Arnav Garg
Arnav Garg is the Machine Learning Lead at Predibase, where he builds high-throughput LLM infrastructure and leads development of efficient fine-tuning techniques like Turbo LoRA. His work focuses on scalable training, fast inference, and practical ML systems for deploying custom models on user data.
![]()
Why Reinforcement Learning Beats SFT with Limited Data
How RFT Beats SFT When Data is Scarce
Read Article![]()
Turbo LoRA: 2-3x faster fine-tuned LLM inference
Turbo LoRA: 2-3x faster fine-tuned LLM inference
Read Article![]()
How to Fine-Tune LLaMA-2 on Your Own Data at Scale
How to Fine-tune LLaMa-2 with Scalable Infra
Read Article![]()
![]()
![]()
![]()
Train AI to Write GPU Code via Reinforcement Fine-Tuning
A Deep Dive into Reinforcement Fine-Tuning
Read Article![]()
Fine-Tune CodeLlama-7B to Generate Python Docstrings
Generate Docstrings with Fine-tuned CodeLlama
Read Article![]()
![]()
