GAIR/SR-Scientist-30B

Partial fine-tune vs Qwen/Qwen3-Coder-30B-A3B-Instruct. Partial change: 88% of the weights changed (attention, mlp, norms), in 48 of 48 layers.

Tensors changed: Embeddings 0/1, Norms 91/193, MLP 2581/2613, Attention 192/192, lm_head 0/1

2 derivatives checked