streaming in completion does not have finish_reason and total_tokens in response | OpenAI | Page 1

violet widget Mar 15, 2023, 9:16 AM

#

streaming in completion does not have finish_reason and total_tokens in response. Is there any any way to get this enabled.

knotty geyser Mar 15, 2023, 4:31 PM

#

each streaming event represents 1 token, so you can determine the tokens that way

knotty geyser Mar 15, 2023, 4:31 PM

#

There is no finish_reason provided during the middle of streaming because it has no knowledge of when it finished

violet widget Mar 15, 2023, 4:41 PM

#

@knotty geyser even at end of stream also we are not getting the finish_reason.

#streaming in completion does not have finish_reason and total_tokens in response