SakanaAI/TAID-VLM-2B
Partial fine-tune vs OpenGVLab/InternVL2-2B. Partial change: 86% of the weights changed (attention, embeddings, mlp, norms, other), in 24 of 24 layers.
Tensors changed: Attention 48/144, Norms 2/145, MLP 72/174, Embeddings 1/5, other 1/49