state_blocks(self):
+        return [b for b in self.blocks if b[1] <= self.prefix_len]
+
+    @property
+    def question_blocks(self):
+        return [b for b in self.blocks if b[0] >= self.prefix_len]
+
+    @property
+    def input_tokens(self):
+        return len(self.ids)
+
+    @property
+    def head_tokens(self):
+        """question + options + readout tokens (everything after the state prefix)"""
+        return len(self.ids) - self.prefix_len
+
+
+class Renderer:
+    def __init__(self, encode, readout_cfg, max_len=CONTEXT_LIMIT):
+        want = {"format": READOUT_FORMAT, "template": TEMPLATE_VERSION, "layout": LAYOUT, "readout": "verdict",
+                "block_size": BLOCK, "total_context_limit": CONTEXT_LIMIT}
+        bad = {k: readout_cfg.get(k) for k, v in want.items() if readout_cfg.get(k) != v}
+        if bad:
+            raise ValueError(f"readout_config.json does not describe this runtime's protocol {want}; got {bad}")
+        if not 0 < int(max_len) <= CONTEXT_LIMIT:
+            raise ValueError(f"max_len must be in 1..{CONTEXT_LIMIT}")
+        self.enc, self.max_len = encode, int(max_len)
+        self.head_max = self.max_len    # no separate question/options cap (field kept for the 0.8B / jev-style API)
+        st = readout_cfg["slot_tokens"]
+        self.yes, self.no, arrow = int(st["yes"]["id"]), int(st["no"]["id"]), int(st["verdict_slot"]["id"])
+        for text, want_id in ((" yes", self.yes), (" no", self.no), (" ->", arrow)):
+            got = self.enc(text)
+            if got != [want_id]:
+                raise ValueError(f"tokenizer mismatch: {text!r} -> {got}, readout_config expects [{want_id}]")
+        self.arrow = [arrow]
+        self.newline = self.enc("\n")
+        self.dash = self.enc("- ")
+        self.rubric_head = self.enc("Judge each numbered option in the complete catalogue above:\n")
+        self._numbered = {}
+
+    def _option_label(self, pos, catalogue):
+        key = (pos, catalogue)
+        if key not in self._numbered:
+            self._numbered[key] = self.enc(f"Option {pos + 1}: " if catalogue else f"Option {pos + 1}")
+        return self._numbered[key]
+
+    def prefix_ids(self, state):
+        return self.enc("State:\n") + self.enc(serialize_state(state)) + self.enc("\n\n")
+
+    def pieces(self, state, question, prefix=None):
+        """Tokenised segments of one question (``prefix``: already tokenised state, to tokenise it once)."""
+        names = option_names(question)
+        return {"prefix": self.prefix_ids(state) if prefix is None else list(prefix),
+                "head": self.enc(f"Question [{question['t']}]: {question['ins']}\nOptions:\n"),
+                "opts": [self.enc(o) for o in render_options(question)], "names": names, "qtype": question["t"]}
+
+    def assemble(self, pieces, max_len=None):
+        maximum = self.max_len if max_len is None else min(self.max_len, int(max_len))
+        prefix, head, opts = list(pieces["prefix"]), list(pieces["head"]), pieces["opts"]
+        k = len(opts)
+        if k < 1:
+            raise QuestionError("at least one option is required")
+        short, rel = list(head), []
+        for o in opts:
+            short += self.dash + list(o) + self.arrow
+            rel.append(len(short) - 1)
+            short += self.newline
+        overflow = len(short) > BLOCK
+        if not overflow:
+            ids = prefix + short
+            slots = [len(prefix) + s for s in rel]
+            blocks = _blocks(0, len(prefix)) + [(len(prefix), len(ids))]
+        else:
+            catalogue = list(head)
+            for pos, o in enumerate(opts):
+                catalogue += self._option_label(pos, True) + list(o) + self.newline
+            doc_end = len(prefix) + len(catalogue)
+            rubric, rel = list(self.rubric_head), []
+            for pos in range(k):
+                rubric += self._option_label(pos, False) + self.arrow
+                rel.append(len(rubric) - 1)
+                rubric += self.newline
+            ids = prefix + catalogue + rubric
+            slots = [doc_end + s for s in rel]
+            blocks = _blocks(0, len(prefix)) + _blocks(len(prefix), doc_end) + _blocks(doc_end, len(ids))
+        if len(ids) > maximum:
+            raise InputBudgetError(f"the complete input needs {len(ids)} tokens (state {len(prefix)} + question/"
+                                   f"options/readout {len(ids) - len(prefix)}); the limit is {maximum} tokens (model "
+                                   f"maximum {CONTEXT_LIMIT}). Nothing was truncated: shorten the state, the "
+                                   f"question or the options.")
+        return Rendered(ids, len(prefix), slots, blocks, list(pieces["names"]), pieces["qtype"], overflow)
+
+    def render(self, state, question, max_len=None, prefix=None):
+        return self.assemble(self.pieces(state, question, prefix), max_len)
+
+
+# -- calibration ----------------------------------------------------------------------------------
+def check_temperature(t):
+    try:
+        t = float(t)
+    except (TypeError, ValueError):
+        raise ValueError(f"temperature must be a number, got {t!r}") from None
+    if not (math.isfinite(t) and t > 0):
+        raise ValueError(f"temperature must be finite and > 0, got {t!r}")
+    return t
+
+
+def softmax_probabilities(scores, temperature):
+    """softmax(scores / T) in float64 (the reference computation)."""
+    z = np.asarray(scores, float) / float(temperature)
+    z = np.exp(z - z.max())
+    return z / z.sum()
+
+
+def concentration(p):
+    k = len(p)
+    if k < 2:
+        return 1.0
+    ent = -(p * np.log(np.clip(p, 1e-12, 1.0))).sum()
+    return float(np.clip(1.0 - ent / math.log(k), 0.0, 1.0))
+
+
+def _sha256(path):
+    h = hashlib.sha256()
+    with open(path, "rb") as f:
+        for b in iter(lambda: f.read(1 << 22), b""):
+            h.update(b)
+    return h.hexdigest()
+
+
+NOT_VERIFIED = ("README.md", "eval_results.json")          # documentation / records: in manifest.json, not checked
+NOT_VERIFIED_DIRS = ("assets/", "figures/", "validation/")
+
+
+def verify_manifest(model_dir, only=None):
+    """Re-hash the files listed in manifest.json (all, or those whose path starts with one of ``only``).
+    Documentation and evaluation records (README.md, eval_results.json, assets/, figures/, validation/) are
+    recorded in the manifest but not checked here (not even when named in ``only``), so a card edit never
+    makes the runtime refuse to load and a download without them still verifies."""
+    model_dir = Path(model_dir)
+    man = json.loads((model_dir / "manifest.json").read_text())
+    bad, missing, checked = [], [], 0
+    for name, rec in man["files"].items():
+        if name in NOT_VERIFIED or name.startswith(NOT_VERIFIED_DIRS):
+            continue
+        if only and not any(name == o or name.startswith(o.rstrip("/") + "/") for o in only):
+            continue
+        p = model_dir / name
+        if not p.exists():
+            missing.append(name)
+        elif _sha256(p) != rec["sha256"]:
+            bad.append(name)
+        checked += 1
+    return {"ok": not bad and not missing, "checked": checked, "bad": bad, "missing": missing}
+
+
+class DecisionBase:
+    """Backend-independent part. A backend implements ``_scores_many(rendered) -> list of list[float]``:
+    ``rendered`` is a non-empty list of Rendered that all share ONE state (identical prefix ids and
+    state blocks); it returns the raw float32 scores of every slot of every Rendered (slot order =
+    canonical option order), computing the state blocks once per call. Backends keep the most recent
+    state for the next call (exact prefix match only), so consecutive calls about the same state (e.g.
+    JSONL rows read from stdin) reuse it."""
+    backend = "base"
+
+    def _setup(self, model_dir, tokenizer_json, max_len=CONTEXT_LIMIT, temperature=None):
+        self.model_dir = Path(model_dir)
+        self.readout_config = json.loads((self.model_dir / "readout_config.json").read_text())
+        self.calibrated_temperature = check_temperature(self.readout_config["temperatures"]["global"])
+        self.default_temperature = self.calibrated_temperature if temperature is None else check_temperature(temperature)
+        self.encode = TextEncoder(tokenizer_json)
+        self.renderer = Renderer(self.encode, self.readout_config, max_len=max_len)
+
+    def _scores_many(self, rendered):
+        raise NotImplementedError
+
+    def close(self):
+        pass
+
+    def __enter__(self):
+        return self
+
+    def __exit__(self, *exc):
+        self.close()
+
+    def _score_all(self, rendered):
+        """Raw scores (lists of floats, canonical order) for Rendered sharing one state; one backend call."""
+        if not rendered:
+            return []
+        p = rendered[0].prefix_len
+        prefix, sblocks = rendered[0].ids[:p], rendered[0].state_blocks
+        for r in rendered[1:]:
+            if r.prefix_len != p or r.ids[:p] != prefix or r.state_blocks != sblocks:
+                raise ValueError("all questions of one backend call must share the same state")
+        scores = self._scores_many(rendered)
+        if len(scores) != len(rendered) or any(len(s) != len(r.slots) for s, r in zip(scores, rendered)):
+            raise RuntimeError("backend returned a wrong number of scores")
+        return [[float(x) for x in s] for s in scores]
+
+    def score_pieces(self, pieces_list, max_len=None):
+        """Raw scores for pre-tokenised questions sharing one state: dicts {"prefix", "head", "opts",
+        "names", "qtype"} of token ids (see Renderer.pieces). For parity checks; no temperature."""
+        return self._score_all([self.renderer.assemble(p, max_len) for p in pieces_list])
+
+    def _result(self, r, scores, temperature):
+        t = self.default_temperature if temperature is None else check_temperature(temperature)
+        bad = [n for n, x in zip(r.names, scores) if not math.isfinite(x)]
+        if bad:
+            raise NonFiniteScoreError(f"non-finite decision scores for option(s) {bad} ({len(r.ids)} input tokens); "
+                                      f"refusing to return probabilities. Check the weights file / engine build.")
+        p = softmax_probabilities(scores, t)
+        if not np.all(np.isfinite(p)):
+            raise NonFiniteScoreError("non-finite probabilities")
+        i = int(p.argmax())
+        return {"answer": r.names[i], "probabilities": dict(zip(r.names, p.tolist())),
+                "scores": dict(zip(r.names, scores)), "temperature": t, "top_probability": float(p[i]),
+                "entropy_concentration": concentration(p), "input_tokens": len(r.ids),
+                "state_tokens": r.prefix_len, "head_tokens": r.head_tokens, "blocks": len(r.blocks),
+                "catalogue_overflow": r.catalogue_overflow, "model": MODEL_NAME, "backend": self.backend}
+
+    def decide(self, state, question, options=None, qtype=None, category=None, temperature=None, head_max=None,
+               max_len=None):
+        """Score one question about ``state``.
+
+        Returns {"answer", "probabilities" {option: p}, "scores" {option: logit(yes)-logit(no)},
+        "temperature", "top_probability", "entropy_concentration", "input_tokens", "state_tokens",
+        "head_tokens", "blocks", "catalogue_overflow", "model", "backend"}. ``temperature`` overrides
+        the calibrated global temperature (1.0 = uncalibrated scores). ``category`` and ``head_max`` are
+        accepted for compatibility with the 0.8B v3 runtime and the jev-style package, and ignored: this
+        model has one global temperature and no separate question/options budget. Raises InputBudgetError
+        (never truncates), QuestionError or NonFiniteScoreError."""
+        q = make_question(question, options, qtype)
+        r = self.renderer.render(state, q, max_len)
+        return self._result(r, self._score_all([r])[0], temperature)
+
+    def score_many(self, state, questions, category=None, temperature=None, head_max=None, max_len=None):
+        """Several questions about ONE state: the state is tokenised and computed once and reused for
+        every question (one backend call). ``questions``: dicts {"t","ins","crit"} (or anything
+        make_question accepts). Results in order, identical to calling decide() per question. All
+        questions are rendered and budget-checked before any scoring. ``category`` / ``head_max``: see
+        decide() (accepted and ignored)."""
+        qs = [make_question(q) for q in questions]
+        if not qs:
+            return []
+        prefix = self.renderer.prefix_ids(state)
+        rs = [self.renderer.render(state, q, max_len, prefix=prefix) for q in qs]
+        return [self._result(r, sc, temperature) for r, sc in zip(rs, self._score_all(rs))]
+
+    decide_many = score_many
+
+
+def base_arg_parser(description):
+    ap = argparse.ArgumentParser(description=description)
+    ap.add_argument("--model-dir", default=str(HERE), help="folder with the weights and readout_config.json")
+    ap.add_argument("--state", help="state as plain text")
+    ap.add_argument("--state-json", help="state as a JSON value")
+    ap.add_argument("--question", help="question text (or a JSON question {'t','ins','crit'})")
+    ap.add_argument("--options", help="JSON: {name: description} or [names] (choice); [levels] (score)")
+    ap.add_argument("--qtype", choices=QTYPES)
+    ap.add_argument("--temperature", type=float,
+                    help="override the calibrated global temperature of readout_config.json (1.0 = raw scores)")
+    ap.add_argument("--max-len", type=int, default=CONTEXT_LIMIT,
+                    help=f"total token budget (state + question + options + readout), at most {CONTEXT_LIMIT}")
+    ap.add_argument("--jsonl", help="batch mode: input JSON lines {id?, state, question, options?, qtype?, "
+                                    "temperature?} ('-' = stdin); one JSON result per line on stdout. Consecutive "
+                                    "rows with an identical state share one state computation")
+    ap.add_argument("--verify", action="store_true", help="check sha256 of the files in manifest.json first")
+    return ap
+
+
+def _jsonl_groups(src, streaming):
+    """(line number, record or error text) grouped into runs of consecutive rows with one state. From a
+    file up to JSONL_GROUP_MAX rows are grouped; from stdin every row is its own group (answered at once;
+    the backend's kept state still makes consecutive identical states cheap)."""
+    group, key = [], None
+    for n, line in enumerate(src):
+        if not line.strip():
+            continue
+        try:
+            rec = json.loads(line)
+            if not isinstance(rec, dict):
+                raise ValueError("a JSONL row must be a JSON object")
+            k = serialize_state(rec.get("state", ""))
+        except ValueError as e:
+            if group:
+                yield group
+            group, key = [], None
+            yield [(n, f"{type(e).__name__}: {e}")]
+            continue
+        if group and (k != key or len(group) >= JSONL_GROUP_MAX):
+            yield group
+            group = []
+        group.append((n, rec))
+        key = k
+        if streaming:
+            yield group
+            group, key = [], None
+    if group:
+        yield group
+
+
+def _run_group(engine, group, args):
+    """Score one group of JSONL rows sharing a state. Returns (output rows, non-finite count)."""
+    out, todo = {}, []
+    prefix = None
+    for n, rec in group:
+        if isinstance(rec, str):
+            out[n] = {"id": n, "error": rec}
+            continue
+        rid = rec.get("id", n)
+        try:
+            if "question" not in rec:
+                raise QuestionError("row has no 'question'")
+            q = make_question(rec["question"], options=rec.get("options"), qtype=rec.get("qtype"))
+            t = rec.get("temperature", args.temperature)
+            t = None if t is None else check_temperature(t)
+            if prefix is None:
+                prefix = engine.renderer.prefix_ids(rec.get("state", ""))
+            todo.append((n, rid, engine.renderer.render(rec.get("state", ""), q, prefix=prefix), t))
+        except (InputBudgetError, QuestionError, ValueError, TypeError, AttributeError) as e:
+            out[n] = {"id": rid, "error": f"{type(e).__name__}: {e}"}
+    nonfinite = 0
+    if todo:
+        scores = engine._score_all([r for _, _, r, _ in todo])
+        for (n, rid, r, t), sc in zip(todo, scores):
+            try:
+                out[n] = {"id": rid, **engine._result(r, sc, t)}
+            except NonFiniteScoreError as e:
+                nonfinite += 1
+                print(f"ERROR row {rid}: NonFiniteScoreError: {e}", file=sys.stderr, flush=True)
+                out[n] = {"id": rid, "error": f"NonFiniteScoreError: {e}"}
+    return [out[n] for n, _ in group], nonfinite
+
+
+def run_cli(args, engine):
+    """Exit status: 0 = ok (JSONL rows with input errors carry an "error" field), 2 = input error
+    (single question), 3 = at least one non-finite score (refused, see stderr)."""
+    if args.jsonl:
+        streaming = args.jsonl == "-"
+        src = sys.stdin if streaming else open(args.jsonl, encoding="utf-8")
+        nonfinite = 0
+        try:
+            for group in _jsonl_groups(src, streaming):
+                rows, bad = _run_group(engine, group, args)
+                nonfinite += bad
+                for row in rows:
+                    print(json.dumps(row, ensure_ascii=False), flush=True)
+        finally:
+            if not streaming:
+                src.close()
+        return 3 if nonfinite else 0
+    if args.question is None:
+        raise SystemExit("--question (or --jsonl) is required")
+    try:
+        state = json.loads(args.state_json) if args.state_json is not None else (args.state or "")
+        question = args.question
+        if question.lstrip().startswith("{"):
+            try:                                            # a JSON question {'t','ins','crit'}; else plain text
+                parsed = json.loads(question)
+            except ValueError:
+                parsed = None
+            if isinstance(parsed, dict):
+                question = parsed
+        options = json.loads(args.options) if args.options else None
+        res = engine.decide(state, question, options=options, qtype=args.qtype, temperature=args.temperature)
+    except (InputBudgetError, QuestionError, ValueError, TypeError) as e:  # NonFiniteScoreError is not a ValueError
+        print(f"error: {type(e).__name__}: {e}", file=sys.stderr)
+        return 2
+    except NonFiniteScoreError as e:
+        print(f"ERROR: NonFiniteScoreError: {e}", file=sys.stderr)
+        return 3
+    print(json.dumps(res, ensure_ascii=False, indent=2))
+    return 0
+# ---------------------------------------------------------------------------- end of shared core
+
+
+# ---------------------------------------------------------------------------------- MLX backend
+# Repository layout (one runtime, two precisions):
+#
+#   jev_style_decision_mlx.py, readout_config.json, release_config.json, manifest.json, ...   shared
+#   bf16/   config.json, model.safetensors (bfloat16), macjev_norms_fp32.safetensors, tokenizer ...
+#   8bit/   config.json, model.safetensors (affine 8-bit, group 64), macjev_norms_fp32.safetensors, tokenizer ...
+#
+# Either folder may be downloaded alone (plus the shared top-level files).
+#
+# Block-causal attention on mlx-lm: the input is fed to the model ONE protocol block per call on an mlx-lm
+# prompt cache. The 6 full-attention layers use BlockKVCache, whose make_mask returns None, so the queries of
+# the current block attend to every cached key plus every key of the block itself, with no causal mask inside
+# the block. The 18 Gated-DeltaNet layers keep their recurrent cache (conv + recurrent state), causal by
+# construction. The state blocks are computed once; each question then runs its block(s) on an independent
+# copy of that cache, so questions never see each other and the state is reused (score_many, JSONL rows).
+#
+# Two corrections to stock mlx-lm 0.31.3 (both needed to match the HF / training numerics):
+#   1. Gated-DeltaNet q/k normalisation: mlx-lm uses rms_norm(x, eps=1e-6) (= l2norm with eps Dk*1e-6); HF / fla /
+#      llama.cpp use l2norm(x, eps=1e-6). A copy of mlx-lm's GatedDeltaNet.__call__ with only that eps changed is
+#      swapped in per instance (never globally).
+#   2. Qwen3.5 RMSNorm is x_hat * (1 + w). mlx-lm's converter adds the 1 in bf16, so the export stores bf16(1 + w).
+#      The exact FP32 (1 + w) ships as the sidecar macjev_norms_fp32.safetensors (ignored by stock mlx-lm); the
+#      runtime installs it and computes those norms in FP32 (cast back to the input dtype, as HF does).
+# Both depend on mlx-lm internals, so the runtime checks the sha256 of the mlx-lm source it patches or relies on
+# and refuses to run on any other mlx-lm version (install mlx-lm==0.31.3). The sidecar and readout_config.json are
+# mandatory: the runtime never falls back to stock norms or to T = 1.
+REPO_ID = "chaoliangUNSW/Jev-Style-2B-Decision-v3-MLX"
+PRECISIONS = ("bf16", "8bit")
+DEFAULT_PRECISION = "bf16"
+# activation dtype per precision: None = the checkpoint's native activations (bf16 residual stream), the
+# setting the release scoring used for both precisions (release_config.json -> runtime.precisions overrides)
+VALIDATED_COMPUTE_DTYPE = {"bf16": None, "8bit": None}
+MLX_CACHE_LIMIT_GIB = 2.0
+MLX_LM_VERSION = "0.31.3"
+NORM_SIDECAR = "macjev_norms_fp32.safetensors"
+NORM_SIDECAR_FORMAT = "macjev-norms-fp32-v1"
+SHIFTED_NORMS = ("input_layernorm", "post_attention_layernorm", "q_norm", "k_norm")
+# sha256 of inspect.getsource(...) in mlx-lm 0.31.3 for every function the MLX path patches (GatedDeltaNet) or relies
+# on for the block mask (a None mask from BlockKVCache must reach scaled_dot_product_attention unchanged)
+MLX_LM_SOURCE_SHA256 = {
+    "qwen3_5.GatedDeltaNet.__call__": "9f27bdc613912fa9b3b0509d8e3f684ac672810bf45e29831f5815a1c3b43091",
+    "qwen3_5.Qwen3_5TextModel.__call__": "45fd379f8885e93b4ba685c3d6c2aa1b891d535bb00fd042c6b1e56613345b6d",
+    "base.create_attention_mask": "a8ebacf963b3f96e9ad838a7a870d1ef6da33aec5233b133b0d0c744c3b33e5e",
+    "qwen3_5.Attention.__call__": "a938424855f1b919d1bcfd62b105e10723f7de295f5ee51a19924b502ddc86a6",
+    "base.scaled_dot_product_attention": "8deba4a8a8fb4e29f1e69ea81701da68f08f90656e1421a1e7a82332aadc41c1",
+}
+
+
+class MLXVersionError(RuntimeError):
+    """The installed mlx-lm is not the version whose internals the runtime was validated against."""
+
+
+def check_mlx_lm():
+    """Raise MLXVersionError unless the mlx-lm source this runtime patches is byte-identical to mlx-lm 0.31.3."""
+    import inspect
+    import mlx_lm
+    from mlx_lm.models import base, qwen3_5
+    objs = {"qwen3_5.GatedDeltaNet.__call__": qwen3_5.GatedDeltaNet.__call__,
+            "qwen3_5.Qwen3_5TextModel.__call__": qwen3_5.Qwen3_5TextModel.__call__,
+            "base.create_attention_mask": base.create_attention_mask,
+            "qwen3_5.Attention.__call__": qwen3_5.Attention.__call__,
+            "base.scaled_dot_product_attention": base.scaled_dot_product_attention}
+    changed = [k for k, f in objs.items()
+               if hashlib.sha256(inspect.getsource(f).encode()).hexdigest() != MLX_LM_SOURCE_SHA256[k]]
+    if changed:
+        raise MLXVersionError(
+            f"this runtime needs mlx-lm=={MLX_LM_VERSION} (installed: {getattr(mlx_lm, '__version__', '?')}): it "
+            f"corrects the Qwen3.5 Gated-DeltaNet q/k norm and the (1 + w) RMSNorm weights inside mlx-lm, and the "
+            f"source it patches differs ({', '.join(changed)}). Install the tested version with: "
+            f"pip install 'mlx-lm=={MLX_LM_VERSION}'")
+
+
+# Third-party code: _gdn_hf_call below is a modified copy of GatedDeltaNet.__call__ from mlx_lm/models/qwen3_5.py,
+# mlx-lm 0.31.3 (https://github.com/ml-explore/mlx-lm), MIT License (mlx-lm LICENSE: "Copyright © 2023 Apple Inc.";
+# qwen3_5.py header: "Copyright © 2026 Apple Inc."). Changes:
+# q/k rms_norm eps 1e-6 -> 1e-6 / Dk (HF-exact l2norm), sharding branches removed (refused), module-level function.
+# The MIT copyright and permission notice is reproduced in THIRD_PARTY_NOTICES.md of this repository.
+def _gdn_hf_call(self, inputs, mask=None, cache=None):
+    """mlx-lm 0.31.3 qwen3_5.GatedDeltaNet.__call__ with HF-exact q/k l2norm (eps 1e-6). No sharding."""
+    import mlx.core as mx
+    import mlx.nn as nn
+    from mlx_lm.models.gated_delta import gated_delta_update
+    if self.sharding_group is not None:
+        raise ValueError("sharded Gated-DeltaNet is not supported by this runtime")
+    B, S, _ = inputs.shape
+    qkv = self.in_proj_qkv(inputs)
+    z = self.in_proj_z(inputs).reshape(B, S, self.num_v_heads, self.head_v_dim)
+    b = self.in_proj_b(inputs)
+    a = self.in_proj_a(inputs)
+    if cache is not None and cache[0] is not None:
+        conv_state = cache[0]
+    else:
+        conv_state = mx.zeros((B, self.conv_kernel_size - 1, self.conv_dim), dtype=inputs.dtype)
+    if mask is not None:
+        qkv = mx.where(mask[..., None], qkv, 0)
+    conv_input = mx.concatenate([conv_state, qkv], axis=1)
+    if cache is not None:
+        n_keep = self.conv_kernel_size - 1
+        if cache.lengths is not None:
+            ends = mx.clip(cache.lengths, 0, S)
+            positions = (ends[:, None] + mx.arange(n_keep))[..., None]
+            cache[0] = mx.take_along_axis(conv_input, positions, axis=1)
+        else:
+            cache[0] = mx.contiguous(conv_input[:, -n_keep:, :])
+    conv_out = nn.silu(self.conv1d(conv_input))
+    q, k, v = [t.reshape(B, S, h, d) for t, h, d in zip(
+        mx.split(conv_out, [self.key_dim, 2 * self.key_dim], -1),
+        [self.num_k_heads, self.num_k_heads, self.num_v_heads],
+        [self.head_k_dim, self.head_k_dim, self.head_v_dim])]
+    state = cache[1] if cache else None
+    dk = k.shape[-1]
+    inv_scale = dk ** -0.5
+    # l2norm(x, 1e-6) == rms_norm(x, eps=1e-6/Dk) / sqrt(Dk); q additionally carries the 1/sqrt(Dk) scale
+    q = (inv_scale ** 2) * mx.fast.rms_norm(q, None, 1e-6 / dk)
+    k = inv_scale * mx.fast.rms_norm(k, None, 1e-6 / dk)
+    out, state = gated_delta_update(q, k, v, a, b, self.A_log, self.dt_bias, state, mask,
+                                    use_kernel=not self.training)
+    if cache is not None:
+        cache[1] = state
+        cache.advance(S)
+    out = self.norm(out, z)
+    return self.out_proj(out.reshape(B, S, -1))
+
+
+def patch_gdn_l2norm(inner):
+    """Swap every GatedDeltaNet instance of ``inner`` to the HF-exact l2norm class (per instance). -> count."""
+    from mlx_lm.models.qwen3_5 import GatedDeltaNet
+    cls = type("GatedDeltaNetHFL2", (GatedDeltaNet,), {"__call__": _gdn_hf_call})
+    n = 0
+    for layer in inner.layers:
+        if getattr(layer, "is_linear", False):
+            if type(layer.linear_attn) is not GatedDeltaNet:
+                raise TypeError(f"unexpected Gated-DeltaNet class {type(layer.linear_attn).__name__}")
+            layer.linear_attn.__class__ = cls
+            n += 1
+    return n
+
+
+def load_norm_sidecar(path):
+    """{mlx parameter name: FP32 (1 + w)} from macjev_norms_fp32.safetensors (format checked)."""
+    import mlx.core as mx
+    arrays, meta = mx.load(str(path), return_metadata=True)
+    if (meta or {}).get("format") != NORM_SIDECAR_FORMAT:
+        raise ValueError(f"{path} is not a {NORM_SIDECAR_FORMAT} norm sidecar (metadata {meta!r})")
+    mx.eval(arrays)                                   # read now (mx.load is lazy)
+    return arrays
+
+
+def apply_exact_norms(inner, norms):
+    """Install FP32 (1 + w) into every shifted RMSNorm of ``inner`` (per-instance class swap). -> count.
+    Refuses a sidecar that misses a norm, names an unknown module or does not belong to these weights."""
+    import mlx.core as mx
+    import mlx.nn as nn
+
+    def _call(self, x):
+        return mx.fast.rms_norm(x.astype(mx.float32), self.weight, self.eps).astype(x.dtype)
+
+    cls = type("RMSNormFP32Weight", (nn.RMSNorm,), {"__call__": _call})
+    modules = dict(inner.named_modules())
+    seen = set()
+    for name, w in norms.items():
+        mod_name = name[:-len(".weight")] if name.endswith(".weight") else name
+        mod = modules.get(mod_name)
+        if mod is None or type(mod) is not nn.RMSNorm:
+            raise KeyError(f"norm sidecar entry {name} is not an RMSNorm of this model")
+        w = w.astype(mx.float32)
+        if w.shape != mod.weight.shape:
+            raise ValueError(f"norm sidecar shape mismatch at {name}")
+        # the sidecar must be the (1 + w) of THESE weights (up to the bf16 rounding it corrects)
+        if float(mx.abs(w - mod.weight.astype(mx.float32)).max()) > 2 ** -7 * float(mx.abs(w).max()) + 1e-6:
+            raise ValueError(f"norm sidecar does not match the model weights at {name}")
+        mod.weight = w
+        mod.__class__ = cls
+        seen.add(mod_name)
+    missing = [k for k, m in modules.items() if type(m) is nn.RMSNorm and k not in seen
+               and (k == "norm" or k.endswith(SHIFTED_NORMS))]
+    if missing:
+        raise KeyError(f"norm sidecar misses {missing[:3]}")
+    mx.eval(inner.parameters())
+    return len(seen)
+
+
+def _block_kv_class():
+    from mlx_lm.models.cache import KVCache
+
+    class BlockKVCache(KVCache):
+        """KV cache whose make_mask is None: the current block attends to all cached keys and to itself
+        (Qwen3_5TextModel builds one mask from the first full-attention layer's cache)."""
+
+        def make_mask(self, *args, **kwargs):
+            return None
+
+    return BlockKVCache
+
+
+def copy_cache(cache):
+    """Independent copy of a prompt cache (KV + ArraysCache). KV copies are trimmed to ``offset`` (the next
+    update re-allocates by concatenation); ArraysCache gets a new list (GDN layers assign new arrays)."""
+    import copy
+    from mlx_lm.models.cache import ArraysCache, KVCache
+    out = []
+    for c in cache:
+        n = copy.copy(c)
+        if isinstance(c, KVCache):
+            if c.keys is not None:
+                n.keys = c.keys[..., :c.offset, :]
+                n.values = c.values[..., :c.offset, :]
+        elif isinstance(c, ArraysCache):
+            n.cache = list(c.cache)
+            if c.lengths is not None or c.left_padding is not None:
+                raise ValueError("batched ArraysCache state is not supported")
+        else:
+            raise TypeError(f"unsupported cache type {type(c).__name__}")
+        out.append(n)
+    return out
+
+
+def _set_mlx_cache_limit(mx, gib):
+    """Bound MLX's freed-buffer cache (process-wide); never raises a lower limit set by the host."""
+    if gib is None:
+        return None
+    want = int(float(gib) * (1 << 30))
+    prev = mx.set_cache_limit(want)
+    if prev is not None and prev < want:
+        mx.set_cache_limit(prev)
+        return int(prev)
+    return want
+
+
+def _weights_precision(weights_dir):
+    """'bf16' or '8bit' (or '<n>bit') from the MLX config.json of a weights folder."""
+    cfg = json.loads((Path(weights_dir) / "config.json").read_text())
+    q = cfg.get("quantization") or cfg.get("quantization_config")
+    return f"{int(q['bits'])}bit" if q else "bf16"
+
+
+def _require_weights(w, precision, repo):
+    need = ["config.json", "model.safetensors", "tokenizer.json", NORM_SIDECAR]
+    miss = [n for n in need if not (w / n).is_file()]
+    if miss:
+        extra = (f" {NORM_SIDECAR} holds the exact FP32 (1 + w) RMSNorm weights; without it mlx-lm would use "
+                 f"bf16-rounded norms, so the runtime refuses to load." if NORM_SIDECAR in miss else "")
+        raise FileNotFoundError(f"{w} is missing {', '.join(miss)}.{extra} Download the folder with: "
+                                f"hf download {REPO_ID} --include \"{precision}/*\" --local-dir {repo}")
+
+
+def resolve_model_dirs(model_dir=HERE, precision=None):
+    """-> (repo_dir, weights_dir, precision).
+
+    ``model_dir`` is the repository folder (holding bf16/ and/or 8bit/ next to readout_config.json),
+    or one precision folder itself (then ``precision`` defaults to the precision stored there).
+    ``precision`` None means "bf16" for a repository folder."""
+    if precision is not None and precision not in PRECISIONS:
+        raise ValueError(f"precision must be one of {PRECISIONS}, got {precision!r}")
+    d = Path(model_dir).resolve()
+    # a precision folder given directly (the repository root also holds a config.json, a copy of bf16/config.json,
+    # so a folder counts as a precision folder only if it has its own weights or no bf16/ / 8bit/ subfolder)
+    if (d / "config.json").is_file() and ((d / "model.safetensors").is_file()
+                                          or not any((d / p).is_dir() for p in PRECISIONS)):
+        found = _weights_precision(d)
+        if precision is not None and precision != found:
+            raise ValueError(f"{d} holds {found} weights, not {precision}")
+        repo = d if (d / "readout_config.json").is_file() else d.parent
+        _require_weights(d, found, repo)
+        return repo, d, found
+    precision = precision or DEFAULT_PRECISION
+    w = d / precision
+    if not ((w / "config.json").is_file() and (w / "model.safetensors").is_file()):
+        have = [p for p in PRECISIONS if (d / p / "config.json").is_file() and (d / p / "model.safetensors").is_file()]
+        msg = f"the {precision} weights are not in {d} (expected {precision}/config.json and {precision}/model.safetensors)."
+        if have:
+            msg += f" This folder has only: {', '.join(p + '/' for p in have)}; use --precision {have[0]} (precision=\"{have[0]}\"), or"
+        msg += f" download them with: hf download {REPO_ID} --include \"{precision}/*\" --local-dir {d}"
+        raise FileNotFoundError(msg)
+    found = _weights_precision(w)
+    if found != precision:
+        raise ValueError(f"{w} holds {found} weights, not {precision}")
+    _require_weights(w, precision, d)
+    return d, w, precision
+
+
+def verify_repo(repo_dir, weights_dir):
+    """manifest check of the shared top-level files and of the selected precision folder only (the other
+    precision may be absent). Documentation and records (README.md, figures/, validation/) are skipped as in
+    verify_manifest."""
+    repo_dir, weights_dir = Path(repo_dir), Path(weights_dir)
+    if weights_dir == repo_dir:
+        return verify_manifest(repo_dir)
+    sub = weights_dir.relative_to(repo_dir).as_posix()
+    names = json.loads((repo_dir / "manifest.json").read_text())["files"]
+    if not any(n.startswith(sub + "/") for n in names):
+        return {"ok": False, "checked": 0, "bad": [], "missing": [f"{sub}/ (not listed in manifest.json)"]}
+    res = verify_manifest(repo_dir, only=[n for n in names if "/" not in n] + [sub + "/"])
+    res["precision_folder"] = sub + "/"
+    return res
+
+
+def release_compute_dtype(repo_dir, precision):
+    """Activation dtype of ``precision`` (release_config.json -> runtime.precisions, else the built-in default)."""
+    p = Path(repo_dir) / "release_config.json"
+    if p.is_file():
+        per = (json.loads(p.read_text()).get("runtime", {}).get("precisions") or {}).get(precision)
+        if per is not None and "mlx_compute_dtype" in per:
+            return per["mlx_compute_dtype"]
+    return VALIDATED_COMPUTE_DTYPE[precision]
+
+
+class _MLXEngine:
+    """Block-causal v2 scoring on mlx-lm (see the MLX backend comment above). Stateless apart from the last
+    state cache (kept so consecutive requests with the same state reuse it)."""
+
+    def __init__(self, weights_dir, yes_id, no_id, compute_dtype=None, cache_limit_gib=MLX_CACHE_LIMIT_GIB):
+        import mlx.core as mx
+        check_mlx_lm()
+        from mlx_lm.utils import load_model
+        self.mx = mx
+        self.cache_limit_bytes = _set_mlx_cache_limit(mx, cache_limit_gib)
+        norms = load_norm_sidecar(Path(weights_dir) / NORM_SIDECAR)
+        self.model, self.config = load_model(Path(weights_dir))
+        if compute_dtype:
+            self.model.set_dtype(getattr(mx, compute_dtype))
+        self.compute_dtype = compute_dtype
+        lm = getattr(self.model, "language_model", self.model)
+        self.inner = lm.model
+        args = getattr(lm, "args", None) or getattr(self.model, "args", None)
+        if not bool(getattr(args, "tie_word_embeddings", True)) or hasattr(lm, "lm_head"):
+            raise ValueError("expected tied input/output embeddings")
+        self.n_gdn_patched = patch_gdn_l2norm(self.inner)
+        self.n_norms_exact = apply_exact_norms(self.inner, norms)
+        self.n_full_attention = sum(1 for layer in self.inner.layers if not layer.is_linear)
+        w = self.inner.embed_tokens(mx.array([int(yes_id), int(no_id)]))   # dequantised rows if quantised
+        w = w.astype(mx.float32)
+        self.direction = w[0] - w[1]
+        mx.eval(self.direction)
+        self._state = None                      # (tuple(prefix ids), cache) of the last state
+
+    def _make_cache(self):
+        from mlx_lm.models.cache import ArraysCache
+        kv = _block_kv_class()
+        return [ArraysCache(size=2) if layer.is_linear else kv() for layer in self.inner.layers]
+
+    def _eval_cache(self, cache):
+        arrays = []
+        for c in cache:
+            if hasattr(c, "keys"):
+                if c.keys is not None:
+                    arrays += [c.keys, c.values]
+            else:
+                arrays += [a for a in c.cache if a is not None]
+        self.mx.eval(arrays)
+
+    def state_cache(self, prefix, blocks):
+        """Cache after the state blocks (computed once; reused while the state is unchanged). -> (cache, reused)"""
+        key = tuple(prefix)
+        if self._state is not None and self._state[0] == key:
+            return self._state[1], True
+        self._state = None
+        cache = self._make_cache()
+        for a, b in blocks:
+            self.inner(self.mx.array(prefix[a:b])[None], cache=cache)
+            self._eval_cache(cache)
+        self._state = (key, cache)
+        return cache, False
+
+    def question_scores(self, ids, slots, qblocks, state_cache):
+        """Question block(s) on a copy of the state cache -> raw FP32 scores, one per slot (rendered order)."""
+        mx = self.mx
+        cache = copy_cache(state_cache)
+        found = {}
+        for i, (a, b) in enumerate(qblocks):
+            h = self.inner(mx.array(ids[a:b])[None], cache=cache)[0]
+            inside = [(j, s) for j, s in enumerate(slots) if a <= s < b]
+            if inside:
+                hs = h[mx.array([s - a for _, s in inside])].astype(mx.float32)
+                v = hs @ self.direction
+                mx.eval(v)
+                for (j, _), x in zip(inside, v.tolist()):
+                    found[j] = x
+            if i + 1 < len(qblocks):
+                self._eval_cache(cache)
+        if len(found) != len(slots):
+            raise RuntimeError("a verdict slot lies outside the question blocks")
+        return [float(found[j]) for j in range(len(slots))]
+
+    def reset(self):
+        self._state = None
+
+    def close(self):
+        self._state = None
+        self.mx.clear_cache()
+
+
+def _check_blocks(r):
+    """The protocol invariants a backend relies on: blocks tile [0, n) in order, each <= BLOCK tokens, the
+    state ends on a block boundary, every slot lies in a question block."""
+    n, p, end = len(r.ids), r.prefix_len, 0
+    for a, b in r.blocks:
+        if a != end or not a < b <= n or b - a > BLOCK:
+            raise ValueError(f"invalid block layout {r.blocks}")
+        end = b
+    sb, qb = r.state_blocks, r.question_blocks
+    if end != n or len(sb) + len(qb) != len(r.blocks) or (sb and sb[-1][1] != p) or not qb:
+        raise ValueError(f"invalid block layout {r.blocks} (prefix {p}, {n} tokens)")
+    if any(not qb[0][0] <= s < n for s in r.slots):
+        raise ValueError("verdict slots must lie in the question blocks")
+
+
+class JevStyleDecisionMLX(DecisionBase):
+    """Apple-silicon runtime on mlx / mlx-lm 0.31.3 (qwen3_5 model code, two numerics corrections).
+
+    ``model_dir``: the repository folder (default: the folder of this script) holding bf16/ and/or 8bit/ next to
+    readout_config.json, or one precision folder. ``precision``: "bf16" (default) or "8bit".
+    ``temperature``: overrides the calibrated global temperature for every call (default: readout_config.json).
+    ``compute_dtype``: "release" (default) = the activation dtype of the chosen precision in release_config.json
+    (native bf16 activations for both precisions); "float32" or None (native) override it.
+    ``max_len``: total token budget (<= 25,600). ``verify``: re-hash the files in manifest.json before loading.
+
+    The most recent state stays computed: score_many() and consecutive calls / JSONL rows with an identical
+    state reuse it (``last_timing["state_reused"]``)."""
+    backend = "mlx"
+
+    def __init__(self, model_dir=HERE, precision=None, temperature=None, max_len=CONTEXT_LIMIT,
+                 compute_dtype="release", cache_limit_gib=MLX_CACHE_LIMIT_GIB, verify=False):
+        repo_dir, weights_dir, precision = resolve_model_dirs(model_dir, precision)
+        self.repo_dir, self.weights_dir, self.precision = repo_dir, weights_dir, precision
+        if not (repo_dir / "readout_config.json").is_file():
+            raise FileNotFoundError(f"readout_config.json (readout + calibration temperature) is not in {repo_dir}; "
+                                    f"the runtime does not guess a temperature. Download it with: hf download "
+                                    f"{REPO_ID} readout_config.json release_config.json --local-dir {repo_dir}")
+        if verify:
+            res = verify_repo(repo_dir, weights_dir)
+            if not res["ok"]:
+                raise RuntimeError(f"integrity check failed: {res}")
+        self._setup(repo_dir, weights_dir / "tokenizer.json", max_len, temperature)
+        if compute_dtype == "release":
+            compute_dtype = release_compute_dtype(repo_dir, precision)
+        if compute_dtype not in (None, "float32", "bfloat16", "float16"):
+            raise ValueError(f"unsupported compute_dtype {compute_dtype!r}")
+        self.engine = _MLXEngine(weights_dir, self.renderer.yes, self.renderer.no, compute_dtype, cache_limit_gib)
+        self.compute_dtype = compute_dtype
+        self.last_timing = {}
+
+    def _scores_many(self, rendered):
+        import time
+        for r in rendered:
+            _check_blocks(r)
+        r0 = rendered[0]
+        t0 = time.perf_counter()
+        cache, reused = self.engine.state_cache(r0.ids[:r0.prefix_len], r0.state_blocks)
+        t1 = time.perf_counter()
+        out = [self.engine.question_scores(r.ids, r.slots, r.question_blocks, cache) for r in rendered]
+        self.last_timing = {"state_tokens": r0.prefix_len, "state_reused": reused, "state_ms": (t1 - t0) * 1000,
+                            "questions": len(rendered), "questions_ms": (time.perf_counter() - t1) * 1000}
+        return out
+
+    def close(self):
+        self.engine.close()
+
+
+def main(argv=None):
+    ap = base_arg_parser(f"{MODEL_NAME}: typed decisions with MLX (Apple silicon)")
+    for a in ap._actions:
+        if a.dest == "model_dir":
+            a.help = ("repository folder holding bf16/ and/or 8bit/ next to readout_config.json (default: the "
+                      "folder of this script), or one precision folder")
+    ap.add_argument("--precision", choices=PRECISIONS, default=None,
+                    help="weights to load: bf16 (default) or 8bit (affine 8-bit, group 64)")
+    ap.add_argument("--compute-dtype", default="release", choices=["release", "float32", "native"],
+                    help="activation dtype: the release setting of the precision (release = native), float32, or "
+                         "the checkpoint's native dtype")
+    ap.add_argument("--cache-limit-gib", type=float, default=MLX_CACHE_LIMIT_GIB)
+    args = ap.parse_args(argv)
+    cdt = None if args.compute_dtype == "native" else args.compute_dtype
+    try:
+        engine = JevStyleDecisionMLX(args.model_dir, precision=args.precision, max_len=args.max_len,
+                                     compute_dtype=cdt, cache_limit_gib=args.cache_limit_gib, verify=args.verify)
+    except (FileNotFoundError, ValueError, MLXVersionError, RuntimeError, ImportError) as e:
+        raise SystemExit(f"error: {e}")
+    try:
+        return run_cli(args, engine)
+    finally:
+        engine.close()
+
+
+if __name__ == "__main__":
+    raise SystemExit(main())

