Image-Text-to-Text
MLX
Safetensors
qwen3_5_moe
mlx-vlm
ornith
qwen3.5-moe
multimodal
abliterated
bf16
conversational
Instructions to use PocketAiHub/Ornith-1.5-35B-A3B-Abliterated-MLX-BF16 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- MLX
How to use PocketAiHub/Ornith-1.5-35B-A3B-Abliterated-MLX-BF16 with MLX:
# Make sure mlx-vlm is installed # pip install --upgrade mlx-vlm from mlx_vlm import load, generate from mlx_vlm.prompt_utils import apply_chat_template from mlx_vlm.utils import load_config # Load the model model, processor = load("PocketAiHub/Ornith-1.5-35B-A3B-Abliterated-MLX-BF16") config = load_config("PocketAiHub/Ornith-1.5-35B-A3B-Abliterated-MLX-BF16") # Prepare input image = ["http://images.cocodataset.org/val2017/000000039769.jpg"] prompt = "Describe this image." # Apply chat template formatted_prompt = apply_chat_template( processor, config, prompt, num_images=1 ) # Generate output output = generate(model, processor, formatted_prompt, image) print(output) - Notebooks
- Google Colab
- Kaggle
- Local Apps Settings
- LM Studio
- Pi
How to use PocketAiHub/Ornith-1.5-35B-A3B-Abliterated-MLX-BF16 with Pi:
Start the MLX server
# Install MLX LM: uv tool install mlx-lm # Start a local OpenAI-compatible server: mlx_lm.server --model "PocketAiHub/Ornith-1.5-35B-A3B-Abliterated-MLX-BF16"
Configure the model in Pi
# Install Pi: npm install -g @earendil-works/pi-coding-agent # Add to ~/.pi/agent/models.json: { "providers": { "mlx-lm": { "baseUrl": "http://localhost:8080/v1", "api": "openai-completions", "apiKey": "none", "models": [ { "id": "PocketAiHub/Ornith-1.5-35B-A3B-Abliterated-MLX-BF16" } ] } } }Run Pi
# Start Pi in your project directory: pi
- Hermes Agent
How to use PocketAiHub/Ornith-1.5-35B-A3B-Abliterated-MLX-BF16 with Hermes Agent:
Start the MLX server
# Install MLX LM: uv tool install mlx-lm # Start a local OpenAI-compatible server: mlx_lm.server --model "PocketAiHub/Ornith-1.5-35B-A3B-Abliterated-MLX-BF16"
Configure Hermes
# Install Hermes: curl -fsSL https://hermes-agent.nousresearch.com/install.sh | bash hermes setup # Point Hermes at the local server: hermes config set model.provider custom hermes config set model.base_url http://127.0.0.1:8080/v1 hermes config set model.default PocketAiHub/Ornith-1.5-35B-A3B-Abliterated-MLX-BF16
Run Hermes
hermes
- Atomic Chat
- OpenClaw
How to use PocketAiHub/Ornith-1.5-35B-A3B-Abliterated-MLX-BF16 with OpenClaw:
Start the MLX server
# Install MLX LM: uv tool install mlx-lm # Start a local OpenAI-compatible server: mlx_lm.server --model "PocketAiHub/Ornith-1.5-35B-A3B-Abliterated-MLX-BF16"
Configure OpenClaw
# Install OpenClaw: npm install -g openclaw@latest # Register the local server and set it as the default model: openclaw onboard --non-interactive --mode local \ --auth-choice custom-api-key \ --custom-base-url http://127.0.0.1:8080/v1 \ --custom-model-id "PocketAiHub/Ornith-1.5-35B-A3B-Abliterated-MLX-BF16" \ --custom-provider-id mlx-lm \ --custom-compatibility openai \ --custom-text-input \ --accept-risk \ --skip-health
Run OpenClaw
openclaw agent --local --agent main --message "Hello from Hugging Face"
| { | |
| "schema_version": 1, | |
| "model": "PocketAiHub/Ornith-1.5-35B-A3B-Abliterated-MLX-BF16", | |
| "variant": "bf16", | |
