zerofata/GLM-4.5-Iceblink-v3-106B-A12B
LoRA-style vs zai-org/GLM-4.5-Air. 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, q_proj, up_proj, v_proj changed, in 46 of 46 layers; norms, biases and embeddings are untouched.
Tensors changed: Embeddings 0/1, Norms 0/93, MLP 2277/2505, Attention 184/322, lm_head 0/1
3 derivatives checked
- zerofata/GLM-4.5-Iceblink-v3-106B-A12B-GGUF: Quantization
- mradermacher/GLM-4.5-Iceblink-v3-106B-A12B-i1-GGUF: Quantization
- mradermacher/GLM-4.5-Iceblink-v3-106B-A12B-GGUF: Quantization