/metalearning

This is a repository and github pages website deployment for my work on the mechanistic analysis of out-of-context meta-learning in LLMs

Primary LanguageSCSSMIT LicenseMIT

A mechanistic analysis of out of context meta-learning in large language models

Link to our arxiv paper

Highlights

  • Key result 1
  • Key result 2
  • Key result 3

Resesarch notes

I am experimenting with making my research notes public.

I believe that it could be helpful to write not just for myself but for an audience interested in the subject, as that may help raise the level of my writing and note taking habits, and also because I want to highlight the importance of getting one's hands dirty.

The final paper and the product looks nice and polished, but the journey of making it so is never so. There's a lot of experimentation, a lot of studying, a lot of conversations, and a lot of frustration. Many new researchers get intimidated by the legendary papers and don't get to see the process of making them into those legendary papers. I want to change that.

You can find the research notes and all the thinking I did for this project at Meta-learning out-of-context research notes