DevQuasar/Qwen.Qwen3-Coder-480B-A35B-Instruct-GGUF

Same weights, converted vs Qwen/Qwen3-Coder-480B-A35B-Instruct. Same weights stored in another dtype (BF16 → F16/F32), plus 186 new tensors.

Tensors changed: Embeddings 0/1, Norms 0/249, Attention 0/248, MLP 0/62, lm_head 0/1

On the Hub · Downloads on Model Pulse