RedHatAI/DeepSeek-R1-Distill-Llama-70B-quantized.w8a8

Quantization vs deepseek-ai/DeepSeek-R1-Distill-Llama-70B. RedHatAI/DeepSeek-R1-Distill-Llama-70B-quantized.w8a8 is a quantized version of the other (I8/BF16).

2 other repos hold byte-for-byte identical weight files.

On the Hub · Downloads on Model Pulse