ChatGPT-DeMod icon indicating copy to clipboard operation
ChatGPT-DeMod copied to clipboard

[Bug] Completely Broken

Open yodaluca23 opened this issue 1 year ago • 5 comments

With the latest update, this is completely broken, I'm using FF so it's not a ManifestV3 (as V2 is still supported, and forever will be on FF) problem, it gets moderated, and the flag shows up. The Demod menu isn't even being injected...

yodaluca23 avatar Sep 03 '24 22:09 yodaluca23

Sometimes it answers, but most times it doesn't image

yodaluca23 avatar Sep 03 '24 22:09 yodaluca23

I've released a new update to DeMod, hopefully it once again works as intended. Can you confirm the problem is gone?

4as avatar Sep 21 '24 20:09 4as

I've released a new update to DeMod, hopefully it once again works as intended. Can you confirm the problem is gone?

ChatGPT is currently down for me, as soon as it's back up I'll let you know

yodaluca23 avatar Sep 21 '24 20:09 yodaluca23

The 4.7 version still has issues and is even less stable than the 4.6 version. When a red warning is triggered, the session instantly clears without any notification. After refreshing the page, DeMod is completely disabled in the current conversation (similar to the effect of DeMod: off), and mildly NSFW content in the conversation is marked in orange. After starting a "New Chat," DeMod partially regains functionality.

DeMod has been doing a great job, at least in preventing orange warnings from appearing. Please focus on fixing the red warnings, as it seems that ChatGPT has modified the review mechanism for this part.

Regardless, your work has been fantastic and has helped me resolve the orange warnings.

worldwarii73 avatar Sep 22 '24 12:09 worldwarii73

@4as It does work! The menu shows up, and it lets me bypass flagged responses, but for blocked one's it'll show red on demod, GPT will load then all of a sudden the response shows up then like a second later it goes back to the start a chat screen...

yodaluca23 avatar Sep 23 '24 22:09 yodaluca23

New 5.0 version of DeMod has proper support for new data formatting OpenAI introduced to the chat. Interception of blocked messages should now work correctly again. If the problems persist open a new issue.

4as avatar Oct 26 '24 12:10 4as