Paper length exceeds maximum context length

talkingwallace / ChatGPT-Paper-Reader

This repo offers a simple interface that helps you to read&summerize research papers in pdf format. You can ask some questions after reading. This interface is developed based on openai API and using GPT-3.5-turbo model.

739 stars 110 forks source link

Paper length exceeds maximum context length #21

Open zhaozewang opened 1 year ago

zhaozewang commented 1 year ago

Got this error: openai.error.InvalidRequestError: This model's maximum context length is 4097 tokens. However, your messages resulted in 7328 tokens. Please reduce the length of the messages.

Is it possible to break the paper into multiple pieces then query one paper piece at a time to avoid this issue?

zhaozewang commented 1 year ago

The issue seems to be cause by the context length, not the paper length. Here's a similar problem https://github.com/hwchase17/langchain/issues/2133.

cc-zehao commented 1 year ago

the same question

Chapoly1305 commented 1 year ago

@cc-zehao @zhaozewang

You may consider modify your "model_interface.py" and use gpt-4-32k instead. I modified and it can process much more than using gpt3.5.

Available options: gpt-3.5-turbo, gpt-4, and gpt-4-32k

REF: https://platform.openai.com/docs/models/continuous-model-upgrades