Text Generation
Transformers
Safetensors
llama
conversational
Inference Endpoints
text-generation-inference

ExL 2-2 or ExL 2

#1
by eramax - opened

Hello
Are all of the recently released models ExL 2-2 or ExL 2 models and what are the differences?

Best,

Everything now uses the updated exllamav2 code, so technically all are -2. I've stopped naming the models -2 now that things have stabilized. Newer models require newer version of exllamav2 Python model and have better quantized model performance, particularly at lower bits.

Sign up or log in to comment