Strings for openAI models?

[protoys-webdev]

Hello again,

I just tried to parse a larger document through a cannoli and I reached the rate limit for the chosen model, which was gpt-3.5-turbo. So I thought I just switch to gpt-3.5-turbo-16k and problem solved. However, I fail at configuring the cannoli:

• what is the right string for gpt-3.5-turbo-16k?

on a similar note:

• what are temperature ranges (1 to 2?)
• are the ranges used in the plugin config and via configuration arrows the same? (I guess so but you never know...)

Thanks! :)

[blindmansion]

Yes, that's the correct string. You can use any of the strings found in the models table of the openai docs (https://platform.openai.com/docs/models/gpt-3-5). Note that choice and field arrows only work when the model is capable of function-calling (also be found on the docs)

I had thought the temperature range is between 0-2, but the docs are saying it's 0-1, so I'm not exactly sure. Might have to do some trial and error to figure that out.

The settings panel setting will be overridden by the config arrow if you have one.

[protoys-webdev]

Thanks for the temperatures info!
could you tell me what I am doing wrong regarding the models (see below)?

[protoys-webdev]

Just to be sure I am not just using the wrong table, I also tried it with gpt-3.5-turbo-1106

[blindmansion]

Yes this is a bug that's been raised by others, you're using the strings correctly. I still need to get in and write a fix.

Sorry it's been some wait on these bugs, I've had trouble finding time the last few weeks.

[protoys-webdev]

No problem. Wasn't aware this is an already found bug.

[blindmansion]

Finally fixed this! The newest release has support for all the new openai language models.