TheBloke

/

Llama-2-70B-Chat-GPTQ

like 257

Hello,

I am trying to use the model to generate me an answer from a context I provide, but when I get to text generation, I get this error: temp_state buffer is too small.
I think it is because my prompt is quite large in terms of tokens, because when I prompt the model without context it works.

I checked and is not an issue of resources, GPU or RAM, and the Llama-2-13B-chat-GPTQ worked when prompted with context.

Does anyone have any suggestions on how to solve this?

Thanks,