TIGER-Lab/VL-Reasoner-72B
Full fine-tune vs Qwen/Qwen2.5-VL-72B-Instruct. Fine-tune: 97% of the weights changed, across 80 of 80 layers.
Tensors changed: Embeddings 1/2, Norms 124/226, Attention 686/688, MLP 436/436, lm_head 0/1
6 derivatives checked
- mradermacher/VL-Rethinker-Preview-i1-GGUF: Quantization
- mradermacher/MM-Thinker-72B-i1-GGUF: Quantization
- mradermacher/VL-Reasoner-72B-i1-GGUF: Quantization
- mradermacher/VL-Rethinker-Preview-GGUF: Quantization
- mradermacher/VL-Reasoner-72B-GGUF: Quantization
- mradermacher/MM-Thinker-72B-GGUF: Quantization