TIGER-Lab/VL-Rethinker-7B

Full fine-tune vs Qwen/Qwen2.5-VL-7B-Instruct. Fine-tune: 100% of the weights changed, across 32 of 32 layers.

Tensors changed: Embeddings 2/2, Norms 56/122, MLP 279/280, Attention 324/324, lm_head 1/1

2 derivatives checked

• mradermacher/VL-Rethinker-7B-i1-GGUF: Quantization (fewer layers)
• mradermacher/VL-Rethinker-7B-GGUF: Same weights, converted (fewer layers)

On the Hub · Downloads on Model Pulse