unsloth/Qwen3-235B-A22B-Thinking-2507-GGUF

Same weights, converted vs Qwen/Qwen3-235B-A22B-Thinking-2507. Same weights stored in another dtype (BF16 → BF16/F32), plus 282 new tensors.

Tensors changed: Embeddings 0/1, Norms 0/377, Attention 0/376, MLP 0/94, lm_head 0/1

On the Hub · Downloads on Model Pulse