I built a tiny physics solver LLM that performs surprisingly well on easy-to-medium difficulty physics problems. Most LLMs today still struggle with physics QA (as PhyBench recently highlighted), so I wanted to see how far I could push a small model with careful data and minimal compute.<p>Model: Qwen3-1.7B<p>Supervised Finetuning: ~1500 curated examples spanning kinematics, EM, acoustics, and more<p>RL Fine-tuning: GRPO, 1-shot RLVR style (single example, 70 steps)<p>Total cost: ~$5 on H100<p>It started with a cold-start SFT (~3 epochs, loss to 0.3), then I ran RL with accuracy reward that climbed from 0.1 → 0.8.<p>Goal: Create a lightweight physics solver that’s small enough to deploy anywhere—think of it as a "pocket tutor" for foundational physics.<p>Still working on evaluations—most benchmarks focus on very hard problems, while I want something that evaluates basic correctness, reasoning, and unit sense on easy/medium problems. If anyone has suggestions, I’d love to hear them.