AvitoTech/avision
Full fine-tune vs Qwen/Qwen2.5-VL-7B-Instruct. Fine-tune: 91% of the weights changed, across 28 of 32 layers.
Tensors changed: Norms 55/122, MLP 84/280, Attention 196/324, Embeddings 0/1
Full fine-tune vs Qwen/Qwen2.5-VL-7B-Instruct. Fine-tune: 91% of the weights changed, across 28 of 32 layers.
Tensors changed: Norms 55/122, MLP 84/280, Attention 196/324, Embeddings 0/1