Upload KETI Llama 7B v0.1

.gitattributes

@@ -33,3 +33,4 @@ 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
+tokenizer.json filter=lfs diff=lfs merge=lfs -text

README.md

@@ -0,0 +1,127 @@
+---
+license: other
+language:
+- en
+- ko
+library_name: transformers
+pipeline_tag: text-generation
+tags:
+- llama
+- causal-lm
+- sft
+- dpo
+- grpo
+- long-context
+- keti
+---
+
+# KETI Llama 7B v0.1
+
+`KETI Llama 7B v0.1` is a long-context causal language model released under the
+Hugging Face repository `KETI-AIR/keti-llama-7b-v0.1`.
+
+This checkpoint was produced from the local base
+model with the following training pipeline:
+
+1. SFT on instruction data with 32K sequence packing.
+2. DPO on preference data.
+3. RL/GRPO on RL data.
+
+The uploaded artifact is the merged Hugging Face model from:
+
+```text
+outputs/llama-8b-keti-dpo-rl-merged
+```
+
+## Model Details
+
+- Architecture: `LlamaForCausalLM`
+- Parameters: 8B-class
+- Context length in config: 131,072 tokens
+- Hidden size: 4096
+- Layers: 32
+- Attention heads: 32
+- KV heads: 8
+- Vocabulary size: 128,256
+- Recommended dtype: `bfloat16`
+
+## Evaluation
+
+Evaluation timestamp: `20260604_202553`
+
+| Category | Dataset | Version | Metric | Mode | Score |
+| --- | --- | --- | --- | --- | ---: |
+| Core | core_average | - | naive_average | gen | 27.77 |
+| Instruction Following | IFEval | 353ae7 | Prompt-level-strict-accuracy | gen | 50.65 |
+| Math Calculation | aime2024 | bc6078 | accuracy | gen | 16.67 |
+| Math Calculation | aime2025 | 5e9f4f | accuracy | gen | 3.33 |
+| Math Calculation | math_prm800k_500 | 11c4b5 | accuracy | gen | 60.20 |
+| General Reasoning | bbh | - | naive_average | gen | 11.87 |
+| General Reasoning | GPQA_diamond | 5aeece | accuracy | gen | 20.71 |
+| Knowledge | mmlu_pro | - | naive_average | gen | 28.26 |
+| Code | openai_humaneval | dcae0e | humaneval_pass@1 | gen | 60.98 |
+| Code | lcb_code_generation | b5b6c5 | pass@1 | gen | 6.00 |
+| Long Context Reasoning | leval | - | naive_average | gen | 39.37 |
+| Long Context Reasoning | longbench | - | naive_average | gen | 20.57 |
+| Long Context Reasoning | LongBenchv2 | 75fbba | accuracy | gen | 24.85 |
+| Long Context Reasoning | keti_long_ctx_gutenberg | - | naive_average | gen | 17.62 |
+
+
+## Quick Start
+
+```python
+import torch
+from transformers import AutoModelForCausalLM, AutoTokenizer
+
+model_id = "KETI-AIR/keti-llama-7b-v0.1"
+
+tokenizer = AutoTokenizer.from_pretrained(model_id, trust_remote_code=True)
+model = AutoModelForCausalLM.from_pretrained(
+    model_id,
+    torch_dtype=torch.bfloat16,
+    device_map="auto",
+    trust_remote_code=True,
+)
+
+messages = [
+    {"role": "user", "content": "Explain why long-context reasoning is useful."}
+]
+inputs = tokenizer.apply_chat_template(
+    messages,
+    add_generation_prompt=True,
+    return_tensors="pt",
+).to(model.device)
+
+outputs = model.generate(
+    inputs,
+    max_new_tokens=512,
+    do_sample=True,
+    temperature=0.7,
+    top_p=0.9,
+)
+print(tokenizer.decode(outputs[0][inputs.shape[-1]:], skip_special_tokens=True))
+```
+
+## Intended Use
+
+This model is intended for research and development on instruction following,
+code generation, mathematical reasoning, and long-context generation tasks.
+
+## Limitations
+
+The model can generate incorrect, unsafe, or biased content. Users should
+evaluate the model for their own deployment setting and apply appropriate safety
+filters and human review where needed.
+
+## Training Framework
+
+- Transformers: 5.8.1
+- PyTorch: 2.11.0+cu130
+- Datasets: 4.8.5
+- Tokenizers: 0.22.2
+- TRL: 1.4.0
+
+## Citation
+
+If you use this model, please cite the corresponding KETI-AIR release and the
+training/evaluation resources used in your work.

