Upload README.md with huggingface_hub

README.md

@@ -70,7 +70,8 @@ only.
 | batch size | 128 |
 | this checkpoint | step 8000 (~1.24 epoch; 1 epoch = 6,444 steps) |
 | planned length | 20,000 steps |
-| lr | peak 2e-4, 1,000 warmup steps, cosine decay over 100,000 |
+| lr | cosine decay, 1,000 warmup steps: peak 2.5e-5 -> 2.5e-6 over 30,000 steps |
+| | (`CosineDecaySchedule` defaults -- `pi05_bi` does not override `lr_schedule`; the `peak_lr=2e-4 / decay_steps=100000` block in `config.py` is referenced only by `pi05_single*`) |
 | LoRA | rank 16 on the LLM, rank 32 on the action expert |
 | vision tower | **fully fine-tuned** -- the freeze filter matches only `.*llm.*` |