Thanks! :) I tried it with something that was coming up for me.
Ray:"How do I do 2d plots of polygons using complex numbers as coordinates"
It worked with explanations!
Unfortunately the code didn't work. Almost though, so we (wolfia.com)
worked on it for a couple of rounds.
We gave up, but I am really impressed. It did self-correct a couple of
things but they weren't the core problem.
No complaints about the performance though; I have a tendency to
uncover bugs LOL
I would appreciate if somebody on the group would try the query and
answer, and if a sagemath/sage bug is found, tell me how to file it.
I would have copied the transcript, but it's disappeared!
Or get it to work:) I am working up a problem on a social media site
and want to illustrate a polygon formula for constructing edge midpoint
polygons from polygons; and reverse midpoint polygons to "full polygons.
In case you think I am lazy; I am.
Let me tell you, some of the manual sections are over 10,000 pages :)
System
Ubuntu mate-local build-python 3.10.6-Sagemat 10.0
No known problems
Both "sage" and "/opt/sage-10.0/sage --notebook jupyterlab"
Had the same error.
This is about a more "grounded" version of ChatGPT + Sage that is now at
https://wolfia.com/?projectId=ed005166-99f2-4e5e-8778-2dc49d4bf930
---------- Forwarded message ---------
From: William Stein<wst...@sagemath.com>
Date: Fri, Jun 30, 2023 at 11:41 AM
Subject: Ask about the SageMath codebase using ChatGPT
To: sage-devel<sage-de...@googlegroups.com>
Cc:<na...@wolfia.com>
Hi,
Summary: Try asking anything about the SageMath codebase at
https://wolfia.com/?projectId=ed005166-99f2-4e5e-8778-2dc49d4bf930
Details:
There's a project called "Wolfia" that does the following with a codebase:
"Wolfia Codex works by indexing an entire codebase. This is done by
first chunking each file into smaller code snippets. Those snippets
and other metadata are then persisted using vector embeddings.
When you ask a question about a codebase, that question is also
converted into a vector embedding. This is compared to all of the
available code and metadata to find the most relevant code.
The relevant code is then inlined as context to various prompts
(general questions, debugging errors, or documentation generation) and
sent to a Large Language Model to generate an answer to your question
based on the provided code."
I asked them to index the sagemath codebase and they did. You can try it here
https://wolfia.com/?projectId=ed005166-99f2-4e5e-8778-2dc49d4bf930
The point is that instead of using GPT's vague memory of what it might
have seen, this instead uses the actual Sage source code, and mainly
uses GPT to make sense of it.
---------- Forwarded message ---------
From: Naren Manoharan<na...@wolfia.com>
Date: Fri, Jun 30, 2023 at 10:53 AM
Subject: Thank you for signing up and Welcome Wolfia!
To:<wst...@sagemath.com>
Hey William,
Thanks for signing up and requesting us to index -
https://github.com/sagemath/sage
This is now available here -
https://wolfia.com/?projectId=ed005166-99f2-4e5e-8778-2dc49d4bf930
Feel free to grab time here if you have any questions or feedback -
https://calendly.com/naren-mano/wolfia
I would love to hear any feedback or thoughts that you have! I'm a big
fan of open core projects. Let me know if I can help in any way.
Best,
Naren
Co-Founder & CTO at Wolfia (YC S22)
(314) 562-8039 | LinkedIn | Twitter
--
Best Regards,
William Stein
CEO, SageMath, Inc.
https://cocalc.com
--
You received this message because you are subscribed to the Google Groups
"sage-support" group.
To unsubscribe from this group and stop receiving emails from it, send an email
to sage-support+unsubscr...@googlegroups.com.
To view this discussion on the web visit
https://groups.google.com/d/msgid/sage-support/08c45d46-5d35-0e57-a44b-f51eb8e29e1b%40gmail.com.
