logic65/Qwen3.8-Whittle-Next-19.8B-A11B-chat
Partial fine-tune vs logic65/Qwen3.8-Whittle-Next-19.8B-A10B-base. Partial change: 18% of the weights changed (attention, embeddings, mlp, norms, other), in 32 of 32 layers.
Tensors changed: lm_head 0/1, Embeddings 2/9, Norms 73/124, other 5/5, Attention 128/328, MLP 160/320
1 derivatives checked
- mradermacher/Qwen3.8-Whittle-Next-19.8B-A11B-chat-GGUF: Not comparable yet