chat_template.jinja

@@ -0,0 +1,250 @@
+{{- bos_token if bos_token is defined else '' }}
+{%- if custom_tools is defined and custom_tools is not none %}
+    {%- set tools = custom_tools %}
+{%- endif %}
+{%- if tools is not defined %}
+    {%- set tools = none %}
+{%- endif %}
+{%- if tools is not none and tools is iterable and tools is not mapping and tools|length == 0 %}
+    {%- set tools = none %}
+{%- endif %}
+{%- if tools_in_user_message is not defined %}
+    {#- Llama 3.1-style tool prompting is most stable when tools are injected into the first user turn. #}
+    {%- set tools_in_user_message = true %}
+{%- endif %}
+{%- if date_string is not defined %}
+    {%- if strftime_now is defined %}
+        {%- set date_string = strftime_now("%d %b %Y") %}
+    {%- else %}
+        {#- Avoid baking a stale Today Date into newer data. Pass date_string explicitly if needed. #}
+        {%- set date_string = none %}
+    {%- endif %}
+{%- endif %}
+{%- if cutoff_date_string is not defined %}
+    {%- set cutoff_date_string = "December 2023" %}
+{%- endif %}
+{%- if image_token is not defined %}
+    {%- set image_token = "<|image|>" %}
+{%- endif %}
+{%- if video_token is not defined %}
+    {%- set video_token = "<|video|>" %}
+{%- endif %}
+
+{#- Modern content renderer inspired by Qwen-style templates.
+    Supports:
+    - string content
+    - None / missing content
+    - list content blocks: {type: text, text: ...}, {type: image/image_url/input_image, ...}, {type: video/input_video, ...}
+    - direct {text: ...}, {image: ...}, {image_url: ...}, {video: ...}
+#}
+{%- set image_count = namespace(value=0) %}
+{%- set video_count = namespace(value=0) %}
+
+{%- macro render_content_item(item, do_vision_count, is_system_content=false) %}
+    {%- if item is string %}
+        {{- item }}
+    {%- elif item is mapping %}
+        {%- set item_type = item.get('type', '') %}
+        {%- if item_type in ['image', 'image_url', 'input_image'] or 'image' in item or 'image_url' in item %}
+            {%- 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 is defined and add_vision_id %}
+                {{- 'Picture ' ~ image_count.value ~ ': ' }}
+            {%- endif %}
+            {{- image_token }}
+        {%- elif item_type in ['video', 'input_video'] or 'video' in item %}
+            {%- 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 is defined and add_vision_id %}
+                {{- 'Video ' ~ video_count.value ~ ': ' }}
+            {%- endif %}
+            {{- video_token }}
+        {%- elif 'text' in item %}
+            {{- item.text }}
+        {%- elif item_type in ['text', 'input_text'] and 'content' in item %}
+            {{- item.content }}
+        {%- elif 'content' in item %}
+            {{- render_content(item.content, do_vision_count, is_system_content) }}
+        {%- else %}
+            {{- raise_exception('Unexpected item type in content.') }}
+        {%- endif %}
+    {%- elif item is none or item is undefined %}
+        {{- '' }}
+    {%- else %}
+        {{- raise_exception('Unexpected item type in content.') }}
+    {%- endif %}
+{%- endmacro %}
+
+{%- macro render_content(content, do_vision_count, is_system_content=false) %}
+    {%- if content is string %}
+        {{- content }}
+    {%- elif content is none or content is undefined %}
+        {{- '' }}
+    {%- elif content is mapping %}
+        {{- render_content_item(content, do_vision_count, is_system_content) }}
+    {%- elif content is iterable %}
+        {%- for item in content %}
+            {{- render_content_item(item, do_vision_count, is_system_content) }}
+        {%- endfor %}
+    {%- else %}
+        {{- raise_exception('Unexpected content type.') }}
+    {%- endif %}
+{%- endmacro %}
+
+{%- macro render_tool_call_json(tool_call) %}
+    {%- if tool_call.function is defined %}
+        {%- set fn = tool_call.function %}
+    {%- else %}
+        {%- set fn = tool_call %}
+    {%- endif %}
+    {%- if fn.name is not defined %}
+        {{- raise_exception('Tool call is missing function name.') }}
+    {%- endif %}
+    {%- set args = fn.arguments if fn.arguments is defined and fn.arguments is not none else {} %}
+    {{- '{"name": ' }}{{- fn.name | tojson }}{{- ', "parameters": ' }}
+    {%- if args is string and args|trim %}
+        {{- args }}
+    {%- else %}
+        {{- args | tojson }}
+    {%- endif %}
+    {{- '}' }}
+{%- endmacro %}
+
+{%- macro render_tool_calls_json(tool_calls) %}
+    {%- if tool_calls|length == 1 %}
+        {{- render_tool_call_json(tool_calls[0]) }}
+    {%- else %}
+        {{- '[' }}
+        {%- for tool_call in tool_calls %}
+            {{- render_tool_call_json(tool_call) }}
+            {%- if not loop.last %}
+                {{- ', ' }}
+            {%- endif %}
+        {%- endfor %}
+        {{- ']' }}
+    {%- endif %}
+{%- endmacro %}
+
+{%- if not messages %}
+    {{- raise_exception('No messages provided.') }}
+{%- endif %}
+
+{#- Extract an initial system/developer message, so we can slot it into the Llama system header. #}
+{%- if messages[0]['role'] == 'system' or messages[0]['role'] == 'developer' %}
+    {%- set system_message = render_content(messages[0]['content'], false, true)|trim %}