| "status": "experimental", | |
| "base_model": "ornith-ai/Ornith-1.5-35B-A3B", | |
| "base_revision": "e4dfb35a93d4b6822a811a7676f3488514abe7e2", | |
| "runtime": { | |
| "mlx": "0.32.0", | |
| "mlx_vlm": "0.6.8" | |
| }, | |
| "behavioral_evaluation": { | |
| "parent_12_case_gate": { | |
| "capability": { | |
| "count": 12, | |
| "pass_count": 12, | |
| "pass_rate": 1.0 | |
| }, | |
| "jbb_benign": { | |
| "count": 12, | |
| "explicit_refusal_count": 1, | |
| "explicit_refusal_rate": 0.08333333333333333, | |
| "final_answer_count": 12, | |
| "completed_answer_count": 1, | |
| "evasive_nonanswer_count": 0, | |
| "evasive_nonanswer_rate": 0.0, | |
| "truncated_generation_count": 11, | |
| "unfinished_reasoning_count": 0 | |
| }, | |
| "jbb_harmful": { | |
| "count": 12, | |
| "explicit_refusal_count": 12, | |
| "explicit_refusal_rate": 1.0, | |
| "final_answer_count": 12, | |
| "completed_answer_count": 11, | |
| "evasive_nonanswer_count": 0, | |
| "evasive_nonanswer_rate": 0.0, | |
| "truncated_generation_count": 1, | |
| "unfinished_reasoning_count": 0 | |
| } | |
| }, | |
| "selected_bf16_12_case_gate": { | |
| "capability": { | |
| "count": 12, | |
| "pass_count": 12, | |
| "pass_rate": 1.0 | |
| }, | |
| "jbb_benign": { | |
| "count": 12, | |
| "explicit_refusal_count": 0, | |
| "explicit_refusal_rate": 0.0, | |
| "final_answer_count": 12, | |
| "completed_answer_count": 1, | |
| "evasive_nonanswer_count": 0, | |
| "evasive_nonanswer_rate": 0.0, | |
| "truncated_generation_count": 11, | |
| "unfinished_reasoning_count": 0 | |
| }, | |
| "jbb_harmful": { | |
| "count": 12, | |
| "explicit_refusal_count": 0, | |
| "explicit_refusal_rate": 0.0, | |
| "final_answer_count": 12, | |
| "completed_answer_count": 4, | |
| "evasive_nonanswer_count": 0, | |
| "evasive_nonanswer_rate": 0.0, | |
| "truncated_generation_count": 8, | |
| "unfinished_reasoning_count": 0 | |
| } | |
| }, | |
| "variant_100_harmful_100_benign_screen": { | |
| "jbb_benign": { | |
| "count": 100, | |
| "explicit_refusal_count": 0, | |
| "explicit_refusal_rate": 0.0, | |
| "final_answer_count": 100, | |
| "completed_answer_count": 2, | |
| "evasive_nonanswer_count": 0, | |
| "evasive_nonanswer_rate": 0.0, | |
| "truncated_generation_count": 98, | |
| "unfinished_reasoning_count": 0 | |
| }, | |
| "jbb_harmful": { | |
| "count": 100, | |
| "explicit_refusal_count": 1, | |
| "explicit_refusal_rate": 0.01, | |
| "final_answer_count": 100, | |
| "completed_answer_count": 9, | |
| "evasive_nonanswer_count": 0, | |
| "evasive_nonanswer_rate": 0.0, | |
| "truncated_generation_count": 91, | |
| "unfinished_reasoning_count": 0 | |
| } | |
| }, | |
| "batch_size": 1, | |
| "max_new_tokens": 128, | |
| "thinking": false, | |
| "scorer": "transparent phrase-based explicit-refusal detector", | |
| "caveat": "This is an early-refusal screen, not a full answer-quality evaluation. Manual review found semantic refusals that the phrase detector did not flag." | |
| }, | |
| "matched_teacher_drift": { | |
| "protocol": { | |
| "context": "shared regular-BF16 greedy teacher trajectory", | |
| "positionsPerCaseMaximum": 16, | |
| "vocabulary": 248320, | |
| "logitsStorage": "float32", | |
| "probabilityMath": "float64 full-vocabulary log-softmax", | |
