allenai/SAGE-MM-Qwen2.5-VL-7B-SFT
Partial fine-tune vs Qwen/Qwen2.5-VL-7B-Instruct. Partial change: 85% of the weights changed (attention, lm_head, mlp, norms), in 28 of 32 layers.
Tensors changed: Embeddings 0/2, Norms 55/122, MLP 84/280, Attention 196/324, lm_head 1/1
4 derivatives checked
- mradermacher/SAGE-MM-Qwen2.5-VL-7B-SFT-i1-GGUF: Quantization (fewer layers)
- mradermacher/SAGE-MM-Qwen2.5-VL-7B-SFT-GGUF: Same weights, converted (fewer layers)
- prithivMLmods/SAGE-MM-Qwen2.5-VL-7B-SFT-GGUF: Same weights, converted (fewer layers)
- allenai/SAGE-MM-Qwen2.5-VL-7B-SFT_RL: Partial fine-tune