wjbmattingly/comma-qwen-3.5-9b-full
Partial fine-tune vs Qwen/Qwen3.5-9B. Partial change: 89% of the weights changed (attention, embeddings, lm_head, mlp, norms), in 32 of 32 layers.
Tensors changed: lm_head 1/1, Embeddings 3/4, MLP 198/208, Attention 287/332, Norms 141/215
1 derivatives checked
- mradermacher/comma-qwen-3.5-9b-full-GGUF: Same weights, converted