• Hello, I am having an issue when configuring my chatGPT AI tool with GPT 3.5 Turbo. After a certain amount of back and forth messaging with the AI tool on my platform, I began receiving this error:

    This model’s maximum context length is 4097 tokens. However, you requested 5369 tokens (3869 in the messages, 1500 in the completion). Please reduce the length of the messages or completion.

    The AI is using embeddings of post page’s post content.


    Switching to GPT 3.5 Turbo 16k had alleviated the issue and the AI began picking up responses. I assume this is because the token max limit was changed to 16k. But I fear even as the conversations keep accumulating through the AI tool, that will reach it’s limit.

    Is there a way to maybe flush the conversation of the AI or segment the embeddings so I can go back to just using GPT 3.5 Turbo?

Viewing 2 replies - 1 through 2 (of 2 total)
  • Plugin Author senols

    (@senols)

    Hi @przalex2

    In order to resolve this issue in a long-term perspective, I recommend dividing your larger content into smaller segments. Splitting your content will help ensure you stay within the maximum token limit for your chosen model. As of now, unfortunately our plugin dont have splitting functionality yet.

    Additionally, you could consider reducing the “remember conversation up to” value. This value plays a significant role in determining the size of the context. By lowering this, you can substantially decrease the overall number of tokens being used at any given time during the conversation.

    I hope this information is helpful.

    Thread Starter przalex2

    (@przalex2)

    Thank you @senols for that information.

    Is there a splitting tool to divide up the content on my wordpress posts or do I have to perform this manually?

    The issue could be that each of my WordPress posts are around 3k tokens of text when the index builder goes over the posts for the AI vector embedding.

Viewing 2 replies - 1 through 2 (of 2 total)

The topic ‘Token Limit Exceeded Error’ is closed to new replies.