Skip to content
View sumitdotml's full-sized avatar
☕
☕
115 followers · 8 following

Achievements

Achievement: QuickdrawAchievement: Pull Sharkx2Achievement: YOLOAchievement: Starstruck

Achievements

Achievement: QuickdrawAchievement: Pull Sharkx2Achievement: YOLOAchievement: Starstruck

Block or report sumitdotml

Block user

Prevent this user from interacting with your repositories and sending you notifications. Learn more about blocking users.

You must be logged in to block users.

Content in all repositories owned by your account will be closed.
Maximum 250 characters. Please don’t include any personal information such as legal names or email addresses. This note will only be visible to you.
Report abuse

Contact GitHub support about this user’s behavior. Learn more about reporting abuse.

Report abuse
sumitdotml/README.md

mle. learning how to think in matrices

i write blogs & research worklogs on my site; take a look if you're interested!

i love coffee, vim, and geoguessr. often just doing one of these if i'm not reading papers, writing code, or walking alongside rivers.

currently learning c/cuda by building and writing gpu kernels.

my git server mirrors what's here.

if you wanna reach out:

  • send me an email: myname{@}mywebsitedomain
  • dm me on twitter

Pinned Loading

  1. sumit.ml sumit.ml Public

    my site

    MDX 3

  2. moe-emergence moe-emergence Public

    a project highlighting the emergent expert specialization in Mixture of Experts (MoEs) across 3 domains: code, math, prose

    Jupyter Notebook 4

  3. lora-and-friends lora-and-friends Public

    self-driven research comparing LoRA fine-tuning approaches (baseline vs. finetuned attention-only LoRA vs. finetuned all-layer LoRA)

    Python 2

  4. dotfiles dotfiles Public

    configs for my neovim, tmux, ghostty, tiling

    Lua 99 2

  5. tinytensor tinytensor Public

    a tiny C/CUDA tensor library for ML runtime internals

    C 1

  6. transformers transformers Public

    my attempt at trying to implement as many transformer architectures as I can until I reach a level of profound understanding of everything

    Python 5