manifest.json

@@ -0,0 +1,190 @@
+{
+  "format": "jev-style-manifest-v1",
+  "repo": "chaoliangUNSW/Jev-Style-2B-Decision-v3-MLX",
+  "created_unix": 1790476271.175225,
+  "files": {
+    "8bit/chat_template.jinja": {
+      "sha256": "273d8e0e683b885071fb17e08d71e5f2a5ddfb5309756181681de4f5a1822d80",
+      "bytes": 7755
+    },
+    "8bit/config.json": {
+      "sha256": "8865dd86a4e561e6b036667cd0d5c1c39ad12527b3feda7c585cf116d890477a",
+      "bytes": 2205
+    },
+    "8bit/generation_config.json": {
+      "sha256": "62153eb6c69f2e1f426beaa8002b7186437e949c7588167085df14e10e9c0a73",
+      "bytes": 116
+    },
+    "8bit/macjev_norms_fp32.safetensors": {
+      "sha256": "02f08c422228929c44dd84856678ea95d41401138aa56b1b771a654376d3e51c",
+      "bytes": 420112
+    },
+    "8bit/model.safetensors": {
+      "sha256": "184b4dda2f1285fe95a00a6271b364c861d370fbf913030f0259b2423f88a740",
+      "bytes": 2000043057
+    },
+    "8bit/model.safetensors.index.json": {
+      "sha256": "54d94d008c07e39846c0b454f68055622bb635a0623b51bce86d724e18733aab",
+      "bytes": 60786
+    },
+    "8bit/tokenizer.json": {
+      "sha256": "06b9509352d2af50381ab2247e083b80d32d5c0aba91c272ca9ff729b6a0e523",
+      "bytes": 19989325
+    },
+    "8bit/tokenizer_config.json": {
+      "sha256": "95c557768e6b88a7128befc7bfd3c7de50e5d51af9b8b33a9f4dee0e04f99679",
+      "bytes": 1161
+    },
+    "LICENSE": {
+      "sha256": "bbedc3fda3305820b977265f01b8619d87570a6739de3a5582c3464840f1e57a",
+      "bytes": 11544
+    },
+    "NOTICE": {
+      "sha256": "39ce15275bd1433d18dd09066e9045529a3c7f1a8c3e761d3f318041167e9fed",
+      "bytes": 2664
+    },
+    "README.md": {
+      "sha256": "965607319a7721d4eb497128179a740b2c9ca9516761c028c44fa8c6b4d20733",
+      "bytes": 13399
+    },
+    "THIRD_PARTY_NOTICES.md": {
+      "sha256": "55ade7a70bb6bfcb8f0a8e9f4aef9222cde16c8bb44b6135aab6f929c706cbc5",
+      "bytes": 2146
+    },
+    "bf16/chat_template.jinja": {
+      "sha256": "273d8e0e683b885071fb17e08d71e5f2a5ddfb5309756181681de4f5a1822d80",
+      "bytes": 7755
+    },
+    "bf16/config.json": {
+      "sha256": "1dd31077b37706dc40e1bbddedf5eeba3e50e1d3c85d1fe34bfd5957a8d6ad14",
+      "bytes": 2000
+    },
+    "bf16/generation_config.json": {
+      "sha256": "62153eb6c69f2e1f426beaa8002b7186437e949c7588167085df14e10e9c0a73",
+      "bytes": 116
+    },
+    "bf16/macjev_norms_fp32.safetensors": {
+      "sha256": "a30e1a72d02598b405978c0eff08c3bfca3b523fa0b3d4b4c80d2a185e0b8081",
+      "bytes": 420112
+    },
+    "bf16/model.safetensors": {
+      "sha256": "00629c8d17ea115c21c3fc0d6ddc5d4aa02887343a8ff3b3cf79ab197624d574",
+      "bytes": 3763691755
+    },
+    "bf16/model.safetensors.index.json": {
+      "sha256": "d49d69fd9257879080337533c6051a14ff4f561a79b9f70d8bdb74e47444a306",
+      "bytes": 28024
+    },
+    "bf16/tokenizer.json": {
+      "sha256": "06b9509352d2af50381ab2247e083b80d32d5c0aba91c272ca9ff729b6a0e523",
+      "bytes": 19989325
+    },
+    "bf16/tokenizer_config.json": {
+      "sha256": "95c557768e6b88a7128befc7bfd3c7de50e5d51af9b8b33a9f4dee0e04f99679",
+      "bytes": 1161
+    },
+    "config.json": {
+      "sha256": "1dd31077b37706dc40e1bbddedf5eeba3e50e1d3c85d1fe34bfd5957a8d6ad14",
+      "bytes": 2000
+    },
+    "figures/banner.data.json": {
+      "sha256": "397cbfb222986f8d59a65143f2fec5a11170385c661a86c5e54d890a24d7d89e",
+      "bytes": 1109
+    },
+    "figures/banner.png": {
+      "sha256": "2623e4afad36239aaa007c34350d6f2fbacb33430ffd3c36ce2f7ee7db849b47",
+      "bytes": 447467
+    },
+    "figures/jevbench.data.json": {
+      "sha256": "508994afea8d8b1bc33e25c675621a9b3b55b01f900daa688c240c8356dffe3b",
+      "bytes": 2387
+    },
+    "figures/jevbench.png": {
+      "sha256": "c8542dba9ecc1bd2a4ea3bda844cbc3fbd661c40d6eb9254098b279a92a74ce8",
+      "bytes": 167843
+    },
+    "figures/jevbench.svg": {
+      "sha256": "13193f8d636ad5e6685cbc58863f27bced87c0ead94c252917839c7c6eed837e",
+      "bytes": 12280
+    },
+    "figures/zeroshot.data.json": {
+      "sha256": "3ad81c3dc926c988db5c339eec4991c2e2cce46faafb89db99cc47d6d8737538",
+      "bytes": 1841
+    },
+    "figures/zeroshot.png": {
+      "sha256": "d51f9a7bc46b88d8dcf5ba46f15d479c781302ca531208ff0838106da8aae755",
+      "bytes": 137714
+    },
+    "figures/zeroshot.svg": {
+      "sha256": "f3234ba4715b0d87e081ec9fc52c0b409010612f54db7aa39307a9eda3b7545f",
+      "bytes": 13874
+    },
+    "jev_style_decision_mlx.py": {
+      "sha256": "025daab9b374d740e743c519bd3aa3385791e08c81f4515b2ae525af57788030",
+      "bytes": 54020
+    },
+    "readout_config.json": {
+      "sha256": "4af0d578c9126d4eb1a545b6e576e0a49a967243a47b8e99e3555f081fdaa1de",
+      "bytes": 2031
+    },
+    "release_config.json": {
+      "sha256": "eddb00c6bf6682f3f62d0c69ffa3ef3820454abd8bea092c7ecddece55f5b551",
+      "bytes": 17829
+    },
+    "requirements.txt": {
+      "sha256": "2dd1d057db2c4f0d906e3665745e85d4647be032e1ffd2cc94cea2db2104c329",
+      "bytes": 479
+    },
+    "validation/SOURCES.json": {
+      "sha256": "fa8510c5c1f3afbdb01a3da6f26500cfa3117b8f4e26ceaf788fbf1eca9a5639",
+      "bytes": 3165
+    },
+    "validation/latency_2b.json": {
+      "sha256": "458d00e0c81b3165072b8547278e6039b9ae0d45d54b7421008ef24c1d24ca3d",
+      "bytes": 40294
+    },
+    "validation/parity/PREDECLARED_RELEASE_GATES_2B.md": {
+      "sha256": "783aec77300779582911cdc57c71cf8b101b0cd5ca040073f7e1fa3ac20af08e",
+      "bytes": 2619
+    },
+    "validation/parity/compare_mlx_affine8-g64_base_vs_block.json": {
+      "sha256": "e1911f21ce73dd42d6c57e8349e2ec3221386a2b44ca461a3e50b9a69057ec34",
+      "bytes": 7697
+    },
+    "validation/parity/compare_mlx_affine8-g64_gate_vs_block.json": {
+      "sha256": "f0e2f44498df00945901c2a9bd5bff6f742d11ab243495c8089eaf0448fda782",
+      "bytes": 6494
+    },
+    "validation/parity/compare_mlx_bf16_base_vs_block.json": {
+      "sha256": "ca16e5281debaf1bc601b4a8c93cf6c1106f7b032f35c5833baa4fc8178def7c",
+      "bytes": 7408
+    },
+    "validation/parity/compare_mlx_bf16_gate_vs_block.json": {
+      "sha256": "7fa30fa414a4ffc12a7aa6371e77579845c3b816b90da3a813a70fec10ccd9fb",
+      "bytes": 6410
+    },
+    "validation/parity/cross_format_dp.json": {
+      "sha256": "c7332df0f02244416b01e49ecb037c76d4bf2643a2b42505ff1dfe896879e533",
+      "bytes": 1550
+    },
+    "validation/runtime/fixture_verify_8bit.json": {
+      "sha256": "305cc657c5ebd539ce92e3fbe7db36c594e40a0566d185b36dc858e338025d01",
+      "bytes": 3078
+    },
+    "validation/runtime/fixture_verify_bf16.json": {
+      "sha256": "87766f14dfdf9ac1d6d97692e523e3942329b4431615a46322992d1dbf315c63",
+      "bytes": 3077
+    },
+    "validation/runtime/render_verify.json": {
+      "sha256": "5333dd7e4083de7e80bd98b4da4d0965c2835782a37e836d55d0e09a26c4f8c1",
+      "bytes": 1550
+    },
+    "validation/runtime/reverify_mlx.json": {
+      "sha256": "1f6f62c8b66e8df2f738acfd96667ef746f9504328e603bdeaa848156ddf22bf",
+      "bytes": 963
+    }
+  },
+  "readme_hashed": true,
+  "readme_placeholder": false,
+  "note": "manifest.json hashes every file of the repo except itself, README.md, figures/ and validation/ included. The runtime --verify check (jev_style_decision_mlx.py) checks the top-level files (except README.md) and the selected precision folder (bf16/ or 8bit/), so either folder may be downloaded alone; README.md, figures/ and validation/ are not checked."
+}

readout_config.json

@@ -0,0 +1,47 @@
+{
+  "format": "macjev-readout-v2",
+  "model_name": "Jev-Style-2B-Decision-v3",
+  "readout": "verdict",
+  "template": "macjev-render-v2-long-options",
+  "layout": "sb",
+  "block_size": 2048,
+  "total_context_limit": 25600,
+  "question_option_limit": null,
+  "slot_tokens": {
+    "yes": {
+      "text": " yes",
+      "id": 9542
+    },
+    "no": {
+      "text": " no",
+      "id": 874
+    },
+    "verdict_slot": {
+      "text": " ->",
+      "id": 1411
+    }
+  },
+  "attention": "block-causal in the 6 full-attention layers: the input is cut into [start, stop) blocks of at most block_size tokens (state blocks; one short question block, or catalogue blocks + rubric blocks when question + options + slots exceed block_size); every block attends to all earlier tokens and to itself with no causal mask inside the block. Gated-DeltaNet layers are ordinary recurrent layers.",
+  "score": "per option k: logit[' yes'] - logit[' no'] at the k-th ' ->' slot, computed as h_slot . (w_yes - w_no) from the final normed hidden state and the tied embedding rows (float32)",
+  "probabilities": "softmax(scores / T) in canonical option order; T = temperatures.global (one global temperature)",
+  "tokenizer_sha256": "06b9509352d2af50381ab2247e083b80d32d5c0aba91c272ca9ff729b6a0e523",
+  "budgets": {
+    "max_len": 25600,
+    "question_option_limit": null,
+    "note": "the complete input (state + question + options + readout) <= max_len tokens; there is no separate question/options cap; larger inputs raise InputBudgetError, nothing is truncated"
+  },
+  "temperatures": {
+    "version": "macjev-temperature-v2-global",
+    "global": 0.8278650620942867,
+    "groups": null,
+    "groups_note": "none: this model has one global temperature (no category / family / option-count temperatures)",
+    "fitted_on": "2,000 independent calibration rows (not dev, not test)",
+    "n_rows": 2000,
+    "fit_quality": {
+      "nll_before": 0.3002617012172898,
+      "nll_after": 0.2900580002426921,
+      "n": 2000
+    },
+    "selected_weights": "ema"
+  }
+}

release_config.json

@@ -0,0 +1,422 @@
+{
+  "format": "jev-style-release-v1",
+  "model_name": "Jev-Style-2B-Decision-v3",
+  "repo": "chaoliangUNSW/Jev-Style-2B-Decision-v3-MLX",
+  "generation": "v3 (third generation of the Jev-Style decision series)",
+  "lineage": {
+    "v1": "chaoliangUNSW/Jev-Style-Qwen3.5-2B-Decision",
+    "v1_public_gguf": "chaoliangUNSW/Jev-Style-Qwen3.5-2B-Decision-GGUF",
+    "v2": "chaoliangUNSW/Jev-Style-Qwen3.5-2B-Decision-v2",
+    "v3_0.8b": "chaoliangUNSW/Jev-Style-0.8B-Decision-v3",
+    "v3_2b": "chaoliangUNSW/Jev-Style-2B-Decision-v3"
+  },
+  "base_model": "Qwen/Qwen3.5-2B",
+  "base_model_revision": "15852e8c16360a2fea060d615a32b45270f8a8fc",
+  "base_model_relation": "finetune",
+  "architecture": "Qwen3_5ForCausalLM (text only, 24 layers: 18 Gated DeltaNet + 6 full attention, hidden 2048, tied embeddings)",
+  "readout": "verdict",
+  "template": "macjev-render-v2-long-options",
+  "layout": "sb",
+  "attention": "block-causal (2,048-token blocks) in the full-attention layers; see readout_config.json -> attention",
+  "readout_config": "readout_config.json",
+  "budgets": {
+    "max_len": 25600,
+    "block_size": 2048,
+    "question_option_limit": null
+  },
+  "source": {
+    "checkpoint_sha256": {
+      "model-00001-of-00002.safetensors": "df592f869cb1232226d31491b783bfa9406ffb3c795eae44a3b7452e1044a8fb",
+      "model-00002-of-00002.safetensors": "c6ff77a32a8cf36767e51c1282bb345a873db6509847f6c75baadd8fb17eeb70"
+    },
+    "selected_weights": "ema",
+    "mlx": "0.32.2",
+    "mlx_lm": "0.31.3",
+    "exports_manifest": {
+      "file": "runs/macjev/release_2b/exports_manifest.json",
+      "sha256": "11c549e479c7022c960c546730d6a3c42957c1c40b1b8901a183a893917d96a5",
+      "note": "sha256 / bytes of the exported GGUF and MLX files (release_2b/build_exports.py)"
+    },
+    "candidate_sha256sums": {
+      "file": "runs/macjev/candidate_2b/hf-candidate/SHA256SUMS.json",
+      "sha256": "7549ed3fe6239b2862998798d8da1457fdced25cd424aa4442557ae852d6e696"
+    }
+  },
+  "calibration": {
+    "version": "macjev-temperature-v2-global",
+    "global_T": 0.8278650620942867,
+    "groups": null,
+    "n_rows": 2000,
+    "fit_quality": {
+      "nll_before": 0.3002617012172898,
+      "nll_after": 0.2900580002426921,
+      "n": 2000
+    },
+    "fitted_on": "2,000 independent calibration rows (not dev, not test)"
+  },
+  "related_repos": {
+    "main": "chaoliangUNSW/Jev-Style-2B-Decision-v3",
+    "gguf": "chaoliangUNSW/Jev-Style-2B-Decision-v3-GGUF",
+    "mlx": "chaoliangUNSW/Jev-Style-2B-Decision-v3-MLX"
+  },
+  "tested_with": {
+    "python": "3.12",
+    "mlx": "0.32.2",
+    "mlx-lm": "0.31.3",
+    "tokenizers": "0.23.2",
+    "numpy": "2.5.3"
+  },
+  "weights": {
+    "bf16/model.safetensors": "bfloat16",
+    "8bit/model.safetensors": "mlx affine 8-bit, group 64",
+    "*/macjev_norms_fp32.safetensors": "exact FP32 (1 + w) of the 61 shifted Qwen3.5 RMSNorms (format macjev-norms-fp32-v1); required by the runtime"
+  },
+  "runtime": {
+    "script": "jev_style_decision_mlx.py",
+    "class": "JevStyleDecisionMLX",
+    "precision_argument": "--precision bf16|8bit (Python: precision=...)",
+    "default_precision": "bf16",
+    "mlx_cache_limit_gib": 2.0,
+    "mlx_lm_required": "0.31.3 (exact; the runtime checks the sha256 of the mlx-lm source it patches and refuses other versions)",
+    "mlx_lm_corrections": {
+      "gdn_qk_l2norm": "Gated-DeltaNet q/k normalised with l2norm eps 1e-6 (HF / fla / llama.cpp) instead of mlx-lm's rms_norm eps 1e-6 (= l2norm eps Dk*1e-6); per-instance class swap",
+      "rmsnorm_fp32_shift": "Qwen3.5 (1 + w) RMSNorm weights taken in FP32 from macjev_norms_fp32.safetensors instead of the bf16-rounded (1 + w) stored by mlx-lm's converter; norms computed in FP32 and cast back"
+    },
+    "state_reuse": "score_many(state, questions) computes the state blocks once and runs every question on a copy of that cache; the last state is kept, so consecutive calls / JSONL rows with an identical state reuse it",
+    "shared_files": "readout_config.json (readout + calibration temperature), release_config.json and manifest.json are shared by both precisions; each precision folder has its own config.json, tokenizer, weights and norm sidecar",
+    "precisions": {
+      "bf16": {
+        "folder": "bf16/",
+        "weights": "bfloat16",
+        "mlx_compute_dtype": null,
+        "mlx_compute_dtype_note": "native activations (bf16 residual stream; norms and readout in FP32), the setting the release scoring used"
+      },
+      "8bit": {
+        "folder": "8bit/",
+        "weights": "mlx affine 8-bit, group 64",
+        "mlx_compute_dtype": null,
+        "mlx_compute_dtype_note": "native activations, the setting the release scoring used"
+      }
+    },
+    "removed_vs_0.8b_runtime": {
+      "--category": "no group temperatures: one global temperature",
+      "--head-max": "no question/options cap: only the 25,600-token total budget; long question+options use the catalogue-overflow layout",
+      "option chunking": "not needed: all options are always scored in one input"
+    }
+  },
+  "format_gates": {
+    "predeclared": {
+      "file": "runs/macjev/runtime_v2_dev/PREDECLARED_RELEASE_GATES_2B.md",
+      "sha256": "783aec77300779582911cdc57c71cf8b101b0cd5ca040073f7e1fa3ac20af08e",
+      "gates": "F16/bf16 top-1 >= 0.99; 8-bit top-1 >= 0.98; |dNLL| <= 0.02; accuracy drop Q8_0 / MLX-8bit <= 0.3 pp, Q4_K_M <= 1.0 pp (4-bit top-1 / NLL report-only); no non-finite scores"
+    },
+    "reference": {
+      "what": "HF transformers FP32 on CPU, exact v2 block attention, released bf16 checkpoint (runs/macjev/candidate_2b/hf-candidate)",
+      "temperature": 0.8278650620942867
+    },
+    "fixtures": {
+      "gate": {
+        "what": "1,000 real dev rows (<= 4,096 tokens); decides the release gates",
+        "file": "runs/macjev/runtime_v2_dev/reference/gate_fixture_n1000.jsonl",
+        "sha256": "1c03b8c570b988b0c83ca1f79e7fd37e6196239a0836f348300c3f23ae667b68"
+      },
+      "long": {
+        "what": "35 requests / 43 questions up to 25,600 tokens incl. catalogue overflow and K <= 151; accuracy gate unresolvable on 43 questions (verdict INCONCLUSIVE means only that), top-1 / dNLL reported",
+        "file": "runs/macjev/runtime_v2_dev/reference_base/fixture.jsonl",
+        "sha256": "2b16d25eafa70c00a1106ba51b94890821b51ecb4fa393abb12403c8739ef314"
+      }
+    },
+    "cross_format_dp": {
+      "file": "runs/macjev/release_2b/parity/cross_format_dp.json",
+      "sha256": "c7332df0f02244416b01e49ecb037c76d4bf2643a2b42505ff1dfe896879e533"
+    },
+    "formats": {
+      "mlx-bf16": {
+        "gate_fixture": {
+          "n": 1000,
+          "top1_agreement": 0.997,
+          "top1_flips": 3,
+          "max_abs_dscore": 0.13961565494537354,
+          "mean_abs_dscore": 0.012822698334419146,
+          "dnll": 0.0001265220422789204,
+          "nll_ref": 0.4436679457518203,
+          "nll_cand": 0.4437944677940992,
+          "acc_ref": 0.808,
+          "acc_cand": 0.805,
+          "acc_drop_pp": 0.30000000000000027,
+          "max_abs_dp": 0.03549758339084519,
+          "mean_max_abs_dp": 0.0025865935031305484,
+          "gates": {
+            "coverage_finite_tokens": {
+              "pass": true,
+              "problems": {},
+              "reference_missing_requests": 0
+            },
+            "top1": {
+              "threshold": 0.99,
+              "value": 0.997,
+              "pass": true
+            },
+            "dnll": {
+              "threshold": 0.02,
+              "value": 0.0001265220422789204,
+              "pass": true
+            }
+          },
+          "verdict": "PASS",
+          "note": "small fixtures cannot resolve 0.3pp accuracy gates; accuracy on a parity fixture is only indicative (synthetic long rows carry the base row's gold label).",
+          "source": {
+            "file": "runs/macjev/release_2b/parity/compare_mlx_bf16_gate_vs_block.json",
+            "sha256": "7fa30fa414a4ffc12a7aa6371e77579845c3b816b90da3a813a70fec10ccd9fb"
+          }
+        },
+        "long_fixture": {
+          "n": 43,
+          "top1_agreement": 1.0,
+          "top1_flips": 0,
+          "max_abs_dscore": 0.08836114406585693,
+          "mean_abs_dscore": 0.01365492056503762,
+          "dnll": -0.0026943261000204055,
+          "nll_ref": 0.670433307194272,
+          "nll_cand": 0.6677389810942516,
+          "acc_ref": 0.7209302325581395,
+          "acc_cand": 0.7209302325581395,
+          "acc_drop_pp": 0.0,
+          "max_abs_dp": 0.010765431767972844,
+          "mean_max_abs_dp": 0.003890125022245109,
+          "gates": {
+            "coverage_finite_tokens": {
+              "pass": true,
+              "problems": {},
+              "reference_missing_requests": 0
+            },
+            "top1": {
+              "threshold": 0.99,
+              "value": 1.0,
+              "pass": true
+            },
+            "dnll": {
+              "threshold": 0.02,
+              "value": -0.0026943261000204055,
+              "pass": true
+            }
+          },
+          "verdict": "PASS",
+          "note": "small fixtures cannot resolve 0.3pp accuracy gates; accuracy on a parity fixture is only indicative (synthetic long rows carry the base row's gold label).",
+          "source": {
+            "file": "runs/macjev/release_2b/parity/compare_mlx_bf16_base_vs_block.json",
+            "sha256": "ca16e5281debaf1bc601b4a8c93cf6c1106f7b032f35c5833baa4fc8178def7c"
+          }
+        },
+        "release_verdict": "PASS"
+      },
+      "mlx-8bit": {
+        "gate_fixture": {
+          "n": 1000,
+          "top1_agreement": 0.996,
+          "top1_flips": 4,
+          "max_abs_dscore": 0.3184394836425781,
+          "mean_abs_dscore": 0.016680597170951345,
+          "dnll": 0.0005035682350758575,
+          "nll_ref": 0.4436679457518203,
+          "nll_cand": 0.44417151398689614,
+          "acc_ref": 0.808,
+          "acc_cand": 0.808,
+          "acc_drop_pp": 0.0,
+          "max_abs_dp": 0.1623697296878569,
+          "mean_max_abs_dp": 0.0039011049015487734,
+          "gates": {
+            "coverage_finite_tokens": {
+              "pass": true,
+              "problems": {},
+              "reference_missing_requests": 0
+            },
+            "top1": {
+              "threshold": 0.98,
+              "value": 0.996,
+              "pass": true
+            },
+            "dnll": {
+              "threshold": 0.02,
+              "value": 0.0005035682350758575,
+              "pass": true
+            },
+            "acc_drop_pp": {
+              "threshold": 0.3,
+              "value": 0.0,
+              "resolution_pp": 0.1,
+              "pass": true,
+              "note": null
+            }
+          },
+          "verdict": "PASS",
+          "note": "small fixtures cannot resolve 0.3pp accuracy gates; accuracy on a parity fixture is only indicative (synthetic long rows carry the base row's gold label).",
+          "source": {
+            "file": "runs/macjev/release_2b/parity/compare_mlx_affine8-g64_gate_vs_block.json",
+            "sha256": "f0e2f44498df00945901c2a9bd5bff6f742d11ab243495c8089eaf0448fda782"
+          }
+        },
+        "long_fixture": {
+          "n": 43,
+          "top1_agreement": 1.0,
+          "top1_flips": 0,
+          "max_abs_dscore": 0.4653351306915283,
+          "mean_abs_dscore": 0.023077200647273945,
+          "dnll": 0.01638063705636561,
+          "nll_ref": 0.670433307194272,
+          "nll_cand": 0.6868139442506376,
+          "acc_ref": 0.7209302325581395,
+          "acc_cand": 0.7209302325581395,
+          "acc_drop_pp": 0.0,
+          "max_abs_dp": 0.04628699142360499,
+          "mean_max_abs_dp": 0.007026009064181663,
+          "gates": {
+            "coverage_finite_tokens": {
+              "pass": true,
+              "problems": {},
+              "reference_missing_requests": 0
+            },
+            "top1": {
+              "threshold": 0.98,
+              "value": 1.0,
+              "pass": true
+            },
+            "dnll": {
+              "threshold": 0.02,
+              "value": 0.01638063705636561,
+              "pass": true
+            },
+            "acc_drop_pp": {
+              "threshold": 0.3,
+              "value": 0.0,
+              "resolution_pp": 2.3255813953488373,
+              "pass": null,
+              "note": "UNRESOLVED: 43 gold questions -> one flip = 2.33pp > 0.3pp; this fixture cannot resolve the gate (use the gate fixture with >= 334 rows; statistical power needs far more)"
+            }
+          },
+          "verdict": "INCONCLUSIVE",
+          "note": "small fixtures cannot resolve 0.3pp accuracy gates; accuracy on a parity fixture is only indicative (synthetic long rows carry the base row's gold label).",
+          "source": {
+            "file": "runs/macjev/release_2b/parity/compare_mlx_affine8-g64_base_vs_block.json",
+            "sha256": "e1911f21ce73dd42d6c57e8349e2ec3221386a2b44ca461a3e50b9a69057ec34"
+          }
+        },
+        "release_verdict": "PASS"
+      }
+    }
+  },
+  "decision_index": "requested from the maintainer after release (not run by us)",
+  "runtime_parity": {
+    "protocol": "the staged runtime scores the long fixture from text/ids through its own renderer; raw scores compared with the development MLX engine that produced the release parity (release_2b/parity/pred_mlx_*_base.jsonl)",
+    "fixture": {
+      "file": "runs/macjev/runtime_v2_dev/reference_base/fixture.jsonl",
+      "sha256": "2b16d25eafa70c00a1106ba51b94890821b51ecb4fa393abb12403c8739ef314"
+    },
+    "bf16": {
+      "precision": "bf16",
+      "dev_seconds": 415.6,
+      "default_T": 0.8278650620942867,
+      "fixture_requests": 35,
+      "fixture_questions": 43,
+      "fixture_tokens": 214700,
+      "overflow_q": 5,
+      "max_abs_dscore": 0.0,
+      "max_abs_dprob_vs_softmax_dev_over_T": 0.0,
+      "text_e2e_max_abs_dscore_vs_dev": 0.0,
+      "text_ids_equal_dev": true,
+      "score_many_vs_separate_fresh_decide_max_abs": 0.0,
+      "order_invariance_max_abs": 0.0,
+      "runtime_seconds": 293.1,
+      "source": {
+        "file": "runs/macjev/runtime_v2_dev/verify_release_mlx/fixture_verify_bf16.json",
+        "sha256": "87766f14dfdf9ac1d6d97692e523e3942329b4431615a46322992d1dbf315c63"
+      }
+    },
+    "8bit": {
+      "precision": "8bit",
+      "dev_seconds": 1324.8,
+      "default_T": 0.8278650620942867,
+      "fixture_requests": 35,
+      "fixture_questions": 43,
+      "fixture_tokens": 214700,
+      "overflow_q": 5,
+      "max_abs_dscore": 0.0,
+      "max_abs_dprob_vs_softmax_dev_over_T": 0.0,
+      "text_e2e_max_abs_dscore_vs_dev": 0.0,
+      "text_ids_equal_dev": true,
+      "score_many_vs_separate_fresh_decide_max_abs": 0.0,
+      "order_invariance_max_abs": 0.0,
+      "runtime_seconds": 286.5,
+      "source": {
+        "file": "runs/macjev/runtime_v2_dev/verify_release_mlx/fixture_verify_8bit.json",
+        "sha256": "305cc657c5ebd539ce92e3fbe7db36c594e40a0566d185b36dc858e338025d01"
+      }
+    },
+    "render_token_identity": {
+      "requests": 463,
+      "questions": 756,
+      "ok_identical": 743,
+      "budget_both": 6,
+      "error_both": 6,
+      "mismatch_count": 1,
+      "source": {
+        "file": "runs/macjev/runtime_v2_dev/verify_release_mlx/render_verify.json",
+        "sha256": "5333dd7e4083de7e80bd98b4da4d0965c2835782a37e836d55d0e09a26c4f8c1"
+      }
+    },
+    "runtime_file": {
+      "file": "jev_style_decision_mlx.py",
+      "sha256_now": "025daab9b374d740e743c519bd3aa3385791e08c81f4515b2ae525af57788030",
+      "mtime_unix": 1790448336.812169
+    },
+    "original_records_predate_last_runtime_edit": true,
+    "reverified_on_current_runtime": true,
+    "recorded_before_last_runtime_edit": false,
+    "note": "the records above were written before the last edit of the runtime file; the runtime as staged now (sha256 runtime_file.sha256_now) was re-verified on the long fixture, see 'reverified' (reverified.runtime_sha256 == runtime_file.sha256_now).",
+    "reverified": {
+      "runtime_file": "jev_style_decision_mlx.py",
+      "runtime_sha256": "025daab9b374d740e743c519bd3aa3385791e08c81f4515b2ae525af57788030",
+      "shared_core_sha256": "56deae095206fc59b9a30d8626ac5d27cacddcd94d7c7f08f6f4bc53d326a002",
+      "what": "MLX bf16 (loaded from the repository root, precision bf16), runtime defaults",
+      "loaded_from": "runs/macjev/hf_staging/Jev-Style-2B-Decision-v3-MLX",
+      "verify_manifest": true,
+      "compared_with": {
+        "file": "runs/macjev/release_2b/parity/pred_mlx_bf16_base.jsonl",
+        "sha256": "c42232870ab8c5410b4f45dc85729249d467b9c0bd5d24213d085e4479a3a60e"
+      },
+      "fixture": {
+        "file": "runs/macjev/runtime_v2_dev/reference_base/fixture.jsonl",
+        "sha256": "2b16d25eafa70c00a1106ba51b94890821b51ecb4fa393abb12403c8739ef314"
+      },
+      "questions": 43,
+      "skipped_questions": 0,
+      "max_abs_score_diff": 0.0,
+      "top1_same": "43/43",
+      "token_or_overflow_mismatches": [],
+      "seconds": 137.2,
+      "finished_unix": 1790449492.3913028,
+      "source": {
+        "file": "runs/macjev/hf_staging/_2b_tools/reverify_mlx.json",
+        "sha256": "1f6f62c8b66e8df2f738acfd96667ef746f9504328e603bdeaa848156ddf22bf"
+      }
+    }
+  },
+  "latency": {
+    "file": "validation/latency_2b.json",
+    "sha256": "458d00e0c81b3165072b8547278e6039b9ae0d45d54b7421008ef24c1d24ca3d",
+    "machine": {
+      "chip": "Apple M1 Max",
+      "memory_bytes": 68719476736,
+      "macos": "15.7.5",
+      "python": "3.12.13"
+    },
+    "backends_shown_on_card": [
+      "mlx-bf16",
+      "mlx-8bit"
+    ],
+    "rows": 12,
+    "rows_in_file": 36,
+    "card_generator": "runs/macjev/release_2b/cards/_build/latency_tables.py"
+  },
+  "root_config_json": "copy of bf16/config.json at the repository root so the Hub counts downloads; the runtime treats the root as the repository folder (it has bf16/ and 8bit/ but no model.safetensors) and loads bf16/ or 8bit/"
+}