+    {%- set messages = messages[1:] %}
+{%- else %}
+    {%- if tools is not none %}
+        {%- set system_message = "You are a helpful assistant with tool calling capabilities. Only reply with a tool call if the function exists in the library provided by the user. If it doesn't exist, just reply directly in natural language. When you receive a tool call response, use the output to format an answer to the original user question." %}
+    {%- else %}
+        {%- set system_message = "" %}
+    {%- endif %}
+{%- endif %}
+
+{#- System message #}
+{{- "<|start_header_id|>system<|end_header_id|>\n\n" }}
+{%- if tools is not none %}
+    {{- "Environment: ipython\n" }}
+{%- endif %}
+{%- if cutoff_date_string %}
+    {{- "Cutting Knowledge Date: " + cutoff_date_string + "\n" }}
+{%- endif %}
+{%- if date_string %}
+    {{- "Today Date: " + date_string + "\n" }}
+{%- endif %}
+{%- if tools is not none or cutoff_date_string or date_string %}
+    {{- "\n" }}
+{%- endif %}
+{%- if tools is not none and not tools_in_user_message %}
+    {{- "You have access to the following functions. To call a function, please respond with JSON for a function call. " }}
+    {{- 'For a single function call, respond in the format {"name": function name, "parameters": dictionary of argument name and its value}. ' }}
+    {{- 'If multiple function calls are required, respond with a JSON array of objects in that same format. ' }}
+    {{- "Do not use variables.\n\n" }}
+    {%- for t in tools %}
+        {{- t | tojson(indent=4) }}
+        {{- "\n\n" }}
+    {%- endfor %}
+{%- endif %}
+{{- system_message }}
+{{- "<|eot_id|>" }}
+
+{#- Custom tools can be passed in the first user message with extra guidance. #}
+{%- if tools_in_user_message and tools is not none %}
+    {%- if messages | length != 0 %}
+        {%- set first_user_message = render_content(messages[0]['content'], true)|trim %}
+        {%- set messages = messages[1:] %}
+    {%- else %}
+        {{- raise_exception("Cannot put tools in the first user message when there's no first user message!") }}
+    {%- endif %}
+    {{- '<|start_header_id|>user<|end_header_id|>\n\n' -}}
+    {{- "Given the following functions, please respond with JSON for a function call " }}
+    {{- "with its proper arguments that best answers the given prompt.\n\n" }}
+    {{- 'For a single function call, respond in the format {"name": function name, "parameters": dictionary of argument name and its value}. ' }}
+    {{- 'If multiple function calls are required, respond with a JSON array of objects in that same format. ' }}
+    {{- "Do not use variables.\n\n" }}
+    {%- for t in tools %}
+        {{- t | tojson(indent=4) }}
+        {{- "\n\n" }}
+    {%- endfor %}
+    {{- first_user_message + "<|eot_id|>" }}
+{%- endif %}
+
+{%- for message in messages %}
+    {%- set role = message.role %}
+    {%- if role == 'system' or role == 'developer' %}
+        {{- '<|start_header_id|>system<|end_header_id|>\n\n' }}
+        {{- render_content(message.content, true, true)|trim }}
+        {{- '<|eot_id|>' }}
+    {%- elif role == 'user' %}
+        {{- '<|start_header_id|>user<|end_header_id|>\n\n' }}
+        {{- render_content(message.content, true)|trim }}
+        {{- '<|eot_id|>' }}
+    {%- elif role == 'assistant' %}
+        {%- set content = render_content(message.content, true)|trim %}
+        {%- set reasoning_content = '' %}
+        {%- if message.reasoning_content is defined and message.reasoning_content is string %}
+            {%- set reasoning_content = message.reasoning_content|trim %}
+        {%- elif '</think>' in content %}
+            {%- set reasoning_content = content.split('</think>')[0].rstrip('\n').split('<think>')[-1].lstrip('\n')|trim %}
+            {%- set content = content.split('</think>')[-1].lstrip('\n')|trim %}
+        {%- endif %}
+        {{- '<|start_header_id|>assistant<|end_header_id|>\n\n' }}
+        {%- if reasoning_content %}
+            {{- '<think>\n' + reasoning_content + '\n</think>\n\n' }}
+        {%- endif %}
+        {{- content }}
+        {%- if message.tool_calls is defined and message.tool_calls and message.tool_calls is iterable and message.tool_calls is not mapping %}
+            {%- if content or reasoning_content %}
+                {{- '\n\n' }}
+            {%- endif %}
+            {{- render_tool_calls_json(message.tool_calls) }}
+        {%- endif %}
+        {{- '<|eot_id|>' }}
+    {%- elif role == 'tool' or role == 'ipython' or role == 'function' %}
+        {{- '<|start_header_id|>ipython<|end_header_id|>\n\n' }}
+        {%- if message.content is mapping %}
+            {{- {"output": message.content} | tojson }}
+        {%- else %}
+            {%- set tool_content = render_content(message.content, true)|trim %}
+            {{- {"output": tool_content} | tojson }}
+        {%- endif %}
+        {{- '<|eot_id|>' }}
+    {%- else %}
+        {{- raise_exception('Unexpected message role: ' + role) }}
+    {%- endif %}
+{%- endfor %}
+
+{%- if add_generation_prompt is defined and add_generation_prompt %}
+    {{- '<|start_header_id|>assistant<|end_header_id|>\n\n' }}
+    {%- if enable_thinking is defined and enable_thinking %}
+        {{- '<think>\n' }}
+    {%- endif %}
+{%- endif %}

