Schiltmans/Ternary-Bonsai-2-27B-DFlash2-ft5
Partial fine-tune vs z-lab/Qwen3.8-27B-DFlash2. Partial change: 90% of the weights changed (attention, mlp, other), in 5 of 5 layers.
Tensors changed: other 2/4, Norms 0/22, Attention 20/30, MLP 15/25
2 derivatives checked
- Schiltmans/Ternary-Bonsai-2-27B-DFlash2-ft5-GGUF: Quantization
- Schiltmans/Ternary-Bonsai-2-27B-DFlash2-ft5-mlx-4bit: Quantization