OpenGVLab/InternVL3_5-4B-Instruct

Full fine-tune vs OpenGVLab/InternVL3_5-4B-Pretrained. Fine-tune: 100% of the weights changed, across 36 of 36 layers.

Tensors changed: Embeddings 5/5, Norms 241/241, MLP 210/210, Attention 240/240, other 48/48, lm_head 1/1

1 derivatives checked

• OpenGVLab/InternVL3_5-4B-MPO: Full fine-tune

On the Hub · Downloads on Model Pulse