requirements.txt

@@ -0,0 +1,8 @@
+# tested with: python 3.12, mlx 0.32.2, mlx-lm 0.31.3, transformers 5.17.0, tokenizers 0.23.2, numpy 2.5.3 (Apple silicon, macOS 15)
+# mlx-lm is pinned: the runtime corrects two Qwen3.5 numerics inside mlx-lm (Gated-DeltaNet q/k l2norm eps,
+# FP32 (1 + w) RMSNorm weights) and checks the sha256 of the mlx-lm source it patches; other versions are refused.
+mlx==0.32.2
+mlx-lm==0.31.3
+transformers==5.17.0  # imported by mlx-lm when the model loads
+tokenizers==0.23.2
+numpy==2.5.3

validation/SOURCES.json

@@ -0,0 +1,60 @@
+{
+  "note": "copies of the verification / benchmark records behind release_config.json and the model card; local absolute paths replaced by repository-relative ones (local_paths_scrubbed); source_sha256 = the unmodified file (equal to the published copy when local_paths_scrubbed is false)",
+  "files": {
+    "parity/compare_mlx_bf16_gate_vs_block.json": {
+      "source": "runs/macjev/release_2b/parity/compare_mlx_bf16_gate_vs_block.json",
+      "source_sha256": "7fa30fa414a4ffc12a7aa6371e77579845c3b816b90da3a813a70fec10ccd9fb",
+      "local_paths_scrubbed": false
+    },
+    "parity/compare_mlx_bf16_base_vs_block.json": {
+      "source": "runs/macjev/release_2b/parity/compare_mlx_bf16_base_vs_block.json",
+      "source_sha256": "ca16e5281debaf1bc601b4a8c93cf6c1106f7b032f35c5833baa4fc8178def7c",
+      "local_paths_scrubbed": false
+    },
+    "parity/compare_mlx_affine8-g64_gate_vs_block.json": {
+      "source": "runs/macjev/release_2b/parity/compare_mlx_affine8-g64_gate_vs_block.json",
+      "source_sha256": "f0e2f44498df00945901c2a9bd5bff6f742d11ab243495c8089eaf0448fda782",
+      "local_paths_scrubbed": false
+    },
+    "parity/compare_mlx_affine8-g64_base_vs_block.json": {
+      "source": "runs/macjev/release_2b/parity/compare_mlx_affine8-g64_base_vs_block.json",
+      "source_sha256": "e1911f21ce73dd42d6c57e8349e2ec3221386a2b44ca461a3e50b9a69057ec34",
+      "local_paths_scrubbed": false
+    },
+    "parity/cross_format_dp.json": {
+      "source": "runs/macjev/release_2b/parity/cross_format_dp.json",
+      "source_sha256": "c7332df0f02244416b01e49ecb037c76d4bf2643a2b42505ff1dfe896879e533",
+      "local_paths_scrubbed": false
+    },
+    "parity/PREDECLARED_RELEASE_GATES_2B.md": {
+      "source": "runs/macjev/runtime_v2_dev/PREDECLARED_RELEASE_GATES_2B.md",
+      "source_sha256": "783aec77300779582911cdc57c71cf8b101b0cd5ca040073f7e1fa3ac20af08e",
+      "local_paths_scrubbed": false
+    },
+    "runtime/fixture_verify_bf16.json": {
+      "source": "runs/macjev/runtime_v2_dev/verify_release_mlx/fixture_verify_bf16.json",
+      "source_sha256": "87766f14dfdf9ac1d6d97692e523e3942329b4431615a46322992d1dbf315c63",
+      "local_paths_scrubbed": false
+    },
+    "runtime/fixture_verify_8bit.json": {
+      "source": "runs/macjev/runtime_v2_dev/verify_release_mlx/fixture_verify_8bit.json",
+      "source_sha256": "305cc657c5ebd539ce92e3fbe7db36c594e40a0566d185b36dc858e338025d01",
+      "local_paths_scrubbed": false
+    },
+    "runtime/render_verify.json": {
+      "source": "runs/macjev/runtime_v2_dev/verify_release_mlx/render_verify.json",
+      "source_sha256": "5333dd7e4083de7e80bd98b4da4d0965c2835782a37e836d55d0e09a26c4f8c1",
+      "local_paths_scrubbed": false
+    },
+    "runtime/reverify_mlx.json": {
+      "source": "runs/macjev/hf_staging/_2b_tools/reverify_mlx.json",
+      "source_sha256": "1f6f62c8b66e8df2f738acfd96667ef746f9504328e603bdeaa848156ddf22bf",
+      "local_paths_scrubbed": false
+    },
+    "latency_2b.json": {
+      "source": "runs/macjev/release_2b/latency_2b.json",
+      "source_sha256": "458d00e0c81b3165072b8547278e6039b9ae0d45d54b7421008ef24c1d24ca3d",
+      "local_paths_scrubbed": false
+    }
+  }
+}

