OpenGVLab/InternVL3_5-4B-HF

Full fine-tune vs OpenGVLab/InternVL3_5-4B-MPO. Fine-tune: 91% of the weights changed, across 36 of 36 layers.

Tensors changed: Embeddings 0/1, Norms 102/145, MLP 108/108, Attention 144/144, lm_head 1/1

1 derivatives checked

• lewiswatson/Frame2KG-InternVL3.5-4b-JSON: LoRA-style (fewer layers)

On the Hub · Downloads on Model Pulse