Delta-Vector/Austral-32B-GLM4-Winton
LoRA-style vs Delta-Vector/Austral-GLM4-KTO. Looks like a merged LoRA (or a fine-tune with frozen norms): only the weight matrices of down_proj, gate_up_proj, k_proj, o_proj, q_proj, v_proj changed, in 61 of 61 layers; norms, biases and embeddings are untouched.
Tensors changed: Embeddings 0/1, Norms 0/245, MLP 122/122, Attention 244/244, lm_head 0/1
5 derivatives checked
- bartowski/Delta-Vector_Austral-32B-GLM4-Winton-GGUF: Same weights, converted
- mradermacher/Austral-GLM4-Winton-i1-GGUF: Quantization
- mradermacher/Austral-32B-GLM4-Winton-i1-GGUF: Quantization
- mradermacher/Austral-GLM4-Winton-GGUF: Quantization
- mradermacher/Austral-32B-GLM4-Winton-GGUF: Quantization