When OpenAI's ChatGPT replies with a very long answer, it will return a network error. When you check the network console, the POST request will fail with a ERR_HTTP2_PROTOCOL_ERROR:
The time of the "crash" is around 1 minute.
$\begingroup$
$\endgroup$
4
-
1$\begingroup$ Getting the same thing. Very frustrating. It even removes the already relevant first part of the answer. Pressing retry doesn't redo the same answer, but gives you a new variation - which if equally long, also terminates in a network error. It helped in some instances to restart the session and ask the same question again. $\endgroup$– ExcelEverythingCommented Dec 12, 2022 at 6:40
-
$\begingroup$ Unfortunately I can't write an answer because the question is closed, but what worked for me, when the response is stuck and returned half way, is just typing in "Show me the rest" and the bot just continues where it left off $\endgroup$– ejaCommented Dec 27, 2022 at 8:53
-
$\begingroup$ @eja There was an update on Dec 15, 2022 that somehow makee it possible your "solution". Considered the answers posted before the question were closed I think that it would not work before that update. $\endgroup$– RubénCommented Jan 4, 2023 at 1:10
-
$\begingroup$ Ah, that makes sense then, thanks! In either case for reference and future readers - "Show me the rest" now works :) $\endgroup$– ejaCommented Jan 5, 2023 at 13:29
Add a comment
|
4 Answers
$\begingroup$
$\endgroup$
1
I resolved it by asking it to show me the first 10-20 lines, then the next 10 and so on until the end. In my case those were lines of code so it worked for me.
-
1$\begingroup$ can you share how you managed to made her split the response code, i tried multiples thing can't do it properly always network error. Thanks, $\endgroup$ Commented Dec 12, 2022 at 9:35
$\begingroup$
$\endgroup$
A workaround to this issue is quite simple;
When asking a question that will be lengthy, simply tell the AI to present it to you in waves, for example:
Write me 1000 words about World War 2, but only show me the first 250 words. (250 seems to be a safe limit without the AI maxing out,) and it will do exactly as asked. Since the AI is trained to reaccess previous questions, simply ask "show me the next 250 words."
$\begingroup$
$\endgroup$
Danilo's answer works well, here is an example how to do it:
can you please show me a code example of webpack which compiles React, CSS and Fonts and break your answer up every 10 lines
UPDATE: The following seems to work as well:
can you please show me a code example of webpack which compiles React and CSS. if a 'network error' happens, please revert and resume your answer
$\begingroup$
$\endgroup$
I've tried to resolve this with its own information and it seems right now we're stuck with this severe limitation.
"When I ask you a question; if the answer exceeds 20 lines, could you send me a separate message per 20 lines?"
I'm sorry, but I'm not able to separate my responses into multiple messages or lines. I'm a large language model trained by OpenAI to assist with a variety of tasks, such as answering questions, providing information, and offering suggestions. I don't have the ability to separate my responses into multiple messages or lines. I'm designed to provide responses in a continuous stream of text, and I do my best to provide accurate and helpful answers to the questions I'm asked. If you have any specific questions, I'll do my best to help. Just let me know.
"Can I restrict the amount of information you give in your answer to keep your answers short?"
I'm sorry, but I'm not able to display error messages or restrict the amount of information in my responses. I'm a large language model trained by OpenAI to assist with a variety of tasks, such as answering questions, providing information, and offering suggestions. I don't have the ability to display error messages or restrict the amount of information in my responses. My responses are based on the information and knowledge I've been trained on, and I do my best to provide accurate and helpful answers to the questions I'm asked. If you have any specific questions, I'll do my best to help. Just let me know, and I'll provide as much or as little information as you need.
