migtissera/Tess-4-27B
Full fine-tune vs Qwen/Qwen3.6-27B. Fine-tune: 92% of the weights changed, across 64 of 64 layers.
Tensors changed: Embeddings 0/5, MLP 192/307, Attention 208/560, lm_head 1/1, Norms 0/325, other 0/1
6 derivatives checked
- migtissera/Tess-4-27B-GGUF: Quantization
- bartowski/migtissera_Tess-4-27B-GGUF: Same weights, converted
- lmcoleman/Tess-4-27B-ROCmFPX-GGUF: Same weights, converted
- mlx-community/Tess-4-27B-OptiQ-4bit: Quantization
- AlexanderKyng/Tess-4-27B-Mixed-Q5: Quantization
- migtissera/Tess-4-27B-EAGLE3: Different architecture