OpenGVLab/InternVL3_5-30B-A3B-HF

Full fine-tune vs OpenGVLab/InternVL3_5-30B-A3B-MPO. Fine-tune: 94% of the weights changed, across 48 of 48 layers.

Tensors changed: Norms 154/193, MLP 2171/2171, Attention 192/192, lm_head 1/1, Embeddings 0/1

On the Hub · Downloads on Model Pulse