Gabriele Venturi
Gabriele Venturi
P.s. @Terrevue just added a couple of tests and made sure the linting is passing
@Terrevue the downside is: how do we make the prompt more specific, considering that the prompt is not only for charts? Do you suggest we implement a CoT to figure...
@Terrevue tried it out with a few prompts and the results are just outstanding. This makes it way more flexible and customizable. We should definitely refactor the code to have...
@Terrevue sounds great, would love it you could submit a PR with all the changes. I give it a couple of tests and seems to be working quite solidly! Super...
@DataScientistTX it seems more like an hallucination. It's showing instead of saving and assuming it's saved. Need to figure it ok
Interesting, sometimes google models tend to be very strict about minor changes. We'll try to investigate further, and we'll try to put in place a more efficient way to prevent...
@YarShev IIRC, the issues were with the SnowFlake connector and probably others as well, but I'm not sure it is highlighted from the tests, as it might be mocked (which...
@dudesparsh I believe this is actually a regression. We recently changed the behavior so that we have a unified way to handle the charts. In fact, given that many people...
Hi @Ridoy302583, you can try to pass `output_type="string"` to the chat method (https://docs.pandas-ai.com/en/latest/API/pandasai/#pandasai.smart_dataframe.SmartDataframe.chat). Let me know if that fixes the issue!
@ArslanSaleem initially reported it. Maybe he can elaborate further!