validation/latency_2b.json

@@ -0,0 +1,1568 @@
+{
+  "format": "jev-style-latency-v1",
+  "model": "Jev-Style-2B-Decision-v3",
+  "machine": {
+    "chip": "Apple M1 Max",
+    "memory_bytes": 68719476736,
+    "macos": "15.7.5",
+    "python": "3.12.13"
+  },
+  "shared_machine_note": "measured while other agents' jobs ran on the same Mac (among them a CPU-heavy PyTorch parity job using ~3.5 cores and ~16 GB); load averages at the end of each run are recorded per row. Treat the numbers as indicative, not as a clean benchmark.",
+  "protocol": {
+    "runtimes": "the staged runtimes of the three repos (jev_style_decision_gguf.py + jev-score-v2 built with build_jev_score.sh against llama.cpp 441df11f, Metal, all layers on the GPU; jev_style_decision_mlx.py, mlx 0.32.2 / mlx-lm 0.31.3; jev_style_decision.py on MPS, float32)",
+    "weights": "trained release weights: release_2b/gguf/model-*.gguf (tensor data identical to the named repo files), release_2b/mlx/{bf16,affine8-g64}, candidate_2b/hf-candidate (torch)",
+    "state": "plain text (repository documentation + source code, English) cut to target-150 tokens; the total input per question is recorded in input_tokens_per_question",
+    "questions": "n_questions=1: one 4-option choice question (decide); n_questions=10: 10 mixed questions (4 choice, 4 true/false, 2 score) about the same state in one score_many call",
+    "one_process_per_row": true,
+    "cold_s": "first scoring call after loading (state + questions; includes GPU warm-up)",
+    "warm_median_s": "median of 3 further calls, cached state dropped before each (state recomputed)",
+    "state_cached_median_s": "median of 3 calls with the state already computed (only the question blocks run)",
+    "load_s": "runtime construction (weights from the OS file cache in most rows)",
+    "wall_time": "time.perf_counter around decide()/score_many(), incl. tokenisation and rendering",
+    "peak_rss_bytes": "ru_maxrss of the Python process (and of the jev-score-v2 child for GGUF); includes memory-mapped weight pages",
+    "peak_phys_footprint_bytes": "macOS lifetime-max physical footprint (proc_pid_rusage v4): memory the process owns, incl. Metal / MLX buffers it allocates. It does NOT count clean memory-mapped file pages, so for GGUF (jev-score-v2 maps the .gguf file) it excludes the weights and is not a memory requirement; use peak_rss_bytes.jev_score_v2 (which includes the mapped weight pages) as the upper bound for GGUF"
+  },
+  "reruns": {
+    "gguf_and_mlx_rows": "all 18 GGUF rows and all 12 MLX rows were re-measured on 2026-09-27 03:11-03:28 (fix round, same script, same state text, same answers) because an independent re-run of the first session's GGUF Q8_0 / Q4_K_M 10-question rows was ~2.4x faster (contention during the first session). The first-session rows are kept in latency/runs_superseded_2026-09-27/ and are not used here. The 6 torch MPS rows are from the first session (not re-measured).",
+    "superseded_dir": "latency/runs_superseded_2026-09-27/"
+  },
+  "complete": true,
+  "missing_runs": [],
+  "rows": [
+    {
+      "backend": "gguf-f16",
+      "n_questions": 1,
+      "state_tokens": 878,
+      "input_tokens_per_question": [
+        943
+      ],
+      "load_s": 2.6054,
+      "cold_s": 0.5624,
+      "warm_median_s": 0.5242,
+      "warm_s": [
+        0.5203,
+        0.5345,
+        0.5242
+      ],
+      "state_cached_median_s": 0.0653,
+      "peak_rss_bytes": {
+        "python": 352616448,
+        "jev_score_v2": 4451008512
+      },
+      "peak_phys_footprint_bytes": {
+        "python": 252413696,
+        "jev_score_v2": 660384576
+      },
+      "results_identical_cold_vs_state_cached": true,
+      "loadavg_at_end": [
+        9.59,
+        9.9,
+        10.98
+      ],
+      "measured_unix": 1790442718.8146281,
+      "settings": {
+        "engine": "jev-score-v2 (staged jev_score_v2.cpp, llama.cpp 441df11f, Metal)",
+        "n_gpu_layers": 999,
+        "flash_attn": "default (llama.cpp auto on Metal)",
+        "gguf": "release_2b/gguf/model-f16.gguf"
+      },
+      "raw": "latency/runs/gguf-f16_1024_1q.json"
+    },
+    {
+      "backend": "gguf-f16",
+      "n_questions": 10,
+      "state_tokens": 878,
+      "input_tokens_per_question": [
+        943,
+        918,
+        925,
+        911,
+        918,
+        921,
+        943,
+        917,
+        912,
+        919
+      ],
+      "load_s": 1.1941,
+      "cold_s": 0.994,
+      "warm_median_s": 0.968,
+      "warm_s": [
+        0.968,
+        0.9464,
+        0.9713
+      ],
+      "state_cached_median_s": 0.4965,
+      "peak_rss_bytes": {
+        "python": 328564736,
+        "jev_score_v2": 4516610048
+      },
+      "peak_phys_footprint_bytes": {
+        "python": 254527296,
+        "jev_score_v2": 676899648
+      },
+      "results_identical_cold_vs_state_cached": true,
+      "loadavg_at_end": [
+        10.0,
+        9.97,
+        10.99
+      ],
+      "measured_unix": 1790442726.19508,
+      "settings": {
+        "engine": "jev-score-v2 (staged jev_score_v2.cpp, llama.cpp 441df11f, Metal)",
+        "n_gpu_layers": 999,
+        "flash_attn": "default (llama.cpp auto on Metal)",
+        "gguf": "release_2b/gguf/model-f16.gguf"
+      },
+      "raw": "latency/runs/gguf-f16_1024_10q.json"
+    },
+    {
+      "backend": "gguf-f16",
+      "n_questions": 1,
+      "state_tokens": 3950,
+      "input_tokens_per_question": [
+        4015
+      ],
+      "load_s": 1.1659,
+      "cold_s": 2.2122,
+      "warm_median_s": 2.1753,
+      "warm_s": [
+        2.1909,
+        2.1753,
+        2.1733
+      ],
+      "state_cached_median_s": 0.078,
+      "peak_rss_bytes": {
+        "python": 354697216,
+        "jev_score_v2": 4568334336
+      },
+      "peak_phys_footprint_bytes": {
+        "python": 257705600,
+        "jev_score_v2": 722676736
+      },
+      "results_identical_cold_vs_state_cached": true,
+      "loadavg_at_end": [
+        9.01,
+        9.76,
+        10.9
+      ],
+      "measured_unix": 1790442737.183161,
+      "settings": {
+        "engine": "jev-score-v2 (staged jev_score_v2.cpp, llama.cpp 441df11f, Metal)",
+        "n_gpu_layers": 999,
+        "flash_attn": "default (llama.cpp auto on Metal)",
+        "gguf": "release_2b/gguf/model-f16.gguf"
+      },
+      "raw": "latency/runs/gguf-f16_4096_1q.json"
+    },
+    {
+      "backend": "gguf-f16",
+      "n_questions": 10,
+      "state_tokens": 3950,
+      "input_tokens_per_question": [
+        4015,
+        3990,
+        3997,
+        3983,
+        3990,
+        3993,
+        4015,
+        3989,
+        3984,
+        3991
+      ],
+      "load_s": 1.2112,
+      "cold_s": 2.6904,
+      "warm_median_s": 2.6583,
+      "warm_s": [
+        2.6583,
+        2.6665,
+        2.6378
+      ],
+      "state_cached_median_s": 0.5467,
+      "peak_rss_bytes": {
+        "python": 331694080,
+        "jev_score_v2": 4564172800
+      },
+      "peak_phys_footprint_bytes": {
+        "python": 248334016,
+        "jev_score_v2": 720382976
+      },
+      "results_identical_cold_vs_state_cached": true,
+      "loadavg_at_end": [
+        8.62,
+        9.62,
+        10.83
+      ],
+      "measured_unix": 1790442751.5065348,
+      "settings": {
+        "engine": "jev-score-v2 (staged jev_score_v2.cpp, llama.cpp 441df11f, Metal)",
+        "n_gpu_layers": 999,
+        "flash_attn": "default (llama.cpp auto on Metal)",
+        "gguf": "release_2b/gguf/model-f16.gguf"
+      },
+      "raw": "latency/runs/gguf-f16_4096_10q.json"
+    },
+    {
+      "backend": "gguf-f16",
+      "n_questions": 1,
+      "state_tokens": 24436,
+      "input_tokens_per_question": [
+        24501
+      ],
+      "load_s": 1.1784,
+      "cold_s": 16.2299,
+      "warm_median_s": 16.1651,
+      "warm_s": [
+        16.1645,
+        16.1651,
+        16.2375
+      ],
+      "state_cached_median_s": 0.1677,
+      "peak_rss_bytes": {
+        "python": 349683712,
+        "jev_score_v2": 4734812160
+      },
+      "peak_phys_footprint_bytes": {
+        "python": 239191744,
+        "jev_score_v2": 882732352
+      },
+      "results_identical_cold_vs_state_cached": true,
+      "loadavg_at_end": [
+        9.74,
+        9.66,
+        10.75
+      ],
+      "measured_unix": 1790442818.798799,
+      "settings": {
+        "engine": "jev-score-v2 (staged jev_score_v2.cpp, llama.cpp 441df11f, Metal)",
+        "n_gpu_layers": 999,
+        "flash_attn": "default (llama.cpp auto on Metal)",
+        "gguf": "release_2b/gguf/model-f16.gguf"
+      },
+      "raw": "latency/runs/gguf-f16_24576_1q.json"
+    },
+    {
+      "backend": "gguf-f16",
+      "n_questions": 10,
+      "state_tokens": 24436,
+      "input_tokens_per_question": [
+        24501,
+        24476,
+        24483,
+        24469,
+        24476,
+        24479,
+        24501,
+        24475,
+        24470,
+        24477
+      ],
+      "load_s": 1.225,
+      "cold_s": 16.9884,
+      "warm_median_s": 16.8115,
+      "warm_s": [
+        16.8115,
+        16.675,
+        16.8156
+      ],
+      "state_cached_median_s": 0.8643,
+      "peak_rss_bytes": {
+        "python": 367919104,
+        "jev_score_v2": 4738170880
+      },
+      "peak_phys_footprint_bytes": {
+        "python": 242992832,
+        "jev_score_v2": 887975168
+      },
+      "results_identical_cold_vs_state_cached": true,
+      "loadavg_at_end": [
+        8.02,
+        9.14,
+        10.46
+      ],
+      "measured_unix": 1790442890.769493,
+      "settings": {
+        "engine": "jev-score-v2 (staged jev_score_v2.cpp, llama.cpp 441df11f, Metal)",
+        "n_gpu_layers": 999,
+        "flash_attn": "default (llama.cpp auto on Metal)",
+        "gguf": "release_2b/gguf/model-f16.gguf"
+      },
+      "raw": "latency/runs/gguf-f16_24576_10q.json"
+    },
+    {
+      "backend": "gguf-q8_0",
+      "n_questions": 1,
+      "state_tokens": 878,
+      "input_tokens_per_question": [
+        943
+      ],
+      "load_s": 1.7934,
+      "cold_s": 0.6136,
+      "warm_median_s": 0.5663,
+      "warm_s": [
+        0.5678,
+        0.5663,
+        0.5663
+      ],
+      "state_cached_median_s": 0.0682,
+      "peak_rss_bytes": {
+        "python": 328974336,
+        "jev_score_v2": 2721579008
+      },
+      "peak_phys_footprint_bytes": {
+        "python": 248874624,
+        "jev_score_v2": 668163520
+      },
+      "results_identical_cold_vs_state_cached": true,
+      "loadavg_at_end": [
+        7.69,
+        9.05,
+        10.42
+      ],
+      "measured_unix": 1790442895.894016,
+      "settings": {
+        "engine": "jev-score-v2 (staged jev_score_v2.cpp, llama.cpp 441df11f, Metal)",
+        "n_gpu_layers": 999,
+        "flash_attn": "default (llama.cpp auto on Metal)",
+        "gguf": "release_2b/gguf/model-q8_0.gguf"
+      },
+      "raw": "latency/runs/gguf-q8_0_1024_1q.json"
+    },
+    {
+      "backend": "gguf-q8_0",
+      "n_questions": 10,
+      "state_tokens": 878,
+      "input_tokens_per_question": [
+        943,
+        918,
+        925,
+        911,
+        918,
+        921,
+        943,
+        917,
+        912,
+        919
+      ],
+      "load_s": 1.0682,
+      "cold_s": 1.073,
+      "warm_median_s": 1.0243,
+      "warm_s": [
+        1.0244,
+        1.0243,
+        1.0231
+      ],
+      "state_cached_median_s": 0.5268,
+      "peak_rss_bytes": {
+        "python": 347635712,
+        "jev_score_v2": 2757804032
+      },
+      "peak_phys_footprint_bytes": {
+        "python": 258377344,
+        "jev_score_v2": 674569792
+      },
+      "results_identical_cold_vs_state_cached": true,
+      "loadavg_at_end": [
+        7.24,
+        8.94,
+        10.37
+      ],
+      "measured_unix": 1790442903.5177228,
+      "settings": {
+        "engine": "jev-score-v2 (staged jev_score_v2.cpp, llama.cpp 441df11f, Metal)",
+        "n_gpu_layers": 999,
+        "flash_attn": "default (llama.cpp auto on Metal)",
+        "gguf": "release_2b/gguf/model-q8_0.gguf"
+      },
+      "raw": "latency/runs/gguf-q8_0_1024_10q.json"
+    },
+    {
+      "backend": "gguf-q8_0",
+      "n_questions": 1,
+      "state_tokens": 3950,
+      "input_tokens_per_question": [
+        4015
+      ],
+      "load_s": 1.0689,
+      "cold_s": 2.3987,
+      "warm_median_s": 2.3345,
+      "warm_s": [
+        2.3333,
+        2.3488,
+        2.3345
+      ],
+      "state_cached_median_s": 0.082,
+      "peak_rss_bytes": {
+        "python": 354189312,
+        "jev_score_v2": 2807808000
+      },
+      "peak_phys_footprint_bytes": {
+        "python": 247596736,
+        "jev_score_v2": 719183680
+      },
+      "results_identical_cold_vs_state_cached": true,
+      "loadavg_at_end": [
+        6.58,
+        8.74,
+        10.28
+      ],
+      "measured_unix": 1790442915.085064,
+      "settings": {
+        "engine": "jev-score-v2 (staged jev_score_v2.cpp, llama.cpp 441df11f, Metal)",
+        "n_gpu_layers": 999,
+        "flash_attn": "default (llama.cpp auto on Metal)",
+        "gguf": "release_2b/gguf/model-q8_0.gguf"
+      },
+      "raw": "latency/runs/gguf-q8_0_4096_1q.json"
+    },
+    {
+      "backend": "gguf-q8_0",
+      "n_questions": 10,
+      "state_tokens": 3950,
+      "input_tokens_per_question": [
+        4015,
+        3990,
+        3997,
+        3983,
+        3990,
+        3993,
+        4015,
+        3989,
+        3984,
+        3991
+      ],
+      "load_s": 1.0684,
+      "cold_s": 2.8854,
+      "warm_median_s": 2.8178,
+      "warm_s": [
+        2.8178,
+        2.8439,
+        2.816
+      ],
+      "state_cached_median_s": 0.5719,
+      "peak_rss_bytes": {
+        "python": 319045632,
+        "jev_score_v2": 2799091712
+      },
+      "peak_phys_footprint_bytes": {
+        "python": 243910336,
+        "jev_score_v2": 716169152
+      },
+      "results_identical_cold_vs_state_cached": true,
+      "loadavg_at_end": [
+        7.36,
+        8.8,
+        10.28
+      ],
+      "measured_unix": 1790442930.063521,
+      "settings": {
+        "engine": "jev-score-v2 (staged jev_score_v2.cpp, llama.cpp 441df11f, Metal)",
+        "n_gpu_layers": 999,
+        "flash_attn": "default (llama.cpp auto on Metal)",
+        "gguf": "release_2b/gguf/model-q8_0.gguf"
+      },
+      "raw": "latency/runs/gguf-q8_0_4096_10q.json"
+    },
+    {
+      "backend": "gguf-q8_0",
+      "n_questions": 1,
+      "state_tokens": 24436,
+      "input_tokens_per_question": [
+        24501
+      ],
+      "load_s": 1.0589,
+      "cold_s": 17.0922,
+      "warm_median_s": 17.0489,
+      "warm_s": [
+        17.0489,
+        17.0376,
+        17.0539
+      ],
+      "state_cached_median_s": 0.1679,
+      "peak_rss_bytes": {
+        "python": 364937216,
+        "jev_score_v2": 2966470656
+      },
+      "peak_phys_footprint_bytes": {
+        "python": 267110080,
+        "jev_score_v2": 875831168
+      },
+      "results_identical_cold_vs_state_cached": true,
+      "loadavg_at_end": [
+        6.7,
+        8.25,
+        9.94
+      ],
+      "measured_unix": 1790443000.691691,
+      "settings": {
+        "engine": "jev-score-v2 (staged jev_score_v2.cpp, llama.cpp 441df11f, Metal)",
+        "n_gpu_layers": 999,
+        "flash_attn": "default (llama.cpp auto on Metal)",
+        "gguf": "release_2b/gguf/model-q8_0.gguf"
+      },
+      "raw": "latency/runs/gguf-q8_0_24576_1q.json"
+    },
+    {
+      "backend": "gguf-q8_0",
+      "n_questions": 10,
+      "state_tokens": 24436,
+      "input_tokens_per_question": [
+        24501,
+        24476,
+        24483,
+        24469,
+        24476,
+        24479,
+        24501,
+        24475,
+        24470,
+        24477
+      ],
+      "load_s": 1.0501,
+      "cold_s": 17.8244,
+      "warm_median_s": 17.7424,
+      "warm_s": [
+        17.7424,
+        17.7312,
+        17.7793
+      ],
+      "state_cached_median_s": 0.8902,
+      "peak_rss_bytes": {
+        "python": 348651520,
+        "jev_score_v2": 2967109632
+      },
+      "peak_phys_footprint_bytes": {
+        "python": 260425536,
+        "jev_score_v2": 878698496
+      },
+      "results_identical_cold_vs_state_cached": true,
+      "loadavg_at_end": [
+        11.28,
+        9.17,
+        10.13
+      ],
+      "measured_unix": 1790443076.341974,
+      "settings": {
+        "engine": "jev-score-v2 (staged jev_score_v2.cpp, llama.cpp 441df11f, Metal)",
+        "n_gpu_layers": 999,
+        "flash_attn": "default (llama.cpp auto on Metal)",
+        "gguf": "release_2b/gguf/model-q8_0.gguf"
+      },
+      "raw": "latency/runs/gguf-q8_0_24576_10q.json"
+    },
+    {
+      "backend": "gguf-q4_k_m",
+      "n_questions": 1,
+      "state_tokens": 878,
+      "input_tokens_per_question": [
+        943
+      ],
+      "load_s": 1.4512,
+      "cold_s": 0.6738,
+      "warm_median_s": 0.6391,
+      "warm_s": [
+        0.6418,
+        0.6391,
+        0.6334
+      ],
+      "state_cached_median_s": 0.0782,
+      "peak_rss_bytes": {
+        "python": 334921728,
+        "jev_score_v2": 1988542464
+      },
+      "peak_phys_footprint_bytes": {
+        "python": 245073728,
+        "jev_score_v2": 669308992
+      },
+      "results_identical_cold_vs_state_cached": true,
+      "loadavg_at_end": [
+        12.77,
+        9.51,
+        10.25
+      ],
+      "measured_unix": 1790443081.420589,
+      "settings": {
+        "engine": "jev-score-v2 (staged jev_score_v2.cpp, llama.cpp 441df11f, Metal)",
+        "n_gpu_layers": 999,
+        "flash_attn": "default (llama.cpp auto on Metal)",
+        "gguf": "release_2b/gguf/model-q4_k_m.gguf"
+      },
+      "raw": "latency/runs/gguf-q4_k_m_1024_1q.json"
+    },
+    {
+      "backend": "gguf-q4_k_m",
+      "n_questions": 10,
+      "state_tokens": 878,
+      "input_tokens_per_question": [
+        943,
+        918,
+        925,
+        911,
+        918,
+        921,
+        943,
+        917,
+        912,
+        919
+      ],
+      "load_s": 0.9937,
+      "cold_s": 1.2045,
+      "warm_median_s": 1.157,
+      "warm_s": [
+        1.1544,
+        1.1643,
+        1.157
+      ],
+      "state_cached_median_s": 0.6009,
+      "peak_rss_bytes": {
+        "python": 352059392,
+        "jev_score_v2": 2018754560
+      },
+      "peak_phys_footprint_bytes": {
+        "python": 256673344,
+        "jev_score_v2": 665065728
+      },
+      "results_identical_cold_vs_state_cached": true,
+      "loadavg_at_end": [
+        13.83,
+        9.78,
+        10.34
+      ],
+      "measured_unix": 1790443089.6937559,
+      "settings": {
+        "engine": "jev-score-v2 (staged jev_score_v2.cpp, llama.cpp 441df11f, Metal)",
+        "n_gpu_layers": 999,
+        "flash_attn": "default (llama.cpp auto on Metal)",
+        "gguf": "release_2b/gguf/model-q4_k_m.gguf"
+      },
+      "raw": "latency/runs/gguf-q4_k_m_1024_10q.json"
+    },
+    {
+      "backend": "gguf-q4_k_m",
+      "n_questions": 1,
+      "state_tokens": 3950,
+      "input_tokens_per_question": [
+        4015
+      ],
+      "load_s": 0.9922,
+      "cold_s": 2.6608,
+      "warm_median_s": 2.6006,
+      "warm_s": [
+        2.6006,
+        2.6206,
+        2.5977
+      ],
+      "state_cached_median_s": 0.0908,
+      "peak_rss_bytes": {
+        "python": 327221248,
+        "jev_score_v2": 2062483456
+      },
+      "peak_phys_footprint_bytes": {
+        "python": 259393408,
+        "jev_score_v2": 714692992
+      },
+      "results_identical_cold_vs_state_cached": true,
+      "loadavg_at_end": [
+        11.99,
+        9.57,
+        10.25
+      ],
+      "measured_unix": 1790443102.1980171,
+      "settings": {
+        "engine": "jev-score-v2 (staged jev_score_v2.cpp, llama.cpp 441df11f, Metal)",
+        "n_gpu_layers": 999,
+        "flash_attn": "default (llama.cpp auto on Metal)",
+        "gguf": "release_2b/gguf/model-q4_k_m.gguf"
+      },
+      "raw": "latency/runs/gguf-q4_k_m_4096_1q.json"
+    },
+    {
+      "backend": "gguf-q4_k_m",
+      "n_questions": 10,
+      "state_tokens": 3950,
+      "input_tokens_per_question": [
+        4015,
+        3990,
+        3997,
+        3983,
+        3990,
+        3993,
+        4015,
+        3989,
+        3984,
+        3991
+      ],
+      "load_s": 0.9859,
+      "cold_s": 3.2053,
+      "warm_median_s": 3.1651,
+      "warm_s": [
+        3.1651,
+        3.1579,
+        3.1711
+      ],
+      "state_cached_median_s": 0.6462,
+      "peak_rss_bytes": {
+        "python": 326434816,
+        "jev_score_v2": 2073214976
+      },
+      "peak_phys_footprint_bytes": {
+        "python": 245614208,
+        "jev_score_v2": 716806400
+      },
+      "results_identical_cold_vs_state_cached": true,
+      "loadavg_at_end": [
+        10.28,
+        9.31,
+        10.15
