DavidAU/Qwen3.5-27B-Claude-4.6-OS-INSTRUCT
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 0/4, MLP 192/304, Attention 112/556, Norms 0/319
17 derivatives checked
- mradermacher/Qwen3.5-27B-Claude-4.6-OS-INSTRUCT-i1-GGUF: Quantization
- mradermacher/Qwen3.5-27B-Claude-4.6-OS-INSTRUCT-GGUF: Quantization
- nightmedia/Qwen3.6-27B-Architect-Polaris2-Fable-B-F451: Full fine-tune
- nightmedia/Qwen3.5-27B-Claude-4.6-OS-INSTRUCT-mxfp8-mlx: Quantization
- nightmedia/Qwen3.6-27B-Miraculix: Full fine-tune
- meddle86/Qwen3.6-27B-ArchPolaris2-Fable-exl3-4.0bpw: Full fine-tune
- nightmedia/Qwen3.8-27B-Holodeck-mxfp4-mlx: Quantization
- nightmedia/Qwen3.6-27B-Architect-Polaris2-Fable-B-F451-Medical-mxfp8-mlx: Quantization
- nightmedia/Qwen3.6-27B-GrandWichtel-mxfp8-mlx: Quantization
- nightmedia/Qwen3.6-27B-Architect-Polaris2-Fable-B-F451-Tess-1M-qx64-hi-mlx: Quantization
- nightmedia/Qwen3.6-27B-Architect-Polaris2-Fable-B-F451-Tess: Full fine-tune
- nightmedia/Qwen3.6-27B-Seven: Full fine-tune
- nightmedia/Qwen3.6-27B-Jormungandr: Full fine-tune
- nightmedia/Qwen3.6-27B-Akka: Full fine-tune
- nightmedia/Qwen3.6-27B-Architect-Polaris2-Fable-B-F451-F32: Full fine-tune
- nightmedia/Qwen3.5-27B-DS9: Full fine-tune
- nightmedia/Qwen3.5-27B-Engineer-Deckard-Claude: Full fine-tune