AXCXEPT/EZO-Qwen2.5-32B-Instruct
LoRA-style vs Qwen/Qwen2.5-32B-Instruct. 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: Embeddings 0/1, Norms 0/129, MLP 192/192, Attention 256/448, lm_head 0/1
11 derivatives checked
- mradermacher/EZO-Qwen2.5-32B-Instruct-i1-GGUF: Quantization
- mradermacher/EZO-Qwen2.5-32B-Instruct-GGUF: Quantization
- tensorblock/EZO-Qwen2.5-32B-Instruct-GGUF: Quantization
- nitky/EZO-QwQ-32B-Preview: Full fine-tune
- YOYO-AI/Qwen2.5-32B-YOYO-V2: Full fine-tune
- YOYO-AI/Qwen2.5-32B-YOYO-karcher: Full fine-tune
- YOYO-AI/Qwen2.5-32B-YOYO-stock: Full fine-tune
- Kaoeiri/Qwenwify-32B-v4: Full fine-tune
- Kaoeiri/Qwenwify-32B-v2.5: Full fine-tune
- Kaoeiri/Qwenwify-32B-v2.6: Full fine-tune
- Kaoeiri/Qwenwify-32B-v2: Full fine-tune
1 other repos hold byte-for-byte identical weight files.