my_act_pusht_seed1000

PushT, reference recipe, training seed 1000, continued 100k->200k

Jacob's from-scratch ACT (scripts/act.py, scripts/train_act.py in https://github.com/JaHeRoth/robot-learning). Checkpoints uploaded: best = step 100000, final = step 200000. Raw torch.save dicts with model_state, opt_state, stats.

step	success	95% CI
20000	0.223	[0.211, 0.235]
40000	0.457	[0.443, 0.470]
60000	0.513	[0.499, 0.527]
80000	0.570	[0.556, 0.584]
100000	0.621	[0.608, 0.635]
120000	0.591	[0.578, 0.605]
140000	0.621	[0.608, 0.635]
160000	0.609	[0.596, 0.623]
180000	0.620	[0.606, 0.633]
200000	0.606	[0.593, 0.620]

5000 held-out episodes (seeds 1000-5999), n_action_steps 16