+      ],
+      "measured_unix": 1790443118.634955,
+      "settings": {
+        "engine": "jev-score-v2 (staged jev_score_v2.cpp, llama.cpp 441df11f, Metal)",
+        "n_gpu_layers": 999,
+        "flash_attn": "default (llama.cpp auto on Metal)",
+        "gguf": "release_2b/gguf/model-q4_k_m.gguf"
+      },
+      "raw": "latency/runs/gguf-q4_k_m_4096_10q.json"
+    },
+    {
+      "backend": "gguf-q4_k_m",
+      "n_questions": 1,
+      "state_tokens": 24436,
+      "input_tokens_per_question": [
+        24501
+      ],
+      "load_s": 1.0041,
+      "cold_s": 18.6914,
+      "warm_median_s": 18.6165,
+      "warm_s": [
+        18.6165,
+        18.6059,
+        18.624
+      ],
+      "state_cached_median_s": 0.1755,
+      "peak_rss_bytes": {
+        "python": 345686016,
+        "jev_score_v2": 2233663488
+      },
+      "peak_phys_footprint_bytes": {
+        "python": 260048576,
+        "jev_score_v2": 880335552
+      },
+      "results_identical_cold_vs_state_cached": true,
+      "loadavg_at_end": [
+        8.43,
+        9.22,
+        10.05
+      ],
+      "measured_unix": 1790443195.534742,
+      "settings": {
+        "engine": "jev-score-v2 (staged jev_score_v2.cpp, llama.cpp 441df11f, Metal)",
+        "n_gpu_layers": 999,
+        "flash_attn": "default (llama.cpp auto on Metal)",
+        "gguf": "release_2b/gguf/model-q4_k_m.gguf"
+      },
+      "raw": "latency/runs/gguf-q4_k_m_24576_1q.json"
+    },
+    {
+      "backend": "gguf-q4_k_m",
+      "n_questions": 10,
+      "state_tokens": 24436,
+      "input_tokens_per_question": [
+        24501,
+        24476,
+        24483,
+        24469,
+        24476,
+        24479,
+        24501,
+        24475,
+        24470,
+        24477
+      ],
+      "load_s": 0.9967,
+      "cold_s": 19.4776,
+      "warm_median_s": 19.4309,
+      "warm_s": [
+        19.4309,
+        19.4251,
+        19.5057
+      ],
+      "state_cached_median_s": 0.9616,
+      "peak_rss_bytes": {
+        "python": 354992128,
+        "jev_score_v2": 2226913280
+      },
+      "peak_phys_footprint_bytes": {
+        "python": 250791552,
+        "jev_score_v2": 870128320
+      },
+      "results_identical_cold_vs_state_cached": true,
+      "loadavg_at_end": [
+        5.63,
+        8.19,
+        9.59
+      ],
+      "measured_unix": 1790443278.075001,
+      "settings": {
+        "engine": "jev-score-v2 (staged jev_score_v2.cpp, llama.cpp 441df11f, Metal)",
+        "n_gpu_layers": 999,
+        "flash_attn": "default (llama.cpp auto on Metal)",
+        "gguf": "release_2b/gguf/model-q4_k_m.gguf"
+      },
+      "raw": "latency/runs/gguf-q4_k_m_24576_10q.json"
+    },
+    {
+      "backend": "mlx-bf16",
+      "n_questions": 1,
+      "state_tokens": 878,
+      "input_tokens_per_question": [
+        943
+      ],
+      "load_s": 3.6602,
+      "cold_s": 0.64,
+      "warm_median_s": 0.5658,
+      "warm_s": [
+        0.5615,
+        0.5665,
+        0.5658
+      ],
+      "state_cached_median_s": 0.0796,
+      "peak_rss_bytes": {
+        "python": 4401463296
+      },
+      "peak_phys_footprint_bytes": {
+        "python": 5205486080
+      },
+      "mlx_peak_memory_bytes": 4536436474,
+      "results_identical_cold_vs_state_cached": true,
+      "loadavg_at_end": [
+        6.5,
+        8.17,
+        9.53
+      ],
+      "measured_unix": 1790443305.918901,
+      "settings": {
+        "engine": "mlx 0.32.2 / mlx-lm 0.31.3 (staged runtime)",
+        "precision": "bf16",
+        "compute_dtype": null
+      },
+      "raw": "latency/runs/mlx-bf16_1024_1q.json"
+    },
+    {
+      "backend": "mlx-bf16",
+      "n_questions": 10,
+      "state_tokens": 878,
+      "input_tokens_per_question": [
+        943,
+        918,
+        925,
+        911,
+        918,
+        921,
+        943,
+        917,
+        912,
+        919
+      ],
+      "load_s": 3.3114,
+      "cold_s": 1.1022,
+      "warm_median_s": 1.0651,
+      "warm_s": [
+        1.0651,
+        1.0647,
+        1.0714
+      ],
+      "state_cached_median_s": 0.5726,
+      "peak_rss_bytes": {
+        "python": 4416684032
+      },
+      "peak_phys_footprint_bytes": {
+        "python": 5575518912
+      },
+      "mlx_peak_memory_bytes": 4536436474,
+      "results_identical_cold_vs_state_cached": true,
+      "loadavg_at_end": [
+        6.03,
+        8.01,
+        9.46
+      ],
+      "measured_unix": 1790443316.3707972,
+      "settings": {
+        "engine": "mlx 0.32.2 / mlx-lm 0.31.3 (staged runtime)",
+        "precision": "bf16",
+        "compute_dtype": null
+      },
+      "raw": "latency/runs/mlx-bf16_1024_10q.json"
+    },
+    {
+      "backend": "mlx-bf16",
+      "n_questions": 1,
+      "state_tokens": 3950,
+      "input_tokens_per_question": [
+        4015
+      ],
+      "load_s": 3.2848,
+      "cold_s": 2.265,
+      "warm_median_s": 2.2579,
+      "warm_s": [
+        2.2462,
+        2.2579,
+        2.2701
+      ],
+      "state_cached_median_s": 0.0903,
+      "peak_rss_bytes": {
+        "python": 4408279040
+      },
+      "peak_phys_footprint_bytes": {
+        "python": 6899099712
+      },
+      "mlx_peak_memory_bytes": 5062198966,
+      "results_identical_cold_vs_state_cached": true,
+      "loadavg_at_end": [
+        5.78,
+        7.9,
+        9.4
+      ],
+      "measured_unix": 1790443330.072621,
+      "settings": {
+        "engine": "mlx 0.32.2 / mlx-lm 0.31.3 (staged runtime)",
+        "precision": "bf16",
+        "compute_dtype": null
+      },
+      "raw": "latency/runs/mlx-bf16_4096_1q.json"
+    },
+    {
+      "backend": "mlx-bf16",
+      "n_questions": 10,
+      "state_tokens": 3950,
+      "input_tokens_per_question": [
+        4015,
+        3990,
+        3997,
+        3983,
+        3990,
+        3993,
+        4015,
+        3989,
+        3984,
+        3991
+      ],
+      "load_s": 3.3381,
+      "cold_s": 2.8286,
+      "warm_median_s": 2.7998,
+      "warm_s": [
+        2.832,
+        2.7968,
+        2.7998
+      ],
+      "state_cached_median_s": 0.622,
+      "peak_rss_bytes": {
+        "python": 4406149120
+      },
+      "peak_phys_footprint_bytes": {
+        "python": 6943402624
+      },
+      "mlx_peak_memory_bytes": 5062395574,
+      "results_identical_cold_vs_state_cached": true,
+      "loadavg_at_end": [
+        5.53,
+        7.71,
+        9.3
+      ],
+      "measured_unix": 1790443347.640775,
+      "settings": {
+        "engine": "mlx 0.32.2 / mlx-lm 0.31.3 (staged runtime)",
+        "precision": "bf16",
+        "compute_dtype": null
+      },
+      "raw": "latency/runs/mlx-bf16_4096_10q.json"
+    },
+    {
+      "backend": "mlx-bf16",
+      "n_questions": 1,
+      "state_tokens": 24436,
+      "input_tokens_per_question": [
+        24501
+      ],
+      "load_s": 3.2657,
+      "cold_s": 15.472,
+      "warm_median_s": 15.5073,
+      "warm_s": [
+        15.5202,
+        15.5073,
+        15.4734
+      ],
+      "state_cached_median_s": 0.1546,
+      "peak_rss_bytes": {
+        "python": 4403707904
+      },
+      "peak_phys_footprint_bytes": {
+        "python": 7889725824
+      },
+      "mlx_peak_memory_bytes": 5942822582,
+      "results_identical_cold_vs_state_cached": true,
+      "loadavg_at_end": [
+        6.15,
+        7.49,
+        9.1
+      ],
+      "measured_unix": 1790443414.47198,
+      "settings": {
+        "engine": "mlx 0.32.2 / mlx-lm 0.31.3 (staged runtime)",
+        "precision": "bf16",
+        "compute_dtype": null
+      },
+      "raw": "latency/runs/mlx-bf16_24576_1q.json"
+    },
+    {
+      "backend": "mlx-bf16",
+      "n_questions": 10,
+      "state_tokens": 24436,
+      "input_tokens_per_question": [
+        24501,
+        24476,
+        24483,
+        24469,
+        24476,
+        24479,
+        24501,
+        24475,
+        24470,
+        24477
+      ],
+      "load_s": 3.278,
+      "cold_s": 16.2813,
+      "warm_median_s": 16.2351,
+      "warm_s": [
+        16.2243,
+        16.2351,
+        16.2657
+      ],
+      "state_cached_median_s": 0.8797,
+      "peak_rss_bytes": {
+        "python": 4405805056
+      },
+      "peak_phys_footprint_bytes": {
+        "python": 7863035904
+      },
+      "mlx_peak_memory_bytes": 5942806198,
+      "results_identical_cold_vs_state_cached": true,
+      "loadavg_at_end": [
+        8.07,
+        7.75,
+        9.06
+      ],
+      "measured_unix": 1790443486.522902,
+      "settings": {
+        "engine": "mlx 0.32.2 / mlx-lm 0.31.3 (staged runtime)",
+        "precision": "bf16",
+        "compute_dtype": null
+      },
+      "raw": "latency/runs/mlx-bf16_24576_10q.json"
+    },
+    {
+      "backend": "mlx-8bit",
+      "n_questions": 1,
+      "state_tokens": 878,
+      "input_tokens_per_question": [
+        943
+      ],
+      "load_s": 3.3345,
+      "cold_s": 0.7573,
+      "warm_median_s": 0.72,
+      "warm_s": [
+        0.7177,
+        0.72,
+        0.7228
+      ],
+      "state_cached_median_s": 0.0834,
+      "peak_rss_bytes": {
+        "python": 2640986112
+      },
+      "peak_phys_footprint_bytes": {
+        "python": 3772211392
+      },
+      "mlx_peak_memory_bytes": 3074287404,
+      "results_identical_cold_vs_state_cached": true,
+      "loadavg_at_end": [
+        8.14,
+        7.77,
+        9.06
+      ],
+      "measured_unix": 1790443494.139926,
+      "settings": {
+        "engine": "mlx 0.32.2 / mlx-lm 0.31.3 (staged runtime)",
+        "precision": "8bit",
+        "compute_dtype": null
+      },
+      "raw": "latency/runs/mlx-8bit_1024_1q.json"
+    },
+    {
+      "backend": "mlx-8bit",
+      "n_questions": 10,
+      "state_tokens": 878,
+      "input_tokens_per_question": [
+        943,
+        918,
+        925,
+        911,
+        918,
+        921,
+        943,
+        917,
+        912,
+        919
+      ],
+      "load_s": 3.1502,
+      "cold_s": 1.3124,
+      "warm_median_s": 1.2657,
+      "warm_s": [
+        1.2719,
+        1.2657,
+        1.2621
+      ],
+      "state_cached_median_s": 0.6288,
+      "peak_rss_bytes": {
+        "python": 2635268096
+      },
+      "peak_phys_footprint_bytes": {
+        "python": 4171391040
+      },
+      "mlx_peak_memory_bytes": 3074287404,
+      "results_identical_cold_vs_state_cached": true,
+      "loadavg_at_end": [
+        7.51,
+        7.65,
+        9.0
+      ],
+      "measured_unix": 1790443505.379929,
+      "settings": {
+        "engine": "mlx 0.32.2 / mlx-lm 0.31.3 (staged runtime)",
+        "precision": "8bit",
+        "compute_dtype": null
+      },
+      "raw": "latency/runs/mlx-8bit_1024_10q.json"
+    },
+    {
+      "backend": "mlx-8bit",
+      "n_questions": 1,
+      "state_tokens": 3950,
+      "input_tokens_per_question": [
+        4015
+      ],
+      "load_s": 3.1569,
+      "cold_s": 2.9904,
+      "warm_median_s": 2.9581,
+      "warm_s": [
+        2.9525,
+        2.9624,
+        2.9581
+      ],
+      "state_cached_median_s": 0.0922,
+      "peak_rss_bytes": {
+        "python": 2639839232
+      },
+      "peak_phys_footprint_bytes": {
+        "python": 5536832896
+      },
+      "mlx_peak_memory_bytes": 3471550184,
+      "results_identical_cold_vs_state_cached": true,
+      "loadavg_at_end": [
+        8.57,
+        7.86,
+        9.04
+      ],
+      "measured_unix": 1790443521.78049,
+      "settings": {
+        "engine": "mlx 0.32.2 / mlx-lm 0.31.3 (staged runtime)",
+        "precision": "8bit",
+        "compute_dtype": null
+      },
+      "raw": "latency/runs/mlx-8bit_4096_1q.json"
+    },
+    {
+      "backend": "mlx-8bit",
+      "n_questions": 10,
+      "state_tokens": 3950,
+      "input_tokens_per_question": [
+        4015,
+        3990,
+        3997,
+        3983,
+        3990,
+        3993,
+        4015,
+        3989,
+        3984,
+        3991
+      ],
+      "load_s": 3.1665,
+      "cold_s": 3.572,
+      "warm_median_s": 3.5432,
+      "warm_s": [
+        3.5584,
+        3.5432,
+        3.5354
+      ],
+      "state_cached_median_s": 0.6768,
+      "peak_rss_bytes": {
+        "python": 2624536576
+      },
+      "peak_phys_footprint_bytes": {
+        "python": 5698149760
+      },
+      "mlx_peak_memory_bytes": 3471615720,
+      "results_identical_cold_vs_state_cached": true,
+      "loadavg_at_end": [
+        7.82,
+        7.73,
+        8.97
+      ],
+      "measured_unix": 1790443542.268524,
+      "settings": {
+        "engine": "mlx 0.32.2 / mlx-lm 0.31.3 (staged runtime)",
+        "precision": "8bit",
+        "compute_dtype": null
+      },
+      "raw": "latency/runs/mlx-8bit_4096_10q.json"
+    },
+    {
+      "backend": "mlx-8bit",
+      "n_questions": 1,
+      "state_tokens": 24436,
+      "input_tokens_per_question": [
+        24501
+      ],
+      "load_s": 3.1447,
+      "cold_s": 19.75,
+      "warm_median_s": 19.8614,
+      "warm_s": [
+        19.8005,
+        19.8614,
+        19.9115
+      ],
+      "state_cached_median_s": 0.1562,
+      "peak_rss_bytes": {
+        "python": 2643197952
+      },
+      "peak_phys_footprint_bytes": {
+        "python": 6518807424
+      },
+      "mlx_peak_memory_bytes": 4350994102,
+      "results_identical_cold_vs_state_cached": true,
+      "loadavg_at_end": [
+        6.7,
+        7.34,
+        8.69
+      ],
+      "measured_unix": 1790443626.3357399,
+      "settings": {
+        "engine": "mlx 0.32.2 / mlx-lm 0.31.3 (staged runtime)",
+        "precision": "8bit",
+        "compute_dtype": null
+      },
+      "raw": "latency/runs/mlx-8bit_24576_1q.json"
+    },
+    {
+      "backend": "mlx-8bit",
+      "n_questions": 10,
+      "state_tokens": 24436,
+      "input_tokens_per_question": [
+        24501,
+        24476,
+        24483,
+        24469,
+        24476,
+        24479,
+        24501,
+        24475,
+        24470,
+        24477
+      ],
+      "load_s": 3.1668,
+      "cold_s": 20.7629,
+      "warm_median_s": 20.7793,
+      "warm_s": [
+        20.6782,
+        20.7793,
+        21.8013
+      ],
+      "state_cached_median_s": 0.9543,
+      "peak_rss_bytes": {
+        "python": 2643705856
+      },
+      "peak_phys_footprint_bytes": {
+        "python": 6522985664
+      },
+      "mlx_peak_memory_bytes": 4350944950,
+      "results_identical_cold_vs_state_cached": true,
+      "loadavg_at_end": [
+        8.26,
+        8.35,
+        8.99
+      ],
+      "measured_unix": 1790443717.49426,
+      "settings": {
+        "engine": "mlx 0.32.2 / mlx-lm 0.31.3 (staged runtime)",
+        "precision": "8bit",
+        "compute_dtype": null
+      },
+      "raw": "latency/runs/mlx-8bit_24576_10q.json"
+    },
+    {
+      "backend": "torch-mps-fp32",
+      "n_questions": 1,
+      "state_tokens": 878,
+      "input_tokens_per_question": [
+        943
+      ],
+      "load_s": 7.2396,
+      "cold_s": 2.0776,
+      "warm_median_s": 1.82,
+      "warm_s": [
+        1.82,
+        1.8185,
+        1.8657
+      ],
+      "state_cached_median_s": 0.2243,
+      "peak_rss_bytes": {
+        "python": 11777310720
+      },
+      "peak_phys_footprint_bytes": {
+        "python": 9488876864
+      },
+      "results_identical_cold_vs_state_cached": true,
+      "loadavg_at_end": [
+        11.08,
+        15.97,
+        14.21
+      ],
+      "measured_unix": 1790433935.3785548,
+      "settings": {
+        "engine": "transformers 5.17 / torch 2.14 (staged runtime)",
+        "device": "mps",
+        "dtype": "float32",
+        "attn_chunk": 1024
+      },
+      "raw": "latency/runs/torch-mps-fp32_1024_1q.json"
+    },
+    {
+      "backend": "torch-mps-fp32",
+      "n_questions": 10,
+      "state_tokens": 878,
+      "input_tokens_per_question": [
+        943,
+        918,
+        925,
+        911,
+        918,
+        921,
+        943,
+        917,
+        912,
+        919
+      ],
+      "load_s": 5.8844,
+      "cold_s": 3.487,
+      "warm_median_s": 3.126,
+      "warm_s": [
+        3.126,
+        3.124,
+        3.1357
+      ],
+      "state_cached_median_s": 1.534,
+      "peak_rss_bytes": {
+        "python": 11766398976
+      },
+      "peak_phys_footprint_bytes": {
+        "python": 9348826496
+      },
+      "results_identical_cold_vs_state_cached": true,
+      "loadavg_at_end": [
+        10.92,
+        15.56,
+        14.11
+      ],
+      "measured_unix": 1790433960.3376808,
+      "settings": {
+        "engine": "transformers 5.17 / torch 2.14 (staged runtime)",
+        "device": "mps",
+        "dtype": "float32",
+        "attn_chunk": 1024
+      },
+      "raw": "latency/runs/torch-mps-fp32_1024_10q.json"
+    },
+    {
+      "backend": "torch-mps-fp32",
+      "n_questions": 1,
+      "state_tokens": 3950,
+      "input_tokens_per_question": [
+        4015
+      ],
+      "load_s": 6.6133,
+      "cold_s": 7.9626,
+      "warm_median_s": 7.7385,
+      "warm_s": [
+        7.7135,
+        7.7385,
+        7.7569
+      ],
+      "state_cached_median_s": 0.2508,
+      "peak_rss_bytes": {
+        "python": 11774099456
+      },
+      "peak_phys_footprint_bytes": {
+        "python": 9830892736
+      },
+      "results_identical_cold_vs_state_cached": true,
+      "loadavg_at_end": [
+        8.41,
+        7.52,
+        7.06
+      ],
+      "measured_unix": 1790438685.325036,
+      "settings": {
+        "engine": "transformers 5.17 / torch 2.14 (staged runtime)",
+        "device": "mps",
+        "dtype": "float32",
+        "attn_chunk": 1024
+      },
+      "raw": "latency/runs/torch-mps-fp32_4096_1q.json"
+    },
+    {
+      "backend": "torch-mps-fp32",
+      "n_questions": 10,
+      "state_tokens": 3950,
+      "input_tokens_per_question": [
+        4015,
+        3990,
+        3997,
+        3983,
+        3990,
+        3993,
+        4015,
+        3989,
+        3984,
+        3991
+      ],
+      "load_s": 5.8354,
+      "cold_s": 9.4943,
+      "warm_median_s": 9.2321,
+      "warm_s": [
+        9.1731,
+        9.2321,
+        9.2805
+      ],
+      "state_cached_median_s": 1.7373,
+      "peak_rss_bytes": {
+        "python": 11766579200
+      },
+      "peak_phys_footprint_bytes": {
+        "python": 9740780928
+      },
+      "results_identical_cold_vs_state_cached": true,
+      "loadavg_at_end": [
+        10.4,
+        8.12,
+        7.31
+      ],
+      "measured_unix": 1790438734.979529,
+      "settings": {
+        "engine": "transformers 5.17 / torch 2.14 (staged runtime)",
+        "device": "mps",
+        "dtype": "float32",
+        "attn_chunk": 1024
+      },
+      "raw": "latency/runs/torch-mps-fp32_4096_10q.json"
+    },
+    {
+      "backend": "torch-mps-fp32",
+      "n_questions": 1,
+      "state_tokens": 24436,
+      "input_tokens_per_question": [
+        24501
+      ],
+      "load_s": 7.5475,
+      "cold_s": 56.4354,
+      "warm_median_s": 61.9146,
+      "warm_s": [
+        55.156,
+        61.9146,
+        80.3434
+      ],
+      "state_cached_median_s": 0.6,
+      "peak_rss_bytes": {
+        "python": 11774935040
+      },
+      "peak_phys_footprint_bytes": {
+        "python": 13703111296
+      },
+      "results_identical_cold_vs_state_cached": true,
+      "loadavg_at_end": [
+        7.27,
+        7.84,
+        7.44
+      ],
+      "measured_unix": 1790438999.6904268,
+      "settings": {
+        "engine": "transformers 5.17 / torch 2.14 (staged runtime)",
+        "device": "mps",
+        "dtype": "float32",
+        "attn_chunk": 1024
+      },
+      "raw": "latency/runs/torch-mps-fp32_24576_1q.json"
+    },
+    {
+      "backend": "torch-mps-fp32",
+      "n_questions": 10,
+      "state_tokens": 24436,
+      "input_tokens_per_question": [
+        24501,
+        24476,
+        24483,
+        24469,
+        24476,
+        24479,
+        24501,
+        24475,
+        24470,
+        24477
+      ],
+      "load_s": 7.9699,
+      "cold_s": 72.6584,
+      "warm_median_s": 72.3035,
+      "warm_s": [
+        83.2192,
+        72.3035,
+        58.3757
+      ],
+      "state_cached_median_s": 2.7217,
+      "peak_rss_bytes": {
+        "python": 11775229952
+      },
+      "peak_phys_footprint_bytes": {
+        "python": 13640000128
+      },
+      "results_identical_cold_vs_state_cached": true,
+      "loadavg_at_end": [
+        9.6,
+        8.35,
+        7.72
+      ],
+      "measured_unix": 1790439304.235886,
+      "settings": {
+        "engine": "transformers 5.17 / torch 2.14 (staged runtime)",
+        "device": "mps",
+        "dtype": "float32",
+        "attn_chunk": 1024
+      },
+      "raw": "latency/runs/torch-mps-fp32_24576_10q.json"
+    }
+  ],
+  "scripts": [
+    "release_2b/latency/latency_run.py",
+    "release_2b/latency/run_all.sh",
+    "release_2b/latency/aggregate.py"
+  ]
+}

