qwp4w3hyb/Hunyuan-A13B-Instruct-hf-WIP-GGUF

Same weights, converted vs bullerwins/Hunyuan-A13B-Instruct-hf. Same weights stored in another dtype (BF16/F32 → BF16/F32), plus 96 new tensors.

Tensors changed: MLP 0/128, Embeddings 0/1, Norms 0/129, Attention 0/128

On the Hub · Downloads on Model Pulse