JPQ24/Natural-Synthesis-8b-3.1-v3-16bit

Quantization vs unsloth/Meta-Llama-3.1-8B-Instruct-bnb-4bit. unsloth/Meta-Llama-3.1-8B-Instruct-bnb-4bit is a quantized version of the other (U8/BF16).

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

On the Hub · Downloads on Model Pulse