yifeihu/TB-OCR-preview-0.1
Full fine-tune vs microsoft/Phi-3.5-vision-instruct. Fine-tune: 100% of the weights changed, across 32 of 32 layers.
Tensors changed: Embeddings 377/398, Norms 65/65, MLP 64/64, Attention 64/64, lm_head 1/1
Full fine-tune vs microsoft/Phi-3.5-vision-instruct. Fine-tune: 100% of the weights changed, across 32 of 32 layers.
Tensors changed: Embeddings 377/398, Norms 65/65, MLP 64/64, Attention 64/64, lm_head 1/1