OpenGVLab/InternVL2-4B
Full fine-tune (fewer layers) vs OpenGVLab/InternViT-300M-448px. OpenGVLab/InternViT-300M-448px matches part of OpenGVLab/InternVL2-4B (340 of its 541 tensors; the rest of OpenGVLab/InternVL2-4B has no counterpart). On the shared part: Fine-tune: 100% of the weights changed, across 24 of 24 layers.
Tensors changed: Embeddings 4/4, Attention 96/96, other 48/48, MLP 96/96, Norms 96/96
3 derivatives checked
- OS-Copilot/OS-Atlas-Base-4B: Full fine-tune
- OpenGVLab/Mini-InternVL2-4B-DA-DriveLM: Full fine-tune
- OpenGVLab/Mini-InternVL2-4B-DA-Medical: Full fine-tune
2 other repos hold byte-for-byte identical weight files.