Tuning doesn't work even with Gemini Pro 1.0 for datasets containing outputs with more than 5000 characters

Gemini API
, ,
1

From the official website, it mentions the output character limit for training dataset as 5000 only for Gemini Flash 1.5 model. But I am unable to tune even with Gemini Pro 1.0 model. Important to note that I was able to do it a few months back, but it doesn’t work now. It feels like a backward step and I need to get this done asap! Any thoughts on this?

1 Like
3

Hi @Abrar,

Gemini pro 1.0 may have worked for tuning earlier, but it’s no longer listed in the official docs, so it’s likely not supported now. And yes, the 5000 character limit is documented only for Gemini Flash 1.5
You can refer to the official source here:
Gemini models | Gemini API | Google AI for Developers