lurf21/Qwen2.5-Coder-32B-NEP
LoRA-style vs Qwen/Qwen2.5-Coder-32B. 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 64 of 64 layers; norms, biases and embeddings are untouched.
Tensors changed: Norms 0/129, MLP 192/192, Attention 256/448
4 derivatives checked
- mradermacher/Qwen2.5-Coder-32B-NEP-new-GGUF: Quantization
- mradermacher/Qwen2.5-coder-32B-NES-i1-GGUF: Quantization
- mradermacher/Qwen2.5-coder-32B-NES-GGUF: Full fine-tune
- mradermacher/Qwen2.5-Coder-32B-NEP-fix-GGUF: Full fine-tune