act_flip_v1_wrist

v1 expert + bare wrist camera (pose A), front+wrist input; extended 30k->60k

Jacob's from-scratch ACT (scripts/act.py, scripts/train_act.py in https://github.com/JaHeRoth/robot-learning). Checkpoints uploaded: best = step 30000, final = step 60000. Raw torch.save dicts with model_state, opt_state, stats.

step	success	95% CI
5000	0.546	[0.502, 0.589]
10000	0.794	[0.756, 0.827]
20000	0.822	[0.786, 0.853]
30000	0.962	[0.941, 0.976]
40000	0.956	[0.934, 0.971]
50000	0.958	[0.937, 0.972]
60000	0.958	[0.937, 0.972]

500 held-out seeds (800000-800499), n_action_steps 40, horizon 450