ddalcu/DeepSeek-V4-Flash-0731-MLX-Serve-mixed-2-3-8bit

Quantization vs deepseek-ai/DeepSeek-V4-Flash-0731. ddalcu/DeepSeek-V4-Flash-0731-MLX-Serve-mixed-2-3-8bit is a quantized version of the other (U32/BF16).

On the Hub · Downloads on Model Pulse