16 5 183

Sourab Mangrulkar

smangrul

https://www.linkedin.com/in/sourab-m/

pacman100

AI & ML interests

Machine Learning, Deep Learning, Natural Language Processing, Natural Language Generation, Computer Vision, Reinforcement Learning

Articles

Making LLMs even more accessible with bitsandbytes, 4-bit quantization and QLoRA

May 24, 2023

• 45

Fine-tuning 20B LLMs with RLHF on a 24GB consumer GPU

Mar 9, 2023

• 16

🤗 PEFT: Parameter-Efficient Fine-Tuning of Billion-Scale Models on Low-Resource Hardware

Feb 10, 2023

• 19

Accelerate Large Model Training using DeepSpeed

Jun 28, 2022

Accelerate Large Model Training using PyTorch Fully Sharded Data Parallel

May 2, 2022

• 1

Organizations

smangrul's activity

upvoted 3 articles about 2 months ago

Article

Welcome Llama 3 - Meta's new open LLM

Apr 18

• 249

Article

Introducing Idefics2: A Powerful 8B Vision-Language Model for the community

Apr 15

• 141

Article

DS-MoE: Making MoE Models More Efficient and Less Memory-Intensive

•

Apr 9

• 27

upvoted a collection about 2 months ago

Meta Llama 3

Collection

This collection hosts the transformers and original repos of the Meta Llama 3 and Llama Guard 2 releases • 5 items • Updated Apr 18 • 584

upvoted a collection 6 months ago

PEFT papers

Collection

A collection of methods that have been implemented in the 🤗 PEFT library • 12 items • Updated Jan 30 • 15

Sourab Mangrulkar

AI & ML interests

Articles

GaLore: Advancing Large Model Training on Consumer-grade Hardware

🤗 PEFT welcomes new merging methods

Mixture of Experts Explained

Personal Copilot: Train Your Own Coding Assistant

Fine-tuning Llama 2 70B using PyTorch FSDP

The Falcon has landed in the Hugging Face ecosystem

Making LLMs even more accessible with bitsandbytes, 4-bit quantization and QLoRA

Fine-tuning 20B LLMs with RLHF on a 24GB consumer GPU

🤗 PEFT: Parameter-Efficient Fine-Tuning of Billion-Scale Models on Low-Resource Hardware

Accelerate Large Model Training using DeepSpeed

Accelerate Large Model Training using PyTorch Fully Sharded Data Parallel

Organizations

smangrul's activity

Welcome Llama 3 - Meta's new open LLM

Introducing Idefics2: A Powerful 8B Vision-Language Model for the community

DS-MoE: Making MoE Models More Efficient and Less Memory-Intensive