Show Your Work: Gemma2 Process Rewards
Google Tunix hackathon submission. GRPO and hybrid process rewards so Gemma2 2B shows its reasoning.
Sanjay J
Machine Learning Engineer
Data science and AI Systems.
Machine Learning Engineer
Data science and AI Systems.
Google Tunix hackathon submission. GRPO and hybrid process rewards so Gemma2 2B shows its reasoning.
Decoder-only transformer in PyTorch with GQA, RoPE, RMSNorm, and SwiGLU. Trained on Tiny Shakespeare.
Vision Transformer trained from scratch on CIFAR-10. Train, eval, inference, Gradio demo, Hub push.