I’m Salman Mohammadi. I love to learn about new things.
Artificial intelligence could be the most transformative technology ever created. I’d like to make sure it has a positive impact on humanity. See some of my writing below:
nanocode: the best Claude Code that $200 can buy. An end-to-end pretraining, SFT, and DPO library written in pure JAX for TPUs.
The theory of Proximal Policy Optimization implementations
Accelerate ND-Parallel: A guide to Efficient Multi-GPU Training
Liger GRPO meets TRL
Training Large Language Models with Interpreter Feedback using WebAssembly
Process Reward Models