gabriellarson/ERNIE-4.5-21B-A3B-Thinking-GGUF

Same weights, converted vs baidu/ERNIE-4.5-21B-A3B-Thinking. Same weights stored in another dtype (BF16/F32 → F16/F32), plus 108 new tensors.

Tensors changed: MLP 0/111, Embeddings 0/1, Norms 0/57, Attention 0/112

On the Hub · Downloads on Model Pulse