lifan0127 / ai-research-assistant

Aria is Your AI Research Assistant Powered by GPT Large Language Models
GNU Affero General Public License v3.0
794 stars 33 forks source link

Error messages when trying to ask for limitations of a set of refs #9

Open wshuyi opened 1 year ago

wshuyi commented 1 year ago

Hi. Thank you for providing such a great addon! I tried to ask the RA to find the limitations of the two papers. And it shows me the errors like this. Would you please look into it? Thanks! CleanShot_2023-04-12_at_09 00 47 CleanShot_2023-04-12_at_09 06 29

lifan0127 commented 1 year ago

Hi @wshuyi Yes. It has been a problem that sometimes GPT ignores the output format instructions and produces unparsible output. I asked about this issue on twitter and got some helpful suggestions that I will look into: https://twitter.com/FanLi_RnD/status/1645827968239845376

lifan0127 commented 1 year ago

Hi @wshuyi Yes. It has been a problem that sometimes GPT ignores the output format instructions and produces unparsible output. I asked about this issue on twitter and got some helpful suggestions that I will look into: https://twitter.com/FanLi_RnD/status/1645827968239845376

Here is a promising new feature in LangChain to allow auto-recovery from output format and other errors: https://github.com/hwchase17/langchain/pull/2656

lifan0127 commented 1 year ago

Hi @wshuyi I have added some logic to handle such "Unable to parse JSON response" error. Please try the latest release and let me know if it works for you.