Problem with context token for gpt-3.5-turbo-0125

Community
1

I finetuned a model based on gpt-3.5-turbo-0125, which is said to have 16,385 tokens. However, I got an error below when I set the max_tokens to 15000. Is there something wrong?

openai.BadRequestError: Error code: 400 - {‘error’: {‘message’: ‘max_tokens is too large: 15000. This model supports at most 4096 completion tokens, whereas you provided 15000.’, ‘type’: ‘invalid_request_error’, ‘param’: ‘max_tokens’, ‘code’: None}}

2

The good news is that nothing is wrong :slight_smile:

The maximum token parameter enables you to control the maximum number of output tokens can produce. For gpt-3.5-turbo along with most other models the maximum output tokens are limited to 4,096.

The 16,385 tokens refers to the context window, which is the sum of both the input and output tokens.