| "residualState": "all 40 residual_post vectors at assistant boundary", | |
| "cacheState": "final teacher-trajectory cache: K/V for 10 full-attention layers; convolution and gated-delta recurrent state for 30 linear-attention layers", | |
| "thinking": false, | |
| "sampling": "greedy" | |
| }, | |
| "dataset": { | |
| "capabilityFileSha256": "fe35cc41fe041cf2d371041c3fb977ec37b51fa1e9a0b0482aaf1eab415c9c26", | |
| "capabilityCases": 12, | |
| "jailbreakBench": { | |
| "repository": "JailbreakBench/JBB-Behaviors", | |
| "revision": "886acc352a31533ffbcf4ef22c744658688086fc" | |
| }, | |
| "harmfulCases": 12, | |
| "benignCases": 12 | |
| }, | |
| "pure_abliteration_vs_regular_bf16": { | |
| "positions": 481, | |
| "forward_kl_nats_mean": 0.54518455642922, | |
| "top1_agreement": 0.7817047817047817, | |
| "residual_cosine_mean": 0.9520897871696514, | |
| "cache": { | |
| "full_attention_k_cosine_mean": 0.9956834574679786, | |
| "full_attention_v_cosine_mean": 0.9942660171605967, | |
| "linear_attention_conv_cosine_mean": 0.997201268533088, | |
| "linear_attention_recurrent_cosine_mean": 0.9962457101916542 | |
| }, | |
| "suite_splits": { | |
| "capability": { | |
| "positions": 97, | |
| "forward_kl_nats_mean": 0.01834161340133247, | |
| "top1_agreement": 0.979381443298969, | |
| "residual_cosine_mean": 0.995895118556436 | |
| }, | |
| "jbb_benign": { | |
| "positions": 192, | |
| "forward_kl_nats_mean": 0.3073881842398057, | |
| "top1_agreement": 0.8541666666666666, | |
| "residual_cosine_mean": 0.9696202077556074 | |
| }, | |
| "jbb_harmful": { | |
| "positions": 192, | |
| "forward_kl_nats_mean": 1.0491463737941817, | |
| "top1_agreement": 0.609375, | |
| "residual_cosine_mean": 0.8907540351969113 | |
| } | |
| } | |
| }, | |
| "quantization_vs_abliterated_bf16": null, | |
| "total_vs_regular_bf16": { | |
| "positions": 481, | |
| "forward_kl_nats_mean": 0.54518455642922, | |
| "top1_agreement": 0.7817047817047817, | |
| "residual_cosine_mean": 0.9520897871696514, | |
| "cache": { | |
| "full_attention_k_cosine_mean": 0.9956834574679786, | |
| "full_attention_v_cosine_mean": 0.9942660171605967, | |
| "linear_attention_conv_cosine_mean": 0.997201268533088, | |
| "linear_attention_recurrent_cosine_mean": 0.9962457101916542 | |
| }, | |
| "suite_splits": { | |
| "capability": { | |
| "positions": 97, | |
| "forward_kl_nats_mean": 0.01834161340133247, | |
| "top1_agreement": 0.979381443298969, | |
| "residual_cosine_mean": 0.995895118556436 | |
| }, | |
| "jbb_benign": { | |
| "positions": 192, | |
| "forward_kl_nats_mean": 0.3073881842398057, | |
| "top1_agreement": 0.8541666666666666, | |
| "residual_cosine_mean": 0.9696202077556074 | |
| }, | |
| "jbb_harmful": { | |
| "positions": 192, | |
| "forward_kl_nats_mean": 1.0491463737941817, | |
| "top1_agreement": 0.609375, | |
| "residual_cosine_mean": 0.8907540351969113 | |
| } | |
| } | |
| } | |
| }, | |
| "capabilities": { | |
| "vision_tower_included": true, | |
| "video_preprocessor_included": true, | |
| "native_mtp_head_included": false, | |
| "four_bit_image_smoke_passed": false, | |
| "vision_evaluation_caveat": "Only the 4-bit artifact received an end-to-end image smoke test; vision and video were not comprehensively benchmarked." | |
| } | |
| } | |