Image-Text-to-Text
MLX
Safetensors
mlx-vlm
qwen
qwen3.8
multimodal
2-bit
4-bit precision
6-bit
8-bit precision
bf16
abliterated
Instructions to use PocketAiHub/Qwen3.8-27B-Abliterated-MLX with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- MLX
How to use PocketAiHub/Qwen3.8-27B-Abliterated-MLX 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/Qwen3.8-27B-Abliterated-MLX") config = load_config("PocketAiHub/Qwen3.8-27B-Abliterated-MLX") # 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
- Atomic Chat
| { | |
| "schema_version": 1, | |
| "method": "refusal-direction orthogonal weight projection", | |
| "base_model": "Qwen/Qwen3.8-27B", | |
| "base_revision": "1d4bf0f2ff6012fd82039f2fa52739d0dd7c60c0", | |
| "direction": { | |
| "dataset_examples_per_class": 256, | |
| "representation": "projected_unit_directions", | |
| "hidden_state": "residual_post at assistant-generation boundary", | |
| "sha256": "3958f6bba70f35e869c2918b61a4858a1fa53fbd3c5f31ac594c5b2d87105c51", | |
| "source_layer": 53 | |
| }, | |
| "projection": { | |
| "destination_layers": [ | |
| 24, | |
| 25, | |
| 26, | |
| 27, | |
| 28, | |
| 29, | |
| 30, | |
| 31, | |
| 32, | |
| 33, | |
| 34, | |
| 35, | |
| 36, | |
| 37, | |
| 38, | |
| 39, | |
| 40, | |
| 41, | |
| 42, | |
| 43, | |
| 44, | |
| 45, | |
| 46, | |
| 47, | |
| 48, | |
| 49, | |
| 50, | |
| 51, | |
| 52, | |
| 53, | |
| 54, | |
| 55, | |
| 56, | |
| 57, | |
| 58, | |
| 59, | |
| 60, | |
| 61, | |
| 62, | |
| 63 | |
| ], | |
| "target_kinds": [ | |
| "full_attention_out", | |
| "linear_attention_out", | |
| "mlp_down" | |
| ], | |
| "scale": 1.0, | |
| "norm_preserve": true, | |
| "modified_tensor_count": 80, | |
| "modified_by_kind": { | |
| "full_attention_out": 10, | |
| "linear_attention_out": 30, | |
| "mlp_down": 40 | |
| } | |
| }, | |
| "architecture": { | |
| "modelType": "qwen3_5", | |
| "hiddenSize": 5120, | |
| "numHiddenLayers": 64, | |
| "fullAttentionInterval": 4 | |
| } | |
| } | |