cucl2/AnyAudio-Judge-30B
Partial fine-tune vs Qwen/Qwen3-Omni-30B-A3B-Captioner. Partial change: 72% of the weights changed (attention, lm_head, mlp, norms), in 48 of 48 layers.
Tensors changed: Embeddings 0/4, Norms 173/439, MLP 1737/1989, Attention 192/556, other 0/11, lm_head 1/1