When using ChatGPT3.5 to generate code I don't have issues. However, whenever I try to do something with ChatGPT4 I get "Network error". And ChatGPT4 is the only reason why I upgraded to the pro plan. Feels like a rip off to me, please fix.
#Network error
brave hawk
This happens all the time. I couldn't get a single complete answer from ChatGPT4 since I upgraded.
sterile spindle
This happens all the time. I couldn't get a single complete answer from ChatGPT4...
It’s literally a network error on your end
brave hawk
It’s literally a network error on your end
It consistently happens only with ChatGPT4, never with 3.5. Also it seems quite unlikely that I have all these network errors, everything else works just fine, not just ChatGPT3.5.
Also, even if it were a network error on my end, why it doesn't reconnect and start streaming the response again? It just stops.
sterile spindle
Never had this isssue, been using gpt4 modes for months
brave hawk
Same with gpt3.5, that's also why I'm saying it doesn't add up. there must be something else going on.
sterile spindle
If you have a vpn, I know that can cause network errors
brave hawk
Nope, I'm not using a vpn.
The error management model should definitely be improved. An open connection should not be required anyway. An HTTP response is sent, the server should store the response and try to send it back as a Response. If a response cannot be sent back, it should still be available, it shouldn't stop generating. Once the client reconnects it can read the generated response from the server.
Making the model asynchronous makes sense, especially considering that apparently the response is stored on the server anyway because it can be read later as well (all discussions are available on the left, and they are not stored on the client).
brave hawk
