tencent/UI-Mate-27B
Full fine-tune vs Qwen/Qwen3.6-27B. Fine-tune: 95% of the weights changed, across 64 of 64 layers.
Tensors changed: Embeddings 3/4, MLP 284/304, Attention 490/556, lm_head 1/1, Norms 221/319
5 derivatives checked
- bartowski/tencent_UI-Mate-27B-GGUF: Same weights, converted
- mradermacher/UI-Mate-27B-i1-GGUF: Quantization
- mradermacher/UI-Mate-27B-GGUF: Quantization
- DevQuasar/tencent.UI-Mate-27B-GGUF: Quantization
- starchicken/UI-Mate-27B-GGUF: Quantization