Fine-tuned gpt-3.5-turbo-0125 Max Output Tokens

API
1

The fine-tuning documentation seems to imply that the fine-tuned model has the same token limits as the model upon which it’s trained:

For gpt-3.5-turbo-0125 , the maximum context length is 16,385 so each training example is also limited to 16,385 tokens.

The documentation states that for gpt-3.5-turbo-0125:

Returns a maximum of 4,096 output tokens.

However, when I test my model in the playground, I cannot set the maximum output length to anything higher than 2048.

Can someone confirm the maximum output tokens for a model fine tuned from gpt-3.5-turbo-0125?

1 Like
2

The playground max_tokens slider maximum is what is incorrect.

1 Like