Sumanth R Hegde SumanthRH

shipit

Software Engineer, Anyscale. Intensity is all you need.

Achievements

SumanthRH/README.md

😄 I'm Sumanth, a software engineer at Anyscale, working on post-training. My primary interests are broadly in machine learning and systems engineering.
🚀 I'm trying to understand generative models, and have worked on finetuning and in-context learning for language models. Addicted to compute 🤖
💻 I'm currently working on SkyThought, and SkyRL.
🌱 I'm trying to learn what it takes to build machine learning systems in practice.
✨ I have a blog: https://sumanthrh.com
💬 Some samples of my writing:
- Everything about distributed training and efficient finetuning: https://sumanthrh.com/post/distributed-and-efficient-finetuning/
- Everything about tokenization (mini-course): https://github.com/SumanthRH/tokenization

NovaSky-AI/SkyRL NovaSky-AI/SkyRL Public

SkyRL: A Modular Full-stack RL Library for LLMs

Python 2.5k 486
NovaSky-AI/SkyThought NovaSky-AI/SkyThought Public

Sky-T1: Train your own O1 preview model within $450

Python 3.4k 348
tokenization tokenization Public

A comprehensive deep dive into the world of tokens

Python 231 11
frankxwang/dpo-prefix-sharing frankxwang/dpo-prefix-sharing Public

DPO, but faster 🚀

Python 51 5
peft peft Public

Forked from huggingface/peft

Fork of 🤗 PEFT: State-of-the-art Parameter-Efficient Fine-Tuning. Our implementation for IA3, a new fine-tuning method is now a part of the official Huggingface library!

Python
varun19299/deep-atrous-guided-filter varun19299/deep-atrous-guided-filter Public

Deep Atrous Guided Filter for Image Restoration in Under Display Cameras (UDC Challenge, ECCV 2020).

Jupyter Notebook 39 6

Something went wrong, please refresh the page to try again.
If the problem persists, check the GitHub status page or contact support.

There was an error while loading. Please reload this page.