OpenGVLab/InternVL-Chat-V1-2
Full fine-tune (fewer layers) vs NousResearch/Nous-Hermes-2-Yi-34B. NousResearch/Nous-Hermes-2-Yi-34B matches part of OpenGVLab/InternVL-Chat-V1-2 (541 of its 1138 tensors; the rest of OpenGVLab/InternVL-Chat-V1-2 has no counterpart). On the shared part: Fine-tune: 100% of the weights changed, across 60 of 60 layers.
Tensors changed: Norms 77/121, MLP 180/180, Attention 240/240