validation/parity/PREDECLARED_RELEASE_GATES_2B.md

@@ -0,0 +1,29 @@
+# Jev-Style-2B-Decision-v3 — release format gates (pre-declared 2026-09-26, BEFORE any trained-weight format was scored)
+
+Reference: HF FP32 CPU, exact v2 block attention, on the released bf16 text-only checkpoint
+runs/macjev/candidate_2b/hf-candidate (SHA256SUMS 279/279 verified). Temperature T = 0.8278650621 (macjev_readout.json).
+
+Fixtures (frozen, sha256 in their .stats/.README):
+- base fixture: runs/macjev/runtime_v2_dev/reference_base/fixture.jsonl (35 req / 43 q, 214,700 tok, up to 25,600 tok, overflow, K<=151)
+- gate fixture: runs/macjev/runtime_v2_dev/reference/gate_fixture_n1000.jsonl (1,000 real dev rows, <=4,096 tok)
+
+Gates (training plan §11.2, unchanged): F16/bf16 top-1 >= .99; 8-bit top-1 >= .98; |dNLL| <= .02 (fp/16/8-bit);
+accuracy drop Q8_0/MLX-8bit <= 0.3 pp, Q4_K_M <= 1.0 pp (measured on the gate fixture; n=1000 => 0.1 pp per flip).
+Block-mask proof on the base fixture: every question closer to the block reference than to the causal and prefix-causal controls.
+
+Release set (user choice): GGUF F16 / Q8_0 / Q4_K_M; MLX bf16 / 8bit (affine g64).
+Recipe 1 (primary): stock llama-quantize defaults; mlx_lm.convert defaults + FP32 norm sidecar + MLXScorerV2.
+Recipe 2 (declared fallback, used ONLY if recipe 1 fails a gate): keep the tied embedding / readout rows at f16
+(llama-quantize --token-embedding-type f16; MLX: quant predicate excluding embed_tokens). Same gates, same fixtures.
+If recipe 2 also fails, that format is NOT shipped. Gates are not relaxed after seeing results.
+
+## Public benchmarks (pre-declared 2026-09-26 22:01:49 AEST, before any 2B benchmark item was scored)
+Timing (erratum added 2026-09-27 03:35 AEST): this section was appended by the same shell command that launched the
+benchmark runs, immediately before the first item (file mtime 22:01:49; run log `runs/macjev/bench_2b/run_trained.log`
+first line `[2026-09-26 22:01:49] START jevbench`; first result seen 22:06:28). The heading originally said
+"22:10" by mistake. The format-gate section above was written when the file was created (21:46:03), before any
+trained-weight format was scored (first format score started 21:53:00).
+- Reported engine: GGUF F16 (runs/macjev/release_2b/gguf/model-f16.gguf, jev-score-v2, Metal), global T = 0.8278650621 only.
+- JevBench v1.4.1 public (231) and elcronos zero-shot sets (tweet_topic / fin_topic / daily_dialog), each run ONCE,
+  same metric code as the 0.8B v3 card. No per-benchmark temperatures, no reruns, no prompt changes after seeing scores.
+- Decision Index 0.2: not run locally; requested from the maintainer after release (user decision 2026-09-26).

validation/parity/compare_mlx_affine8-g64_base_vs_block.json

@@ -0,0 +1,276 @@
+{
+ "format": "8bit",
+ "temperature": 0.8278650620942867,
+ "overall": {
+  "n": 43,
+  "top1_agreement": 1.0,
+  "top1_flips": 0,
+  "max_abs_dscore": 0.4653351306915283,
+  "mean_abs_dscore": 0.023077200647273945,
+  "mean_kl": 0.0006975644872166873,
+  "n_gold": 43,
+  "nll_ref": 0.670433307194272,
+  "nll_cand": 0.6868139442506376,
+  "dnll": 0.01638063705636561,
+  "mean_abs_row_dnll": 0.030139399526949388,
+  "acc_ref": 0.7209302325581395,
+  "acc_cand": 0.7209302325581395,
+  "acc_drop_pp": 0.0
+ },
+ "per_category": {
+  "catalogue_overflow": {
+   "n": 3,
+   "top1_agreement": 1.0,
+   "top1_flips": 0,
+   "max_abs_dscore": 0.2735496163368225,
+   "mean_abs_dscore": 0.019241230377298316,
+   "mean_kl": 0.0012829640019479873,
+   "n_gold": 3,
+   "nll_ref": 0.5023960827418966,
+   "nll_cand": 0.46349011750257935,
+   "dnll": -0.038905965239317275,
+   "mean_abs_row_dnll": 0.039379824994875905,
+   "acc_ref": 0.6666666666666666,
+   "acc_cand": 0.6666666666666666,
+   "acc_drop_pp": 0.0
+  },
+  "catalogue_overflow_long_state": {
+   "n": 1,
+   "top1_agreement": 1.0,
+   "top1_flips": 0,
+   "max_abs_dscore": 0.09829854965209961,
+   "mean_abs_dscore": 0.02608197888001701,
+   "mean_kl": 0.00037418306666026774,
+   "n_gold": 1,
+   "nll_ref": 1.868953922917252,
+   "nll_cand": 1.8305656122814864,
+   "dnll": -0.03838831063576564,
+   "mean_abs_row_dnll": 0.03838831063576564,
+   "acc_ref": 0.0,
+   "acc_cand": 0.0,
+   "acc_drop_pp": 0.0
+  },
+  "long_16k": {
+   "n": 3,
+   "top1_agreement": 1.0,
+   "top1_flips": 0,
+   "max_abs_dscore": 0.4653351306915283,
+   "mean_abs_dscore": 0.04898447410160343,
+   "mean_kl": 0.005652873428140343,
+   "n_gold": 3,
+   "nll_ref": 1.098996986024778,
+   "nll_cand": 1.3040987437659517,
+   "dnll": 0.20510175774117378,
+   "mean_abs_row_dnll": 0.2051017577411738,
+   "acc_ref": 0.6666666666666666,
+   "acc_cand": 0.6666666666666666,
+   "acc_drop_pp": 0.0
+  },
+  "long_24k": {
+   "n": 2,
+   "top1_agreement": 1.0,
+   "top1_flips": 0,
+   "max_abs_dscore": 0.068747878074646,
+   "mean_abs_dscore": 0.03896120687325796,
+   "mean_kl": 0.0007690578648421433,
+   "n_gold": 2,
+   "nll_ref": 0.43411292245601785,
+   "nll_cand": 0.43100727732191557,
+   "dnll": -0.003105645134102275,
+   "mean_abs_row_dnll": 0.019953016950989222,
+   "acc_ref": 1.0,
+   "acc_cand": 1.0,
+   "acc_drop_pp": 0.0
+  },
+  "many_options": {
+   "n": 2,
+   "top1_agreement": 1.0,
+   "top1_flips": 0,
+   "max_abs_dscore": 0.06005406379699707,
+   "mean_abs_dscore": 0.01609148249030113,
+   "mean_kl": 3.102768330017338e-06,
+   "n_gold": 2,
+   "nll_ref": 0.012124599411053285,
+   "nll_cand": 0.012000711496263293,
+   "dnll": -0.00012388791478999163,
+   "mean_abs_row_dnll": 0.00012388791478999076,
+   "acc_ref": 1.0,
+   "acc_cand": 1.0,
+   "acc_drop_pp": 0.0
+  },
+  "multi_block_3k": {
+   "n": 2,
+   "top1_agreement": 1.0,
+   "top1_flips": 0,
+   "max_abs_dscore": 0.037885069847106934,
+   "mean_abs_dscore": 0.015739992260932922,
+   "mean_kl": 0.00011035902286322221,
+   "n_gold": 2,
+   "nll_ref": 0.10264503770560551,
+   "nll_cand": 0.09882922575086513,
+   "dnll": -0.0038158119547403863,
+   "mean_abs_row_dnll": 0.00429795147361977,
+   "acc_ref": 1.0,
+   "acc_cand": 1.0,
+   "acc_drop_pp": 0.0
+  },
+  "multi_block_5k": {
+   "n": 2,
+   "top1_agreement": 1.0,
+   "top1_flips": 0,
+   "max_abs_dscore": 0.0428202748298645,
+   "mean_abs_dscore": 0.02182019054889679,
+   "mean_kl": 4.99818269133305e-05,
+   "n_gold": 2,
+   "nll_ref": 0.41414880456154163,
+   "nll_cand": 0.40767085022183286,
+   "dnll": -0.006477954339708769,
+   "mean_abs_row_dnll": 0.006477954339708783,
+   "acc_ref": 1.0,
+   "acc_cand": 1.0,
+   "acc_drop_pp": 0.0
+  },
+  "multi_block_9k": {
+   "n": 2,
+   "top1_agreement": 1.0,
+   "top1_flips": 0,
+   "max_abs_dscore": 0.02951335906982422,
+   "mean_abs_dscore": 0.02005617693066597,
+   "mean_kl": 6.540709706121381e-06,
+   "n_gold": 2,
+   "nll_ref": 1.387121106483142,
+   "nll_cand": 1.3792866299819007,
+   "dnll": -0.00783447650124125,
+   "mean_abs_row_dnll": 0.00783447650124136,
+   "acc_ref": 0.5,
+   "acc_cand": 0.5,
+   "acc_drop_pp": 0.0
+  },
+  "multi_question": {
+   "n": 10,
+   "top1_agreement": 1.0,
+   "top1_flips": 0,
+   "max_abs_dscore": 0.06343865394592285,
+   "mean_abs_dscore": 0.018984448437889417,
+   "mean_kl": 0.00014264548450480086,
+   "n_gold": 10,
+   "nll_ref": 0.7606667410227165,
+   "nll_cand": 0.7742217565907421,
+   "dnll": 0.01355501556802563,
+   "mean_abs_row_dnll": 0.016602018627922027,
+   "acc_ref": 0.7,
+   "acc_cand": 0.7,
+   "acc_drop_pp": 0.0
+  },
+  "prefix_boundary": {
+   "n": 7,
+   "top1_agreement": 1.0,
+   "top1_flips": 0,
+   "max_abs_dscore": 0.07407116889953613,
+   "mean_abs_dscore": 0.02506211300690969,
+   "mean_kl": 0.000275167227242767,
+   "n_gold": 7,
+   "nll_ref": 0.9445710499909146,
+   "nll_cand": 0.9592024225354618,
+   "dnll": 0.014631372544547272,
+   "mean_abs_row_dnll": 0.026803385065686018,
+   "acc_ref": 0.5714285714285714,
+   "acc_cand": 0.5714285714285714,
+   "acc_drop_pp": 0.0
+  },
+  "qtype_choice": {
+   "n": 1,
+   "top1_agreement": 1.0,
+   "top1_flips": 0,
+   "max_abs_dscore": 0.039629340171813965,
+   "mean_abs_dscore": 0.01868373155593872,
+   "mean_kl": 9.758721027128357e-05,
+   "n_gold": 1,
+   "nll_ref": 1.2471588298811915,
+   "nll_cand": 1.2644290223938441,
+   "dnll": 0.017270192512652605,
+   "mean_abs_row_dnll": 0.017270192512652605,
+   "acc_ref": 0.0,
+   "acc_cand": 0.0,
+   "acc_drop_pp": 0.0
+  },
+  "qtype_noul": {
+   "n": 1,
+   "top1_agreement": 1.0,
+   "top1_flips": 0,
+   "max_abs_dscore": 0.04642695188522339,
+   "mean_abs_dscore": 0.02744242548942566,
+   "mean_kl": 0.00046207429765575576,
+   "n_gold": 1,
+   "nll_ref": 0.3643240477162341,
+   "nll_cand": 0.3445434412562239,
+   "dnll": -0.01978060646001023,
+   "mean_abs_row_dnll": 0.01978060646001023,
+   "acc_ref": 1.0,
+   "acc_cand": 1.0,
+   "acc_drop_pp": 0.0
+  },
+  "qtype_score": {
+   "n": 1,
+   "top1_agreement": 1.0,
+   "top1_flips": 0,
+   "max_abs_dscore": 0.04161477088928223,
+   "mean_abs_dscore": 0.015557724237442016,
+   "mean_kl": 1.4547309373278143e-05,
+   "n_gold": 1,
+   "nll_ref": 0.9393375234945696,
+   "nll_cand": 0.9372245712212975,
+   "dnll": -0.0021129522732720174,
+   "mean_abs_row_dnll": 0.0021129522732720174,
+   "acc_ref": 0.0,
+   "acc_cand": 0.0,
+   "acc_drop_pp": 0.0
+  },
+  "short_sb": {
+   "n": 6,
+   "top1_agreement": 1.0,
+   "top1_flips": 0,
+   "max_abs_dscore": 0.12482678890228271,
+   "mean_abs_dscore": 0.01820988009964663,
+   "mean_kl": 0.0005014431591724873,
+   "n_gold": 6,
+   "nll_ref": 0.11428482960768895,
+   "nll_cand": 0.123207743102961,
+   "dnll": 0.00892291349527205,
+   "mean_abs_row_dnll": 0.008996485578208901,
+   "acc_ref": 1.0,
+   "acc_cand": 1.0,
+   "acc_drop_pp": 0.0
+  }
+ },
+ "gates": {
+  "coverage_finite_tokens": {
+   "pass": true,
+   "problems": {},
+   "reference_missing_requests": 0
+  },
+  "top1": {
+   "threshold": 0.98,
+   "value": 1.0,
+   "pass": true
+  },
+  "dnll": {
+   "threshold": 0.02,
+   "value": 0.01638063705636561,
+   "pass": true
+  },
+  "acc_drop_pp": {
+   "threshold": 0.3,
+   "value": 0.0,
+   "resolution_pp": 2.3255813953488373,
+   "pass": null,
+   "note": "UNRESOLVED: 43 gold questions -> one flip = 2.33pp > 0.3pp; this fixture cannot resolve the gate (use the gate fixture with >= 334 rows; statistical power needs far more)"
+  }
+ },
+ "verdict": "INCONCLUSIVE",
+ "resolution_pp_per_flip": 2.3255813953488373,
+ "note": "small fixtures cannot resolve 0.3pp accuracy gates; accuracy on a parity fixture is only indicative (synthetic long rows carry the base row's gold label).",
+ "ref": "runs/macjev/runtime_v2_dev/reference_trained/ref_block_fp32.jsonl",
+ "cand": "runs/macjev/release_2b/parity/pred_mlx_affine8-g64_base.jsonl",
+ "fixture": "runs/macjev/runtime_v2_dev/reference_base/fixture.jsonl"
+}

