MBZUAI/MediX-R1-2B
Full fine-tune vs Qwen/Qwen3-VL-2B-Instruct. Fine-tune: 100% of the weights changed, across 28 of 28 layers.
Tensors changed: Embeddings 4/4, Norms 156/217, MLP 185/196, Attention 208/208
3 derivatives checked
- MBZUAI/MediX-R1-2B-GGUF: Same weights, converted (fewer layers)
- mradermacher/MediX-R1-2B-i1-GGUF: Quantization (fewer layers)
- mradermacher/MediX-R1-2B-GGUF: Same weights, converted (fewer layers)