OpenGVLab/InternVL2-8B-MPO
Full fine-tune vs OpenGVLab/InternVL2-8B. Fine-tune: 95% of the weights changed, across 32 of 32 layers.
Tensors changed: Attention 160/160, Norms 107/161, MLP 197/198, Embeddings 4/5, other 1/49
Full fine-tune vs OpenGVLab/InternVL2-8B. Fine-tune: 95% of the weights changed, across 32 of 32 layers.
Tensors changed: Attention 160/160, Norms 107/161, MLP 197/198, Embeddings 4/5, other 1/49