validation/parity/compare_mlx_affine8-g64_gate_vs_block.json

@@ -0,0 +1,228 @@
+{
+ "format": "8bit",
+ "temperature": 0.8278650620942867,
+ "overall": {
+  "n": 1000,
+  "top1_agreement": 0.996,
+  "top1_flips": 4,
+  "max_abs_dscore": 0.3184394836425781,
+  "mean_abs_dscore": 0.016680597170951345,
+  "mean_kl": 0.0002134925530053613,
+  "n_gold": 1000,
+  "nll_ref": 0.4436679457518203,
+  "nll_cand": 0.44417151398689614,
+  "dnll": 0.0005035682350758575,
+  "mean_abs_row_dnll": 0.008527335296640175,
+  "acc_ref": 0.808,
+  "acc_cand": 0.808,
+  "acc_drop_pp": 0.0
+ },
+ "per_category": {
+  "di_knowledge": {
+   "n": 93,
+   "top1_agreement": 0.978494623655914,
+   "top1_flips": 2,
+   "max_abs_dscore": 0.10516834259033203,
+   "mean_abs_dscore": 0.01826677256272753,
+   "mean_kl": 0.0001961899290394059,
+   "n_gold": 93,
+   "nll_ref": 0.7907936240501159,
+   "nll_cand": 0.7950259330695674,
+   "dnll": 0.004232309019451486,
+   "mean_abs_row_dnll": 0.012335049730692368,
+   "acc_ref": 0.6559139784946236,
+   "acc_cand": 0.6559139784946236,
+   "acc_drop_pp": 0.0
+  },
+  "di_language": {
+   "n": 93,
+   "top1_agreement": 1.0,
+   "top1_flips": 0,
+   "max_abs_dscore": 0.12582671642303467,
+   "mean_abs_dscore": 0.019231186946019475,
+   "mean_kl": 0.0002764365455637818,
+   "n_gold": 93,
+   "nll_ref": 0.43828192127009435,
+   "nll_cand": 0.4350339331964042,
+   "dnll": -0.0032479880736901445,
+   "mean_abs_row_dnll": 0.01105854314657161,
+   "acc_ref": 0.8494623655913979,
+   "acc_cand": 0.8494623655913979,
+   "acc_drop_pp": 0.0
+  },
+  "di_retrieval": {
+   "n": 93,
+   "top1_agreement": 1.0,
+   "top1_flips": 0,
+   "max_abs_dscore": 0.16489863395690918,
+   "mean_abs_dscore": 0.013032539303739148,
+   "mean_kl": 9.17728737128296e-05,
+   "n_gold": 93,
+   "nll_ref": 0.07565367364179643,
+   "nll_cand": 0.07596664403414172,
+   "dnll": 0.0003129703923452909,
+   "mean_abs_row_dnll": 0.0027333661229381077,
+   "acc_ref": 0.978494623655914,
+   "acc_cand": 0.978494623655914,
+   "acc_drop_pp": 0.0
+  },
+  "format": {
+   "n": 75,
+   "top1_agreement": 1.0,
+   "top1_flips": 0,
+   "max_abs_dscore": 0.15371152758598328,
+   "mean_abs_dscore": 0.014854268128133183,
+   "mean_kl": 8.706661370645814e-05,
+   "n_gold": 75,
+   "nll_ref": 0.09087868406020969,
+   "nll_cand": 0.09241995785860305,
+   "dnll": 0.001541273798393361,
+   "mean_abs_row_dnll": 0.003729830247715702,
+   "acc_ref": 0.9733333333333334,
+   "acc_cand": 0.9733333333333334,
+   "acc_drop_pp": 0.0
+  },
+  "general": {
+   "n": 93,
+   "top1_agreement": 1.0,
+   "top1_flips": 0,
+   "max_abs_dscore": 0.1003103256225586,
+   "mean_abs_dscore": 0.01342117658100456,
+   "mean_kl": 7.21948109281326e-05,
+   "n_gold": 93,
+   "nll_ref": 0.13061020512422436,
+   "nll_cand": 0.12993521810236933,
+   "dnll": -0.0006749870218550336,
+   "mean_abs_row_dnll": 0.003786342376918514,
+   "acc_ref": 0.967741935483871,
+   "acc_cand": 0.967741935483871,
+   "acc_drop_pp": 0.0
+  },
+  "hard_short": {
+   "n": 93,
+   "top1_agreement": 0.989247311827957,
+   "top1_flips": 1,
+   "max_abs_dscore": 0.12274712324142456,
+   "mean_abs_dscore": 0.020271506775084727,
+   "mean_kl": 0.00027143138298962177,
+   "n_gold": 93,
+   "nll_ref": 0.7835153258109905,
+   "nll_cand": 0.7861608190184016,
+   "dnll": 0.0026454932074111426,
+   "mean_abs_row_dnll": 0.014292107382686168,
+   "acc_ref": 0.6021505376344086,
+   "acc_cand": 0.5913978494623656,
+   "acc_drop_pp": 1.0752688172043001
+  },
+  "long": {
+   "n": 92,
+   "top1_agreement": 1.0,
+   "top1_flips": 0,
+   "max_abs_dscore": 0.3184394836425781,
+   "mean_abs_dscore": 0.020006766643251753,
+   "mean_kl": 0.0007467326030115854,
+   "n_gold": 92,
+   "nll_ref": 0.10827285500173092,
+   "nll_cand": 0.10447898892584594,
+   "dnll": -0.0037938660758849857,
+   "mean_abs_row_dnll": 0.008173341485227314,
+   "acc_ref": 0.967391304347826,
+   "acc_cand": 0.967391304347826,
+   "acc_drop_pp": 0.0
+  },
+  "mac": {
+   "n": 92,
+   "top1_agreement": 1.0,
+   "top1_flips": 0,
+   "max_abs_dscore": 0.05875861644744873,
+   "mean_abs_dscore": 0.013405789823635765,
+   "mean_kl": 3.8362254891210755e-05,
+   "n_gold": 92,
+   "nll_ref": 0.3262944821095264,
+   "nll_cand": 0.3269720723076003,
+   "dnll": 0.0006775901980738963,
+   "mean_abs_row_dnll": 0.0040734288260641455,
+   "acc_ref": 0.8369565217391305,
+   "acc_cand": 0.8369565217391305,
+   "acc_drop_pp": 0.0
+  },
+  "public_ingests": {
+   "n": 92,
+   "top1_agreement": 0.9891304347826086,
+   "top1_flips": 1,
+   "max_abs_dscore": 0.146925687789917,
+   "mean_abs_dscore": 0.020489049231511028,
+   "mean_kl": 0.00030913724098080613,
+   "n_gold": 92,
+   "nll_ref": 0.7667467143475788,
+   "nll_cand": 0.7701994879203283,
+   "dnll": 0.0034527735727495346,
+   "mean_abs_row_dnll": 0.016463952376233926,
+   "acc_ref": 0.6521739130434783,
+   "acc_cand": 0.6630434782608695,
+   "acc_drop_pp": -1.0869565217391242
+  },
+  "themes": {
+   "n": 92,
+   "top1_agreement": 1.0,
+   "top1_flips": 0,
+   "max_abs_dscore": 0.05321455001831055,
+   "mean_abs_dscore": 0.012158645316958427,
+   "mean_kl": 2.255066396046878e-05,
+   "n_gold": 92,
+   "nll_ref": 0.27025191463643466,
+   "nll_cand": 0.27031868131249703,
+   "dnll": 6.676667606236864e-05,
+   "mean_abs_row_dnll": 0.003036639894596646,
+   "acc_ref": 0.8695652173913043,
+   "acc_cand": 0.8695652173913043,
+   "acc_drop_pp": 0.0
+  },
+  "typed_synthetic": {
+   "n": 92,
+   "top1_agreement": 1.0,
+   "top1_flips": 0,
+   "max_abs_dscore": 0.07491475343704224,
+   "mean_abs_dscore": 0.018002478546206502,
+   "mean_kl": 0.00021491486269545232,
+   "n_gold": 92,
+   "nll_ref": 1.0338530850662833,
+   "nll_cand": 1.0343635982006703,
+   "dnll": 0.0005105131343869918,
+   "mean_abs_row_dnll": 0.0132145397374374,
+   "acc_ref": 0.5652173913043478,
+   "acc_cand": 0.5652173913043478,
+   "acc_drop_pp": 0.0
+  }
+ },
+ "gates": {
+  "coverage_finite_tokens": {
+   "pass": true,
+   "problems": {},
+   "reference_missing_requests": 0
+  },
+  "top1": {
+   "threshold": 0.98,
+   "value": 0.996,
+   "pass": true
+  },
+  "dnll": {
+   "threshold": 0.02,
+   "value": 0.0005035682350758575,
+   "pass": true
+  },
+  "acc_drop_pp": {
+   "threshold": 0.3,
+   "value": 0.0,
+   "resolution_pp": 0.1,
+   "pass": true,
+   "note": null
+  }
+ },
+ "verdict": "PASS",
+ "resolution_pp_per_flip": 0.1,
+ "note": "small fixtures cannot resolve 0.3pp accuracy gates; accuracy on a parity fixture is only indicative (synthetic long rows carry the base row's gold label).",
+ "ref": "runs/macjev/runtime_v2_dev/reference_trained/gate_ref_block_fp32.jsonl",
+ "cand": "runs/macjev/release_2b/parity/pred_mlx_affine8-g64_gate.jsonl",
+ "fixture": "runs/macjev/runtime_v2_dev/reference/gate_fixture_n1000.jsonl"
+}

validation/parity/compare_mlx_bf16_base_vs_block.json

@@ -0,0 +1,269 @@
+{
+ "format": "bf16",
+ "temperature": 0.8278650620942867,
+ "overall": {
+  "n": 43,
+  "top1_agreement": 1.0,
+  "top1_flips": 0,
+  "max_abs_dscore": 0.08836114406585693,
+  "mean_abs_dscore": 0.01365492056503762,
+  "mean_kl": 9.883488035170103e-05,
+  "n_gold": 43,
+  "nll_ref": 0.670433307194272,
+  "nll_cand": 0.6677389810942516,
+  "dnll": -0.0026943261000204055,
+  "mean_abs_row_dnll": 0.009347213165255148,
+  "acc_ref": 0.7209302325581395,
+  "acc_cand": 0.7209302325581395,
+  "acc_drop_pp": 0.0
+ },
+ "per_category": {
+  "catalogue_overflow": {
+   "n": 3,
+   "top1_agreement": 1.0,
+   "top1_flips": 0,
+   "max_abs_dscore": 0.08836114406585693,
+   "mean_abs_dscore": 0.012576654652096578,
+   "mean_kl": 9.049678076068058e-05,
+   "n_gold": 3,
+   "nll_ref": 0.5023960827418966,
+   "nll_cand": 0.49688457645611867,
+   "dnll": -0.005511506285777956,
+   "mean_abs_row_dnll": 0.005740943453378988,
+   "acc_ref": 0.6666666666666666,
+   "acc_cand": 0.6666666666666666,
+   "acc_drop_pp": 0.0
+  },
+  "catalogue_overflow_long_state": {
+   "n": 1,
+   "top1_agreement": 1.0,
+   "top1_flips": 0,
+   "max_abs_dscore": 0.04151296615600586,
+   "mean_abs_dscore": 0.012695910914844235,
+   "mean_kl": 0.00019275121223053925,
+   "n_gold": 1,
+   "nll_ref": 1.868953922917252,
+   "nll_cand": 1.8621515972563385,
+   "dnll": -0.006802325660913544,
+   "mean_abs_row_dnll": 0.006802325660913544,
+   "acc_ref": 0.0,
+   "acc_cand": 0.0,
+   "acc_drop_pp": 0.0
+  },
+  "long_16k": {
+   "n": 3,
+   "top1_agreement": 1.0,
+   "top1_flips": 0,
+   "max_abs_dscore": 0.05213227868080139,
+   "mean_abs_dscore": 0.014879310828434184,
+   "mean_kl": 0.000149667886473795,
+   "n_gold": 3,
+   "nll_ref": 1.098996986024778,
+   "nll_cand": 1.0956083262917284,
+   "dnll": -0.0033886597330494705,
+   "mean_abs_row_dnll": 0.006144886758009865,
+   "acc_ref": 0.6666666666666666,
+   "acc_cand": 0.6666666666666666,
+   "acc_drop_pp": 0.0
+  },
+  "long_24k": {
+   "n": 2,
+   "top1_agreement": 1.0,
+   "top1_flips": 0,
+   "max_abs_dscore": 0.03363656997680664,
+   "mean_abs_dscore": 0.012938144306341806,
+   "mean_kl": 0.0001715685914165032,
+   "n_gold": 2,
+   "nll_ref": 0.43411292245601785,
+   "nll_cand": 0.42479629433910976,
+   "dnll": -0.009316628116908088,
+   "mean_abs_row_dnll": 0.00931662811690806,
+   "acc_ref": 1.0,
+   "acc_cand": 1.0,
+   "acc_drop_pp": 0.0
+  },
+  "many_options": {
+   "n": 2,
+   "top1_agreement": 1.0,
+   "top1_flips": 0,
+   "max_abs_dscore": 0.03784656524658203,
+   "mean_abs_dscore": 0.0131730338682731,
+   "mean_kl": 4.283448514828303e-06,
+   "n_gold": 2,
+   "nll_ref": 0.012124599411053285,
+   "nll_cand": 0.01233819862308103,
+   "dnll": 0.0002135992120277444,
+   "mean_abs_row_dnll": 0.00021359921202774614,
+   "acc_ref": 1.0,
+   "acc_cand": 1.0,
+   "acc_drop_pp": 0.0
+  },
+  "multi_block_3k": {
+   "n": 2,
+   "top1_agreement": 1.0,
+   "top1_flips": 0,
+   "max_abs_dscore": 0.053168416023254395,
+   "mean_abs_dscore": 0.011156398057937621,
+   "mean_kl": 0.00020998625581104233,
+   "n_gold": 2,
+   "nll_ref": 0.10264503770560551,
+   "nll_cand": 0.09699945875175084,
+   "dnll": -0.005645578953854674,
+   "mean_abs_row_dnll": 0.005792563090663002,
+   "acc_ref": 1.0,
+   "acc_cand": 1.0,
+   "acc_drop_pp": 0.0
+  },
+  "multi_block_5k": {
+   "n": 2,
+   "top1_agreement": 1.0,
+   "top1_flips": 0,
+   "max_abs_dscore": 0.018996506929397583,
+   "mean_abs_dscore": 0.010537676513195038,
+   "mean_kl": 7.5737981061909e-05,
+   "n_gold": 2,
+   "nll_ref": 0.41414880456154163,
+   "nll_cand": 0.40761999179423464,
+   "dnll": -0.006528812767306991,
+   "mean_abs_row_dnll": 0.0065288127673070045,
+   "acc_ref": 1.0,
+   "acc_cand": 1.0,
+   "acc_drop_pp": 0.0
+  },
+  "multi_block_9k": {
+   "n": 2,
+   "top1_agreement": 1.0,
+   "top1_flips": 0,
+   "max_abs_dscore": 0.033649444580078125,
+   "mean_abs_dscore": 0.00982309877872467,
+   "mean_kl": 8.772959532626077e-05,
+   "n_gold": 2,
+   "nll_ref": 1.387121106483142,
+   "nll_cand": 1.389387240331538,
+   "dnll": 0.0022661338483960236,
+   "mean_abs_row_dnll": 0.009883718382939666,
+   "acc_ref": 0.5,
+   "acc_cand": 0.5,
+   "acc_drop_pp": 0.0
+  },
+  "multi_question": {
+   "n": 10,
+   "top1_agreement": 1.0,
+   "top1_flips": 0,
+   "max_abs_dscore": 0.06624865531921387,
+   "mean_abs_dscore": 0.014526768724123637,
+   "mean_kl": 8.529135561399713e-05,
+   "n_gold": 10,
+   "nll_ref": 0.7606667410227165,
+   "nll_cand": 0.7570780864187747,
+   "dnll": -0.0035886546039417544,
+   "mean_abs_row_dnll": 0.010879836749203488,
+   "acc_ref": 0.7,
+   "acc_cand": 0.7,
+   "acc_drop_pp": 0.0
+  },
+  "prefix_boundary": {
+   "n": 7,
+   "top1_agreement": 1.0,
+   "top1_flips": 0,
+   "max_abs_dscore": 0.047842979431152344,
+   "mean_abs_dscore": 0.014937927183650791,
+   "mean_kl": 0.00011731551621384146,
+   "n_gold": 7,
+   "nll_ref": 0.9445710499909146,
+   "nll_cand": 0.9344505690212561,
+   "dnll": -0.010120480969658452,
+   "mean_abs_row_dnll": 0.01758201360584486,
+   "acc_ref": 0.5714285714285714,
+   "acc_cand": 0.5714285714285714,
+   "acc_drop_pp": 0.0
+  },
+  "qtype_choice": {
+   "n": 1,
+   "top1_agreement": 1.0,
+   "top1_flips": 0,
+   "max_abs_dscore": 0.02133503556251526,
+   "mean_abs_dscore": 0.013152092695236206,
+   "mean_kl": 8.789457820938037e-05,
+   "n_gold": 1,
+   "nll_ref": 1.2471588298811915,
+   "nll_cand": 1.267813737179954,
+   "dnll": 0.020654907298762515,
+   "mean_abs_row_dnll": 0.020654907298762515,
+   "acc_ref": 0.0,
+   "acc_cand": 0.0,
+   "acc_drop_pp": 0.0
+  },
+  "qtype_noul": {
+   "n": 1,
+   "top1_agreement": 1.0,
+   "top1_flips": 0,
+   "max_abs_dscore": 0.010687410831451416,
+   "mean_abs_dscore": 0.006474539637565613,
+   "mean_kl": 2.599908743399675e-05,
+   "n_gold": 1,
+   "nll_ref": 0.3643240477162341,
+   "nll_cand": 0.3691259380231449,
+   "dnll": 0.004801890306910805,
+   "mean_abs_row_dnll": 0.004801890306910805,
+   "acc_ref": 1.0,
+   "acc_cand": 1.0,
+   "acc_drop_pp": 0.0
+  },
+  "qtype_score": {
+   "n": 1,
+   "top1_agreement": 1.0,
+   "top1_flips": 0,
+   "max_abs_dscore": 0.03596752882003784,
+   "mean_abs_dscore": 0.013961112499237061,
+   "mean_kl": 0.0002653113090994117,
+   "n_gold": 1,
+   "nll_ref": 0.9393375234945696,
+   "nll_cand": 0.9660497441698873,
+   "dnll": 0.026712220675317755,
+   "mean_abs_row_dnll": 0.026712220675317755,
+   "acc_ref": 0.0,
+   "acc_cand": 0.0,
+   "acc_drop_pp": 0.0
+  },
+  "short_sb": {
+   "n": 6,
+   "top1_agreement": 1.0,
+   "top1_flips": 0,
+   "max_abs_dscore": 0.04204094409942627,
+   "mean_abs_dscore": 0.015570025255400981,
+   "mean_kl": 3.078595875807344e-05,
+   "n_gold": 6,
+   "nll_ref": 0.11428482960768895,
+   "nll_cand": 0.11598987452733013,
+   "dnll": 0.0017050449196411854,
+   "mean_abs_row_dnll": 0.0019930376095433936,
+   "acc_ref": 1.0,
+   "acc_cand": 1.0,
+   "acc_drop_pp": 0.0
+  }
+ },
+ "gates": {
+  "coverage_finite_tokens": {
+   "pass": true,
+   "problems": {},
+   "reference_missing_requests": 0
+  },
+  "top1": {
+   "threshold": 0.99,
+   "value": 1.0,
+   "pass": true
+  },
+  "dnll": {
+   "threshold": 0.02,
+   "value": -0.0026943261000204055,
+   "pass": true
+  }
+ },
+ "verdict": "PASS",
+ "resolution_pp_per_flip": 2.3255813953488373,
+ "note": "small fixtures cannot resolve 0.3pp accuracy gates; accuracy on a parity fixture is only indicative (synthetic long rows carry the base row's gold label).",
+ "ref": "runs/macjev/runtime_v2_dev/reference_trained/ref_block_fp32.jsonl",
+ "cand": "runs/macjev/release_2b/parity/pred_mlx_bf16_base.jsonl",
+ "fixture": "runs/macjev/runtime_v2_dev/reference_base/fixture.jsonl"
+}

