Qwen/Qwen2-VL-72B-Instruct
Full fine-tune vs Qwen/Qwen2-VL-72B. Fine-tune: 99% of the weights changed, across 80 of 80 layers.
Tensors changed: Embeddings 1/2, Norms 131/291, Attention 560/688, MLP 244/372, lm_head 1/1
9 derivatives checked
- bartowski/Qwen2-VL-72B-Instruct-GGUF: Quantization
- second-state/Qwen2-VL-72B-Instruct-GGUF: Same weights, converted
- mradermacher/Qwen2-VL-72B-Instruct-i1-GGUF: Quantization
- Qwen/Qwen2-VL-72B-Instruct-AWQ: Quantization
- mradermacher/Qwen2-VL-72B-Instruct-GGUF: Quantization
- gaianet/Qwen2-VL-72B-Instruct-GGUF: Same weights, converted
- Qwen/Qwen2-VL-72B-Instruct-GPTQ-Int4: Quantization
- lmstudio-community/Qwen2-VL-72B-Instruct-GGUF: Quantization
- Qwen/Qwen2-VL-72B-Instruct-GPTQ-Int8: Quantization
2 other repos hold byte-for-byte identical weight files.