Max tokens

#39
by MatrixIA - opened

I have deployed this model on a Text Generation Inference with an Nvidia L4 路 1x GPU 路 24 GB,
and I get this error
{'error': 'Input validation error: `inputs` tokens + `max_new_tokens` must be <= 1512. Given: 6644 `inputs` tokens and 100 `max_new_tokens`', 'error_type': 'validation'}

but looking the the model config I can see that the model has : "max_position_embeddings": 16384

I'm not sure what the issue is !

Defog.ai org

Hi @MatrixIA this is a question for TGI, could you raise it on their github instead? We don't use TGI much here so we're unable to provide assistance on this unfortunately.

jp-defog changed discussion status to closed

Sign up or log in to comment