rodrigomt/Qwen-3.5-Opus-GLM-27B
LoRA-style vs Jackrong/Qwen3.5-27B-GLM5.1-Distill-v1. 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 0/1, Norms 0/209, Attention 112/448, MLP 192/192
4 derivatives checked
- mradermacher/Qwopus-GLM-27B-dare-ties-i1-GGUF: Quantization
- mradermacher/Qwen-3.5-Opus-GLM-27B-i1-GGUF: Quantization
- mradermacher/Qwen-3.5-Opus-GLM-27B-GGUF: Quantization
- mradermacher/Qwopus-GLM-27B-dare-ties-GGUF: Quantization