tsinghua-ee/video-SALMONN-2
Full fine-tune (fewer layers) vs Qwen/Qwen2-7B. Qwen/Qwen2-7B matches part of tsinghua-ee/video-SALMONN-2 (339 of its 1316 tensors; the rest of tsinghua-ee/video-SALMONN-2 has no counterpart). On the shared part: Fine-tune: 100% of the weights changed, across 28 of 28 layers.
Tensors changed: Embeddings 1/1, Norms 57/57, MLP 84/84, Attention 196/196, lm_head 1/1