wanglne/EVA-Qwen2.5-VL-7B-Instruct
Abliterated vs Qwen/Qwen2.5-VL-7B-Instruct. Looks like abliteration (refusal removal): only the output projections (down_proj) changed, in 2 of 32 layers. Abliteration edits exactly these matrices to remove a "refusal direction"; everything else is untouched.
Tensors changed: Embeddings 0/2, Norms 0/122, MLP 2/280, Attention 0/324, lm_head 0/1
1 derivatives checked
- mradermacher/EVA-Qwen2.5-VL-7B-Instruct-GGUF: Same weights, converted (fewer layers)