Context Limits?

#5
by ArihantSheth - opened

What is the context limit for this model? I am facing the following error:
InternalServerError: 500 {"error":"Incomplete generation","error_type":"Incomplete generation"}
on an input of length 2727 characters containing 415 words. Model is deployed using VertexAI on Google Cloud using the following configuration:
Machine type: g2-standard-12
Accelerator type: NVIDIA_L4
Accelerator count: 1

ARC Lab, Tencent PCG org

We use 8192 as the maximum ctx length during training like Mistral and we did not revise its original config's maximum length. I think maybe that is an environment problem.

Alright, thanks

ArihantSheth changed discussion status to closed

Sign up or log in to comment