lovedheart/Qwen3-Next-REAP-20B-A3B-Thinking-GGUF

Same weights, converted vs Qwen/Qwen3-Next-80B-A3B-Thinking. Same weights stored in another dtype (BF16 → BF16/F32), plus 252 new tensors.

Tensors changed: Embeddings 0/1, Norms 0/121, MLP 0/192, Attention 0/48, lm_head 0/1

On the Hub · Downloads on Model Pulse