Downloads · 30 days
68
3% of all-time downloads
vanch007/Qwen3.5-122B-A10B-abliterated-4bit-vlm-mlx-cs2764-pass2
Qwen3.5-122B-A10B-abliterated-4bit-vlm-mlx-cs2764-pass2 is a image-text-to-text model from vanch007. Use it for the image-text-to-text task on the model card, and read the license before you ship it in a product. It is set up for mlx-vlm. The card lists the license as apache-2.0.
4-bit MLX VLM release of a Qwen3.5-122B-A10B abliterated checkpoint, with an additional broad cs2764/mlx-abliteration pass applied on the MLX model.
Downloads · 30 days
68
3% of all-time downloads
All-time downloads
2K
Public
Parameters
123B
69.9 GB on disk
Likes
0
Public
Click a slice to open those files.
.safetensors69.9 GB · 100%
How the weights are stored.
U32122B · 100%
From the Hugging Face model README
4-bit MLX VLM release of a Qwen3.5-122B-A10B abliterated checkpoint, with an additional broad cs2764/mlx-abliteration pass applied on the MLX model.
A newer, more speed-balanced variant is available here:
qwen3_5_moe / Qwen3_5MoeForConditionalGenerationgroup_size=64, mode=affine)65.1 GB14cs2764/mlx-abliterationfinal Repo| Aspect | This pass2 repo | Newer final repo |
|---|---|---|
| Extra ablation scope | broader pass over layers 0..47 | targeted hot-path pass on layers 36..47 |
| Goal | maximize refusal reduction | better speed / behavior tradeoff |
| Local behavior spot checks | stronger refusal weakening | milder than pass2, but still weaker refusal than direct 4-bit base |
| Local short generation speed | about 12.6 tok/s | about 24.0 tok/s |
| Recommendation: |
pass2 repo if you want the more aggressive refusal-removal behavior.final repo if you want a faster model with a narrower targeted extra pass.The release chain for this repository is:
vision_config, preprocessor_config.json, processor_config.json, and video_preprocessor_config.jsoncs2764/mlx-abliteration on the converted MLX modelThis repository contains a complete MLX VLM checkpoint:
config.json with vision_configmodel.safetensors.index.jsonmodel-00001-of-00014.safetensors through model-00014-of-00014.safetensorstokenizer.json, tokenizer_config.json, vocab.jsonpreprocessor_config.json, processor_config.json, video_preprocessor_config.jsonabliteration_log.json for the pass-2 MLX abliteration runNote that ablation_meta.json is inherited from the input checkpoint. For the extra MLX pass in this repository, use abliteration_log.json and this model card as the authoritative record.
The additional MLX abliteration pass was run with the following settings:
| Parameter | Value |
|---|---|
| Toolkit | cs2764/mlx-abliteration |
| Refusal vector policy | per-layer |
| Ablation vector source | per-layer |
| Ablation strength | 2.0 |
| Refusal direction method | projected |
| Probed layers | 0..47 |
| Adaptive search | False |
| Attention only | False |
| MoE safe mode | True |
| Timestamp | 2026-03-07T02:40:35Z |
This run was done on the MLX model, not on a raw Transformers checkpoint.
This repository is for MLX / Apple Silicon usage. It is not a standard Transformers-only release.
Verified locally:
mlx_vlm.load(..., lazy=True) loads successfullyprocessor_class = Qwen3VLProcessorhas_vision_tower = Trueconfig.json retains vision_configmlx-vlmfrom mlx_vlm import load, generate
model, processor = load(
"vanch007/Qwen3.5-122B-A10B-abliterated-4bit-vlm-mlx-cs2764-pass2",
lazy=True,
)
result = generate(
model,
processor,
prompt="Describe the image briefly.",
image="/absolute/path/to/example.jpg",
max_tokens=128,
verbose=False,
)
print(result.text)
This repository contains a model with reduced refusal behavior. It may produce harmful, offensive, or unsafe content.
Do not use it in consumer-facing or safety-sensitive systems without independent safety controls.
You are responsible for ensuring compliance with applicable law, policy, and platform rules.