We are using the Google Gemini 1.0 Pro API in a production environment, but we have noticed extremely truncated answers for any case that we try on it, even in the vertex AI studio (but for some reason it's not acting in such a way in the https://aistudio.google.com/. We noticed this problem started somewhere around 2:40 PM ET today (3/21/2024) (Before that, it was working normally and responding with correct responses).
Here's a screen grab from the vertex ai studio (default model parameters):
UPDATE: I noticed that this problem only occurs when region is set to us-central1, it is responding correctly for other regions:
I found what I was doing wrong, I was streaming the response and only parsing the first candidate, I disabled streaming and it started giving complete responses, hope this helps.
Are you using the free API key or you had to switch to a paid one ?
User | Count |
---|---|
2 | |
1 | |
1 | |
1 | |
1 |