mradermacher/UnifiedReward-Think-qwen3vl-2b-GGUF
Same weights, converted (fewer layers) vs CodeGoat24/UnifiedReward-Think-qwen3vl-2b. mradermacher/UnifiedReward-Think-qwen3vl-2b-GGUF matches part of CodeGoat24/UnifiedReward-Think-qwen3vl-2b (310 of its 625 tensors; the rest of CodeGoat24/UnifiedReward-Think-qwen3vl-2b has no counterpart). On the shared part: Same weights stored in another dtype (BF16 → F16/F32).
Tensors changed: Embeddings 0/1, Norms 0/113, MLP 0/84, Attention 0/112