RedHatAI/granite-3.1-8b-instruct-FP8-dynamic

Quantization vs ibm-granite/granite-3.1-8b-instruct. RedHatAI/granite-3.1-8b-instruct-FP8-dynamic is a quantized version of the other (F8_E4M3/BF16).

On the Hub · Downloads on Model Pulse