Phora68/rapha
Partial fine-tune vs unsloth/Qwen2.5-3B-Instruct-bnb-4bit. Partial change: 69% of the weights changed (attention, mlp), in 36 of 36 layers.
Tensors changed: MLP 408/628, Attention 525/921, Embeddings 0/1, Norms 0/73
Partial fine-tune vs unsloth/Qwen2.5-3B-Instruct-bnb-4bit. Partial change: 69% of the weights changed (attention, mlp), in 36 of 36 layers.
Tensors changed: MLP 408/628, Attention 525/921, Embeddings 0/1, Norms 0/73