config.json

@@ -0,0 +1,36 @@
+{
+  "architectures": [
+    "LlamaForCausalLM"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "bos_token_id": 128000,
+  "dtype": "float32",
+  "eos_token_id": 128001,
+  "head_dim": 128,
+  "hidden_act": "silu",
+  "hidden_size": 4096,
+  "initializer_range": 0.02,
+  "intermediate_size": 14336,
+  "max_position_embeddings": 131072,
+  "mlp_bias": false,
+  "model_type": "llama",
+  "num_attention_heads": 32,
+  "num_hidden_layers": 32,
+  "num_key_value_heads": 8,
+  "pad_token_id": 128001,
+  "pretraining_tp": 1,
+  "rms_norm_eps": 1e-05,
+  "rope_parameters": {
+    "factor": 8.0,
+    "high_freq_factor": 4.0,
+    "low_freq_factor": 1.0,
+    "original_max_position_embeddings": 8192,
+    "rope_theta": 500000.0,
+    "rope_type": "llama3"
+  },
+  "tie_word_embeddings": false,
+  "transformers_version": "5.8.1",
+  "use_cache": false,
+  "vocab_size": 128256
+}

generation_config.json

@@ -0,0 +1,12 @@
+{
+  "_from_model_config": true,
+  "bos_token_id": 128000,
+  "eos_token_id": [
+    128001
+  ],
+  "output_attentions": false,
+  "output_hidden_states": false,
+  "pad_token_id": 128001,
+  "transformers_version": "5.8.1",
+  "use_cache": true
+}

model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ca40dc6a43271c21057f48b1a49d0c2adaf74022584a49d9683bde34b0460d2c
+size 32121079032

tokenizer.json

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6b9e4e7fb171f92fd137b777cc2714bf87d11576700a1dcd7a399e7bbe39537b
+size 17209920

tokenizer_config.json

@@ -0,0 +1,15 @@
+{
+  "backend": "tokenizers",
+  "bos_token": "<|begin_of_text|>",
+  "clean_up_tokenization_spaces": true,
+  "eos_token": "<|end_of_text|>",
+  "is_local": true,
+  "local_files_only": false,
+  "model_input_names": [
+    "input_ids",
+    "attention_mask"
+  ],
+  "model_max_length": 131072,
+  "pad_token": "<|end_of_text|>",
+  "tokenizer_class": "TokenizersBackend"
+}