Jackrong/Qwen3.5-27B-Claude-4.6-Opus-Reasoning-Distilled-v2
LoRA-style vs Qwen/Qwen3.5-27B. Looks like a merged LoRA (or a fine-tune with frozen norms): only the weight matrices of down_proj, gate_proj, 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: lm_head 0/1, Embeddings 0/5, MLP 192/307, Attention 112/560, other 0/1, Norms 0/325
7 derivatives checked
- QuantTrio/Qwen3.5-27B-Claude-4.6-Opus-Reasoning-Distilled-v2-AWQ: Quantization
- Jackrong/MLX-Qwen3.5-27B-Claude-4.6-Opus-Reasoning-Distilled-v2-4bit: Quantization
- Flakily6416/Qwen3.5-27B-Opus-Reasoning-v2-Abliterated-EvoPress-GGUF: Quantization
- cpatonn/Qwen3.5-27B-Claude-4.6-Opus-Reasoning-Distilled-v2-AWQ-4bit: Quantization
- Jackrong/MLX-Qwen3.5-27B-Claude-4.6-Opus-Reasoning-Distilled-v2-8bit: Quantization
- Jackrong/MLX-Qwen3.5-27B-Claude-4.6-Opus-Reasoning-Distilled-v2-6bit: Quantization
- mconcat/Qwen3.5-27B-Claude-4.6-Opus-Reasoning-Distilled-v2-NVFP4: Quantization
2 other repos hold byte-for-byte identical weight files.