vidore/ColSmolVLM-base

Same weights, converted vs HuggingFaceTB/SmolVLM-Instruct. Same weights stored in another dtype (BF16 → F32), plus 2 new tensors.

Tensors changed: other 0/1, Embeddings 0/4, Norms 0/159, MLP 0/180, Attention 0/312

1 derivatives checked