CodeGoat24/UnifiedReward-Flex-qwen3vl-4b

Full fine-tune vs CodeGoat24/UnifiedReward-Think-qwen3vl-4b. Fine-tune: 91% of the weights changed, across 36 of 36 layers.

Tensors changed: Embeddings 1/4, Norms 127/249, MLP 108/220, Attention 144/240

2 derivatives checked

• mradermacher/UnifiedReward-Flex-qwen3vl-4b-i1-GGUF: Full fine-tune (fewer layers)
• mradermacher/UnifiedReward-Flex-qwen3vl-4b-GGUF: Full fine-tune (fewer layers)

On the Hub · Downloads on Model Pulse