OpenGVLab/InternVL2-26B

Full fine-tune (fewer layers) vs OpenGVLab/InternViT-6B-448px-V1-5. OpenGVLab/InternViT-6B-448px-V1-5 matches part of OpenGVLab/InternVL2-26B (589 of its 934 tensors; the rest of OpenGVLab/InternVL2-26B has no counterpart). On the shared part: Fine-tune: 100% of the weights changed, across 45 of 45 layers.

Tensors changed: Embeddings 4/4, Norms 180/180, Attention 135/135, other 90/90, MLP 180/180

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

On the Hub · Downloads on Model Pulse