validation/parity/compare_mlx_bf16_gate_vs_block.json

@@ -0,0 +1,221 @@
+{
+ "format": "bf16",
+ "temperature": 0.8278650620942867,
+ "overall": {
+  "n": 1000,
+  "top1_agreement": 0.997,
+  "top1_flips": 3,
+  "max_abs_dscore": 0.13961565494537354,
+  "mean_abs_dscore": 0.012822698334419146,
+  "mean_kl": 7.013363207503902e-05,
+  "n_gold": 1000,
+  "nll_ref": 0.4436679457518203,
+  "nll_cand": 0.4437944677940992,
+  "dnll": 0.0001265220422789204,
+  "mean_abs_row_dnll": 0.005641898326944814,
+  "acc_ref": 0.808,
+  "acc_cand": 0.805,
+  "acc_drop_pp": 0.30000000000000027
+ },
+ "per_category": {
+  "di_knowledge": {
+   "n": 93,
+   "top1_agreement": 1.0,
+   "top1_flips": 0,
+   "max_abs_dscore": 0.07215514779090881,
+   "mean_abs_dscore": 0.014889913475799058,
+   "mean_kl": 0.00013722336018850683,
+   "n_gold": 93,
+   "nll_ref": 0.7907936240501159,
+   "nll_cand": 0.7927041608827371,
+   "dnll": 0.0019105368326212124,
+   "mean_abs_row_dnll": 0.01039648465267476,
+   "acc_ref": 0.6559139784946236,
+   "acc_cand": 0.6559139784946236,
+   "acc_drop_pp": 0.0
+  },
+  "di_language": {
+   "n": 93,
+   "top1_agreement": 1.0,
+   "top1_flips": 0,
+   "max_abs_dscore": 0.0584871768951416,
+   "mean_abs_dscore": 0.013199496934170362,
+   "mean_kl": 4.376886766709004e-05,
+   "n_gold": 93,
+   "nll_ref": 0.43828192127009435,
+   "nll_cand": 0.43613660599650544,
+   "dnll": -0.0021453152735889103,
+   "mean_abs_row_dnll": 0.005769073926221885,
+   "acc_ref": 0.8494623655913979,
+   "acc_cand": 0.8494623655913979,
+   "acc_drop_pp": 0.0
+  },
+  "di_retrieval": {
+   "n": 93,
+   "top1_agreement": 1.0,
+   "top1_flips": 0,
+   "max_abs_dscore": 0.13961565494537354,
+   "mean_abs_dscore": 0.011278054842327665,
+   "mean_kl": 6.320183776936612e-05,
+   "n_gold": 93,
+   "nll_ref": 0.07565367364179643,
+   "nll_cand": 0.07529180302084762,
+   "dnll": -0.0003618706209488065,
+   "mean_abs_row_dnll": 0.0018865181948961214,
+   "acc_ref": 0.978494623655914,
+   "acc_cand": 0.978494623655914,
+   "acc_drop_pp": 0.0
+  },
+  "format": {
+   "n": 75,
+   "top1_agreement": 1.0,
+   "top1_flips": 0,
+   "max_abs_dscore": 0.07049310207366943,
+   "mean_abs_dscore": 0.012186950511526283,
+   "mean_kl": 1.866440251394625e-05,
+   "n_gold": 75,
+   "nll_ref": 0.09087868406020969,
+   "nll_cand": 0.09027726831006917,
+   "dnll": -0.0006014157501405132,
+   "mean_abs_row_dnll": 0.0017343322558420642,
+   "acc_ref": 0.9733333333333334,
+   "acc_cand": 0.9733333333333334,
+   "acc_drop_pp": 0.0
+  },
+  "general": {
+   "n": 93,
+   "top1_agreement": 1.0,
+   "top1_flips": 0,
+   "max_abs_dscore": 0.053872108459472656,
+   "mean_abs_dscore": 0.011199172480311083,
+   "mean_kl": 8.964417186979084e-06,
+   "n_gold": 93,
+   "nll_ref": 0.13061020512422436,
+   "nll_cand": 0.12994049456566417,
+   "dnll": -0.0006697105585601881,
+   "mean_abs_row_dnll": 0.0015611860129022736,
+   "acc_ref": 0.967741935483871,
+   "acc_cand": 0.967741935483871,
+   "acc_drop_pp": 0.0
+  },
+  "hard_short": {
+   "n": 93,
+   "top1_agreement": 0.989247311827957,
+   "top1_flips": 1,
+   "max_abs_dscore": 0.06802543997764587,
+   "mean_abs_dscore": 0.014637837174057167,
+   "mean_kl": 0.00013160361749015227,
+   "n_gold": 93,
+   "nll_ref": 0.7835153258109905,
+   "nll_cand": 0.782783474188176,
+   "dnll": -0.0007318516228145278,
+   "mean_abs_row_dnll": 0.01093632719811279,
+   "acc_ref": 0.6021505376344086,
+   "acc_cand": 0.5913978494623656,
+   "acc_drop_pp": 1.0752688172043001
+  },
+  "long": {
+   "n": 92,
+   "top1_agreement": 1.0,
+   "top1_flips": 0,
+   "max_abs_dscore": 0.1342533826828003,
+   "mean_abs_dscore": 0.013823732117665294,
+   "mean_kl": 6.168921301697641e-05,
+   "n_gold": 92,
+   "nll_ref": 0.10827285500173092,
+   "nll_cand": 0.10768749311257943,
+   "dnll": -0.000585361889151495,
+   "mean_abs_row_dnll": 0.002564256755867923,
+   "acc_ref": 0.967391304347826,
+   "acc_cand": 0.967391304347826,
+   "acc_drop_pp": 0.0
+  },
+  "mac": {
+   "n": 92,
+   "top1_agreement": 1.0,
+   "top1_flips": 0,
+   "max_abs_dscore": 0.04618799686431885,
+   "mean_abs_dscore": 0.01143809092109618,
+   "mean_kl": 2.6581458539860547e-05,
+   "n_gold": 92,
+   "nll_ref": 0.3262944821095264,
+   "nll_cand": 0.3269180887247012,
+   "dnll": 0.0006236066151747988,
+   "mean_abs_row_dnll": 0.003772370527716574,
+   "acc_ref": 0.8369565217391305,
+   "acc_cand": 0.8369565217391305,
+   "acc_drop_pp": 0.0
+  },
+  "public_ingests": {
+   "n": 92,
+   "top1_agreement": 1.0,
+   "top1_flips": 0,
+   "max_abs_dscore": 0.10585364699363708,
+   "mean_abs_dscore": 0.015097456673118565,
+   "mean_kl": 0.00015103258358910773,
+   "n_gold": 92,
+   "nll_ref": 0.7667467143475788,
+   "nll_cand": 0.7682457652594957,
+   "dnll": 0.0014990509119169326,
+   "mean_abs_row_dnll": 0.010491535553020102,
+   "acc_ref": 0.6521739130434783,
+   "acc_cand": 0.6521739130434783,
+   "acc_drop_pp": 0.0
+  },
+  "themes": {
+   "n": 92,
+   "top1_agreement": 0.9891304347826086,
+   "top1_flips": 1,
+   "max_abs_dscore": 0.04355478286743164,
+   "mean_abs_dscore": 0.010605669458923132,
+   "mean_kl": 2.4907957260022938e-05,
+   "n_gold": 92,
+   "nll_ref": 0.27025191463643466,
+   "nll_cand": 0.2710970978349834,
+   "dnll": 0.0008451831985487601,
+   "mean_abs_row_dnll": 0.0033373064356656897,
+   "acc_ref": 0.8695652173913043,
+   "acc_cand": 0.8586956521739131,
+   "acc_drop_pp": 1.0869565217391242
+  },
+  "typed_synthetic": {
+   "n": 92,
+   "top1_agreement": 0.9891304347826086,
+   "top1_flips": 1,
+   "max_abs_dscore": 0.04864954948425293,
+   "mean_abs_dscore": 0.01256397343500986,
+   "mean_kl": 9.395103279401365e-05,
+   "n_gold": 92,
+   "nll_ref": 1.0338530850662833,
+   "nll_cand": 1.0353560613294195,
+   "dnll": 0.0015029762631362242,
+   "mean_abs_row_dnll": 0.008864003979572451,
+   "acc_ref": 0.5652173913043478,
+   "acc_cand": 0.5543478260869565,
+   "acc_drop_pp": 1.0869565217391242
+  }
+ },
+ "gates": {
+  "coverage_finite_tokens": {
+   "pass": true,
+   "problems": {},
+   "reference_missing_requests": 0
+  },
+  "top1": {
+   "threshold": 0.99,
+   "value": 0.997,
+   "pass": true
+  },
+  "dnll": {
+   "threshold": 0.02,
+   "value": 0.0001265220422789204,
+   "pass": true
+  }
+ },
+ "verdict": "PASS",
+ "resolution_pp_per_flip": 0.1,
+ "note": "small fixtures cannot resolve 0.3pp accuracy gates; accuracy on a parity fixture is only indicative (synthetic long rows carry the base row's gold label).",
+ "ref": "runs/macjev/runtime_v2_dev/reference_trained/gate_ref_block_fp32.jsonl",
+ "cand": "runs/macjev/release_2b/parity/pred_mlx_bf16_gate.jsonl",
+ "fixture": "runs/macjev/runtime_v2_dev/reference/gate_fixture_n1000.jsonl"
+}

validation/parity/cross_format_dp.json

@@ -0,0 +1,68 @@
+{
+ "base:torch_fp32_cpu": {
+  "n": 43,
+  "top1_agree": 1.0,
+  "max_abs_dp": 4.782633115096857e-06,
+  "mean_max_abs_dp": 6.972375795818997e-07
+ },
+ "base:gguf_f16": {
+  "n": 43,
+  "top1_agree": 1.0,
+  "max_abs_dp": 0.0006297677378335198,
+  "mean_max_abs_dp": 0.00017737997866918594
+ },
+ "base:gguf_q8_0": {
+  "n": 43,
+  "top1_agree": 1.0,
+  "max_abs_dp": 0.021289327356281362,
+  "mean_max_abs_dp": 0.0044768677260933745
+ },
+ "base:gguf_q4_k_m": {
+  "n": 43,
+  "top1_agree": 0.9767441860465116,
+  "max_abs_dp": 0.15778925110927658,
+  "mean_max_abs_dp": 0.047022506153486646
+ },
+ "base:mlx_bf16": {
+  "n": 43,
+  "top1_agree": 1.0,
+  "max_abs_dp": 0.010765431767972844,
+  "mean_max_abs_dp": 0.003890125022245109
+ },
+ "base:mlx_8bit": {
+  "n": 43,
+  "top1_agree": 1.0,
+  "max_abs_dp": 0.04628699142360499,
+  "mean_max_abs_dp": 0.007026009064181663
+ },
+ "gate:gguf_f16": {
+  "n": 1000,
+  "top1_agree": 1.0,
+  "max_abs_dp": 0.0013728371250730786,
+  "mean_max_abs_dp": 0.00012138780447113503
+ },
+ "gate:gguf_q8_0": {
+  "n": 1000,
+  "top1_agree": 0.997,
+  "max_abs_dp": 0.033479594816581415,
+  "mean_max_abs_dp": 0.0020247272188215755
+ },
+ "gate:gguf_q4_k_m": {
+  "n": 1000,
+  "top1_agree": 0.957,
+  "max_abs_dp": 0.34599917206983777,
+  "mean_max_abs_dp": 0.02317308476687987
+ },
+ "gate:mlx_bf16": {
+  "n": 1000,
+  "top1_agree": 0.997,
+  "max_abs_dp": 0.03549758339084519,
+  "mean_max_abs_dp": 0.0025865935031305484
+ },
+ "gate:mlx_8bit": {
+  "n": 1000,
+  "top1_agree": 0.996,
+  "max_abs_dp": 0.1623697296878569,
+  "mean_max_abs_dp": 0.0039011049015487734
+ }
+}

validation/runtime/fixture_verify_8bit.json

@@ -0,0 +1,319 @@
+{
+ "precision": "8bit",
+ "dev_seconds": 1324.8,
+ "default_T": 0.8278650620942867,
+ "fixture_requests": 35,
+ "fixture_questions": 43,
+ "fixture_tokens": 214700,
+ "overflow_q": 5,
+ "max_abs_dscore": 0.0,
+ "max_abs_dprob_vs_softmax_dev_over_T": 0.0,
+ "text_e2e_max_abs_dscore_vs_dev": 0.0,
+ "text_ids_equal_dev": true,
+ "score_many_vs_separate_fresh_decide_max_abs": 0.0,
+ "order_invariance_max_abs": 0.0,
+ "runtime_seconds": 286.5,
+ "per_question": [
+  [
+   "multi_question-0",
+   0,
+   450,
+   5,
+   0.0
+  ],
+  [
+   "multi_question-0",
+   1,
+   364,
+   4,
+   0.0
+  ],
+  [
+   "multi_question-0",
+   2,
+   356,
+   2,
+   0.0
+  ],
+  [
+   "multi_question-0",
+   3,
+   345,
+   4,
+   0.0
+  ],
+  [
+   "multi_question-0",
+   4,
+   319,
+   2,
+   0.0
+  ],
+  [
+   "multi_question-1",
+   0,
+   5416,
+   2,
+   0.0
+  ],
+  [
+   "multi_question-1",
+   1,
+   5471,
+   2,
+   0.0
+  ],
+  [
+   "multi_question-1",
+   2,
+   5425,
+   2,
+   0.0
+  ],
+  [
+   "multi_question-1",
+   3,
+   5517,
+   6,
+   0.0
+  ],
+  [
+   "multi_question-1",
+   4,
+   5428,
+   2,
+   0.0
+  ],
+  [
+   "short_sb-themes-0",
+   0,
+   105,
+   2,
+   0.0
+  ],
+  [
+   "short_sb-public_ingests-0",
+   0,
+   382,
+   16,
+   0.0
+  ],
+  [
+   "short_sb-di_retrieval-0",
+   0,
+   197,
+   2,
+   0.0
+  ],
+  [
+   "short_sb-di_knowledge-0",
+   0,
+   72,
+   4,
+   0.0
+  ],
+  [
+   "short_sb-hard_short-0",
+   0,
+   517,
+   3,
+   0.0
+  ],
+  [
+   "short_sb-format-0",
+   0,
+   166,
+   21,
+   0.0
+  ],
+  [
+   "qtype_choice-0",
+   0,
+   594,
+   3,
+   0.0
+  ],
+  [
+   "qtype_noul-0",
+   0,
+   246,
+   2,
+   0.0
+  ],
+  [
+   "qtype_score-0",
+   0,
+   873,
+   5,
+   0.0
+  ],
+  [
+   "multi_block_3k-0",
+   0,
+   2927,
+   5,
+   0.0
+  ],
+  [
+   "multi_block_3k-1",
+   0,
+   2932,
+   2,
+   0.0
+  ],
+  [
+   "multi_block_5k-0",
+   0,
+   5011,
+   5,
+   0.0
+  ],
+  [
+   "multi_block_5k-1",
+   0,
+   4767,
+   2,
+   0.0
+  ],
+  [
+   "multi_block_9k-s0",
+   0,
+   9083,
+   3,
+   0.0
+  ],
+  [
+   "multi_block_9k-s1",
+   0,
+   9568,
+   2,
+   0.0
+  ],
+  [
+   "prefix_boundary-2047",
+   0,
+   2178,
+   2,
+   0.0
+  ],
+  [
+   "prefix_boundary-2048",
+   0,
+   2203,
+   3,
+   0.0
+  ],
+  [
+   "prefix_boundary-2049",
+   0,
+   2289,
+   6,
+   0.0
+  ],
+  [
+   "prefix_boundary-4095",
+   0,
+   4163,
+   4,
+   0.0
+  ],
+  [
+   "prefix_boundary-4096",
+   0,
+   4312,
+   5,
+   0.0
+  ],
+  [
+   "prefix_boundary-4097",
+   0,
+   4180,
+   2,
+   0.0
+  ],
+  [
+   "prefix_boundary-6144",
+   0,
+   6259,
+   3,
+   0.0
+  ],
+  [
+   "catalogue_overflow-0",
+   0,
+   4722,
+   151,
+   0.0
+  ],
+  [
+   "catalogue_overflow-1",
+   0,
+   4721,
+   151,
+   0.0
+  ],
+  [
+   "catalogue_overflow-2",
+   0,
+   4729,
+   151,
+   0.0
+  ],
+  [
+   "catalogue_overflow_long_state-0",
+   0,
+   8806,
+   151,
+   0.0
+  ],
+  [
+   "many_options-0",
+   0,
+   500,
+   75,
+   0.0
+  ],
+  [
+   "many_options-1",
+   0,
+   423,
+   64,
+   0.0
+  ],
+  [
+   "long_16k-0",
+   0,
+   15900,
+   6,
+   0.0
+  ],
+  [
+   "long_16k-1",
+   0,
+   16384,
+   2,
+   0.0
+  ],
+  [
+   "long_16k-2",
+   0,
+   16800,
+   151,
+   0.0
+  ],
+  [
+   "long_24k-0",
+   0,
+   24000,
+   2,
+   0.0
+  ],
+  [
+   "long_24k-1",
+   0,
+   25600,
+   6,
+   0.0
+  ]
+ ]
+}

validation/runtime/fixture_verify_bf16.json

@@ -0,0 +1,319 @@
+{
+ "precision": "bf16",
+ "dev_seconds": 415.6,
+ "default_T": 0.8278650620942867,
+ "fixture_requests": 35,
+ "fixture_questions": 43,
+ "fixture_tokens": 214700,
+ "overflow_q": 5,
+ "max_abs_dscore": 0.0,
+ "max_abs_dprob_vs_softmax_dev_over_T": 0.0,
+ "text_e2e_max_abs_dscore_vs_dev": 0.0,
+ "text_ids_equal_dev": true,
+ "score_many_vs_separate_fresh_decide_max_abs": 0.0,
+ "order_invariance_max_abs": 0.0,
+ "runtime_seconds": 293.1,
+ "per_question": [
+  [
+   "multi_question-0",
+   0,
+   450,
+   5,
+   0.0
+  ],
+  [
+   "multi_question-0",
+   1,
+   364,
+   4,
+   0.0
+  ],
+  [
+   "multi_question-0",
+   2,
+   356,
+   2,
+   0.0
+  ],
+  [
+   "multi_question-0",
+   3,
+   345,
+   4,
+   0.0
+  ],
+  [
+   "multi_question-0",
+   4,
+   319,
+   2,
+   0.0
+  ],
+  [
+   "multi_question-1",
+   0,
+   5416,
+   2,
+   0.0
+  ],
+  [
+   "multi_question-1",
+   1,
+   5471,
+   2,
+   0.0
+  ],
+  [
+   "multi_question-1",
+   2,
+   5425,
+   2,
+   0.0
+  ],
+  [
+   "multi_question-1",
+   3,
+   5517,
+   6,
+   0.0
+  ],
+  [
+   "multi_question-1",
+   4,
+   5428,
+   2,
+   0.0
+  ],
+  [
+   "short_sb-themes-0",
+   0,
+   105,
+   2,
+   0.0
+  ],
+  [
+   "short_sb-public_ingests-0",
+   0,
+   382,
+   16,
+   0.0
+  ],
+  [
+   "short_sb-di_retrieval-0",
+   0,
+   197,
+   2,
+   0.0
+  ],
+  [
+   "short_sb-di_knowledge-0",
+   0,
+   72,
+   4,
+   0.0
+  ],
+  [
+   "short_sb-hard_short-0",
+   0,
+   517,
+   3,
+   0.0
+  ],
+  [
+   "short_sb-format-0",
+   0,
+   166,
+   21,
+   0.0
+  ],
+  [
+   "qtype_choice-0",
+   0,
+   594,
+   3,
+   0.0
+  ],
+  [
+   "qtype_noul-0",
+   0,
+   246,
+   2,
+   0.0
+  ],
+  [
+   "qtype_score-0",
+   0,
+   873,
+   5,
+   0.0
+  ],
+  [
+   "multi_block_3k-0",
+   0,
+   2927,
+   5,
+   0.0
+  ],
+  [
+   "multi_block_3k-1",
+   0,
+   2932,
+   2,
+   0.0
+  ],
+  [
+   "multi_block_5k-0",
+   0,
+   5011,
+   5,
+   0.0
+  ],
+  [
+   "multi_block_5k-1",
+   0,
+   4767,
+   2,
+   0.0
+  ],
+  [
+   "multi_block_9k-s0",
+   0,
+   9083,
+   3,
+   0.0
+  ],
+  [
+   "multi_block_9k-s1",
+   0,
+   9568,
+   2,
+   0.0
+  ],
+  [
+   "prefix_boundary-2047",
+   0,
+   2178,
+   2,
+   0.0
+  ],
+  [
+   "prefix_boundary-2048",
+   0,
+   2203,
+   3,
+   0.0
+  ],
+  [
+   "prefix_boundary-2049",
+   0,
+   2289,
+   6,
+   0.0
+  ],
+  [
+   "prefix_boundary-4095",
+   0,
+   4163,
+   4,
+   0.0
+  ],
+  [
+   "prefix_boundary-4096",
+   0,
+   4312,
+   5,
+   0.0
+  ],
+  [
+   "prefix_boundary-4097",
+   0,
+   4180,
+   2,
+   0.0
+  ],
+  [
+   "prefix_boundary-6144",
+   0,
+   6259,
+   3,
+   0.0
+  ],
+  [
+   "catalogue_overflow-0",
+   0,
+   4722,
+   151,
+   0.0
+  ],
+  [
+   "catalogue_overflow-1",
+   0,
+   4721,
+   151,
+   0.0
+  ],
+  [
+   "catalogue_overflow-2",
+   0,
+   4729,
+   151,
+   0.0
+  ],
+  [
+   "catalogue_overflow_long_state-0",
+   0,
+   8806,
+   151,
+   0.0
+  ],
+  [
+   "many_options-0",
+   0,
+   500,
+   75,
+   0.0
+  ],
+  [
+   "many_options-1",
+   0,
+   423,
+   64,
+   0.0
+  ],
+  [
+   "long_16k-0",
+   0,
+   15900,
+   6,
+   0.0
+  ],
+  [
+   "long_16k-1",
+   0,
+   16384,
+   2,
+   0.0
+  ],
+  [
+   "long_16k-2",
+   0,
+   16800,
+   151,
+   0.0
+  ],
+  [
+   "long_24k-0",
+   0,
+   24000,
+   2,
+   0.0
+  ],
+  [
+   "long_24k-1",
+   0,
+   25600,
+   6,
+   0.0
+  ]
+ ]
+}

validation/runtime/render_verify.json

@@ -0,0 +1,88 @@
+{
+ "requests": 463,
+ "questions": 756,
+ "ok_identical": 743,
+ "budget_both": 6,
+ "error_both": 6,
+ "mismatch": [
+  {
+   "q_repr": "{'t': 'score', 'ins': '\\n', 'crit': [' <tool_response>', '\\n<tool_response>->', '<|box_start|>\\nOptions:\\nOption 2: ', '<tool_response> -> yesabc', '<|endoftext|><|endoftext|>Option 1 ->', '', '<|video_pad|>\ud83d\ude42Option 2: ', '<tool_call>Judge each numbered option in the complete catalogue above:\\n<think",
+   "state_repr": "'abc<|im_start|><|vision_pad|>Judge each numbered option in the complete catalogue above:\\n<think>\\n",
+   "id": "special:16",
+   "q": 0,
+   "ref": [
+    "ok",
+    245
+   ],
+   "rt": "error",
+   "ref_err": null,
+   "rt_err": [
+    "QuestionError"
+   ]
+  }
+ ],
+ "overflow_q": 17,
+ "max_tokens": 25600,
+ "max_k": 1000,
+ "prefix_path_mismatch": 0,
+ "boundary_head_lens": [
+  2046,
+  2047,
+  2048,
+  2095,
+  2096
+ ],
+ "state_prefix_lens": [
+  2047,
+  2048,
+  2049,
+  4095,
+  4096,
+  4097,
+  20479,
+  20480,
+  20481
+ ],
+ "overflow_budget_edge_total": 25600,
+ "budget_ids": [
+  "total_edge:18657",
+  "total_edge:18658",
+  "bigk:1500",
+  "maxlen:64",
+  "maxlen:300",
+  "huge_question"
+ ],
+ "error_ids": [
+  [
+   "ctrl:1",
+   "TypeError",
+   "TypeError"
+  ],
+  [
+   "score_11",
+   "RecordError",
+   "QuestionError"
+  ],
+  [
+   "score_1",
+   "RecordError",
+   "QuestionError"
+  ],
+  [
+   "choice_empty",
+   "RecordError",
+   "QuestionError"
+  ],
+  [
+   "choice_numkeys",
+   "ValueError",
+   "TypeError"
+  ],
+  [
+   "bad_type",
+   "RecordError",
+   "QuestionError"
+  ]
+ ],
+ "n_mismatch": 1
+}

validation/runtime/reverify_mlx.json

@@ -0,0 +1,23 @@
+{
+  "runtime_file": "jev_style_decision_mlx.py",
+  "runtime_sha256": "025daab9b374d740e743c519bd3aa3385791e08c81f4515b2ae525af57788030",
+  "shared_core_sha256": "56deae095206fc59b9a30d8626ac5d27cacddcd94d7c7f08f6f4bc53d326a002",
+  "what": "MLX bf16 (loaded from the repository root, precision bf16), runtime defaults",
+  "loaded_from": "runs/macjev/hf_staging/Jev-Style-2B-Decision-v3-MLX",
+  "verify_manifest": true,
+  "compared_with": {
+    "file": "runs/macjev/release_2b/parity/pred_mlx_bf16_base.jsonl",
+    "sha256": "c42232870ab8c5410b4f45dc85729249d467b9c0bd5d24213d085e4479a3a60e"
+  },
+  "fixture": {
+    "file": "runs/macjev/runtime_v2_dev/reference_base/fixture.jsonl",
+    "sha256": "2b16d25eafa70c00a1106ba51b94890821b51ecb4fa393abb12403c8739ef314"
+  },
+  "questions": 43,
+  "skipped_questions": 0,
+  "max_abs_score_diff": 0.0,
+  "top1_same": "43/43",
+  "token_or_overflow_mismatches": [],
+  "seconds": 137.2,
+  "finished_unix": 1790449492.3913028
+}