OpenGVLab/InternVL3-9B-Instruct

Full fine-tune vs OpenGVLab/InternVL3-9B-Pretrained. Fine-tune: 100% of the weights changed, across 48 of 48 layers.

Tensors changed: Attention 192/192, Norms 186/193, MLP 246/246, Embeddings 5/5, other 49/49

1 derivatives checked

• OpenGVLab/InternVL3-9B: Full fine-tune

1 other repos hold byte-for-byte identical weight files.

On the Hub · Downloads on Model Pulse