DevQuasar/Kwaipilot.KwaiCoder-23B-A4B-v1-GGUF

Same weights, converted vs Kwaipilot/KwaiCoder-23B-A4B-v1. Same weights stored in another dtype (BF16 → F16/F32), plus 60 new tensors.

Tensors changed: Embeddings 0/1, other 0/20, Norms 0/41, Attention 0/80, lm_head 0/1

On the Hub · Downloads on Model Pulse