laion/jtd-d5-69-30B

Full fine-tune vs Qwen/Qwen3-Coder-30B-A3B-Instruct. Fine-tune: 94% of the weights changed, across 48 of 48 layers.

Tensors changed: Embeddings 0/1, Norms 173/193, MLP 2611/2613, Attention 192/192, lm_head 1/1

1 derivatives checked