Upload README.md with huggingface_hub

README.md

@@ -0,0 +1,8 @@
+Top-level structure:
+
+- `RL/`
+  Trained GRPO runs and checkpoints.
+- `EVAL/`
+  Post-hoc GSM8K evaluation outputs.
+- `SFT/`
+  SFT-side artifacts and runs.