Soumendra Ganguly
Soumendra Ganguly
@wermos I completely agree with your blog post. ~~If only you could help me figure out why my Lark backend is **so** slow (see output in previous comment; I might...
Being able to take conjugate transpose while working with matrices with literal complex number entries is a must for me (major reason for this PR). The macros `\sympyi`, `\sympytr`, and...
@wermos Thank you so much for running the timing test. Looks like the Lark issue is on my end. Please do not worry about that anymore :) From your error...
@wermos It should be easy to break this up into 2 PRs. Working on all suggested changes now.
@wermos All requested changes made. Since you are the expert on Lark and are deeply familiar with the source, can you please point me toward the correct way of implementing...
@asmeurer Thank you for the feedback. It might be a specific issue with my setup that is causing Lark to be **that** slow. Anyway, I have removed the ANTLR code...
@asmeurer I am a number theory postdoc and currently have very limited time, so I cannot be the *principal maintainer* of any repository. That being said, I am really interested...
@asmeurer @wermos @sylee957
Great questions. > Is locally trained model is a direction we should take, or is prompt engineering a direction we should take? | | Cloud | Local | | :---...
To make the process more democratic, I think we should vote after considering the above table of pros and cons. To vote for *local LLMs*, please react with a :+1:...