HuggingFaceTB/SmolVLM2-2.2B-Base
Full fine-tune vs HuggingFaceTB/SmolVLM-Base. Fine-tune: 100% of the weights changed, across 27 of 27 layers.
Tensors changed: other 1/1, Norms 159/159, MLP 126/180, Attention 204/312, Embeddings 3/3
Full fine-tune vs HuggingFaceTB/SmolVLM-Base. Fine-tune: 100% of the weights changed, across 27 of 27 layers.
Tensors changed: other 1/1, Norms 159/159, MLP 126/180, Attention 204/312, Embeddings 3/3