-
Notifications
You must be signed in to change notification settings - Fork 44.6k
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Logs output to console if we're not on speak mode #3715
Logs output to console if we're not on speak mode #3715
Conversation
The latest updates on your projects. Learn more about Vercel for Git ↗︎
|
Codecov ReportPatch coverage:
Additional details and impacted files@@ Coverage Diff @@
## master #3715 +/- ##
==========================================
- Coverage 62.68% 62.67% -0.01%
==========================================
Files 74 74
Lines 3398 3400 +2
Branches 494 495 +1
==========================================
+ Hits 2130 2131 +1
Misses 1120 1120
- Partials 148 149 +1
☔ View full report in Codecov by Sentry. |
Do we have an example where SPEAK actually has some more information value than THOUGHTS? Because I think with printing THOUGHTS we really don't need to print SPEAK. The idea in the prompt for SPEAK is to be used only for say_text imo. |
you don't need this if/else. Just log the thoughts. The typewriter log has internal speak logic: https://github.com/Significant-Gravitas/Auto-GPT/blob/master/autogpt/logs.py#L90 |
The LLM doesn't really know if we have speak mode enabled or not, so it often happens that it puts relevant info in that section, because that is the section that communicates with the user. I actually had the LLM asked me things over several tasks or claryfing what she was about to do better than the previous thoughts. That's why I think that info should not be lost, no reason to lose some of the output of the LLM. |
This is a mass message from the AutoGPT core team. For more details (and for infor on joining our Discord), please refer to: |
…as#3715) Co-authored-by: k-boikov <[email protected]>
I had a PR approved for this some weeks ago, but I think it got lost the very same night against another PR that did a major refactor. I think it is an important point so I am proposing it again, feel free to approve or discard of course.
Background
I noticed that if we're not on speak mode, what the assistant want to say gets lost because it is not logged to console, sometimes losing precious info
Changes
I added an "else" so that if the assistant doesn't speak directly (because we're not in speak mode), at least it is printed to console like the rest of the output.
PR Quality Checklist