Minthy/ToriiGate-v0.3
Full fine-tune vs HuggingFaceM4/Idefics3-8B-Llama3. Fine-tune: 94% of the weights changed, across 32 of 32 layers.
Tensors changed: other 1/1, Embeddings 3/4, Norms 86/175, MLP 194/204, Attention 343/344, lm_head 1/1
Full fine-tune vs HuggingFaceM4/Idefics3-8B-Llama3. Fine-tune: 94% of the weights changed, across 32 of 32 layers.
Tensors changed: other 1/1, Embeddings 3/4, Norms 86/175, MLP 194/204, Attention 343/344, lm_head 1/1