chess-sim: RL pilot (ReinFlow-style PPO on SmolVLA), stopped after round 7: checks, logs, round-4 checkpoint
e534973 verified | { | |
| "policy": "models/baseline", | |
| "out": "outputs/rl_pilot", | |
| "iterations": 24, | |
| "envs": 10, | |
| "episodes": 100, | |
| "train_seed": 7700003, | |
| "val_seed": 9300003, | |
| "val_episodes": 60, | |
| "val_every": 4, | |
| "save_every": 4, | |
| "seconds": 20.0, | |
| "lr": 3e-07, | |
| "sigma_lr": 0.003, | |
| "critic_lr": 0.0003, | |
| "sigma_start": 0.2, | |
| "sigma_end": 0.002, | |
| "learn_sigma": false, | |
| "expert_bf16": false, | |
| "gamma": 0.99, | |
| "lam": 0.95, | |
| "clip": 0.2, | |
| "epochs": 2, | |
| "minibatch": 8, | |
| "target_kl": 0.02, | |
| "vf_coef": 0.5, | |
| "max_grad_norm": 1.0, | |
| "critic_warmup": 1, | |
| "torch_threads": 2, | |
| "reward": { | |
| "success": 1.0, | |
| "clean_lift": 0.2, | |
| "disturbed": -0.5, | |
| "toppled": -0.3 | |
| } | |
| } |