TeichAI/Qwen3.8-27B-Fable-Distill
LoRA-style vs Qwen/Qwen3.8-27B. Looks like a merged LoRA (or a fine-tune with frozen norms): only the weight matrices of down_proj, gate_proj, in_proj_a, in_proj_b, in_proj_qkv, in_proj_z, k_proj, o_proj, out_proj, q_proj, up_proj, v_proj changed, in 64 of 64 layers; norms, biases and embeddings are untouched.
Tensors changed: Norms 0/325, Attention 304/560, MLP 192/307, Embeddings 0/5, lm_head 0/1, other 0/1
10 derivatives checked
- TeichAI/Qwen3.8-27B-Fable-Distill-GGUF: Same weights, converted
- mradermacher/Qwen3.8-27B-Fable-Distill-i1-GGUF: Quantization
- mradermacher/Qwen3.8-27B-Fable-Distill-GGUF: Quantization
- vwdubb/Qwen3.8-27B-Fable-Distill-FP8: Quantization
- caslca/Qwen3.8-27B-Fable-Distill-OptiQ-4.5bpw-mixed: Quantization
- vwdubb/Qwen3.8-27B-Fable-Distill-NVFP4: Quantization
- caslca/Qwen3.8-27B-Fable-Distill-mlx-uniform-4bit: Quantization
- McG-221/Qwen3.8-27B-Fable-Distill-mlx-8Bit: Quantization
- lued/Qwen3.8-27B-TeichAI-Fable-INT8-W8A16-MTP: Quantization
- armand0e/Qwen3.8-27B-Fable-Distill-Heretic-ara: Abliterated