Datasetpublic

PranayTest/piper-speedup-evals

PiPER speed-up evals Real-robot policy rollouts on the PiPER bimanual cell (plate pick, hand-over, place), run from the operator stack's Policy panel at execution speed multipliers of 1x to 6x. One folder per policy, one sub-folder per run (start time, IST): the three camera clips (top, left-arm, right-arm; re-encoded from the recorded chunks with every frame at its capture time, so they play in real time; clip t = 0 is 1 s before the run start and frame_ts.json lists each… See the full description on the dataset page: https://huggingface.co/datasets/PranayTest/piper-speedup-evals.

sourceHugging Faceupdated 12d agoView on Hugging Face

0likes335downloads

Dataset Card

PiPER speed-up evals

Real-robot policy rollouts on the PiPER bimanual cell (plate pick, hand-over, place), run from the operator stack's Policy panel at execution speed multipliers of 1x to 6x. One folder per policy, one sub-folder per run (start time, IST): the three camera clips (top, left-arm, right-arm; re-encoded from the recorded chunks with every frame at its capture time, so they play in real time; clip t = 0 is 1 s before the run start and frame_ts.json lists each frame's capture time per camera), telemetry.jsonl / telemetry.parquet (every BigQuery signal in the window, including the policy's own policy_command_positions / policy_gripper_command) and manifest.json.

manifest.json carries the run parameters (params.speed is the time-scaling multiplier, 1.0 = the checkpoint's training speed), the gripper command events, the hand-over times (right close → left close → right release, seconds after the start), the peak commanded and measured joint speeds per arm and the stop reason. index.json is the same table as below. Telemetry shows commands and joint feedback, not whether the plate physically transferred.

Runs: 31; hand-over sequences completed: 10. Fastest complete hand-over: act_PhubR_c30_50k_best_reach_v6 at speed 6.0 on 2026-09-10T03:13:42.563000+05:30, right release at 8.9 s.

start (IST)	policy	speed	re-plan s	length s	R close	L close	R release	complete	R cmd/meas peak deg/s	stop
02:14:32	`act_PhubR_50k_hub_prior`	1.0	0.50	4.0	—	—	—	no	44.7 / 59.3	dashboard stop
02:16:08	`act_PhubR_50k_hub_prior`	1.0	0.50	71.5	19.8	36.1	39.6	yes	18.6 / 26.1	dashboard stop
02:18:32	`act_PhubR_50k_hub_prior`	3.0	0.50	15.5	11.4	—	—	no	36.8 / 45.7	dashboard stop
02:19:30	`act_PhubR_50k_hub_prior`	3.0	0.50	32.9	7.2	—	—	no	46.0 / 104.5	dashboard stop
02:20:40	`act_PhubR_50k_hub_prior`	3.0	0.50	19.7	—	—	—	no	30.0 / 32.3	dashboard stop
02:21:49	`act_PhubR_50k_hub_prior`	1.0	0.50	79.6	28.9	—	—	no	34.5 / 26.2	dashboard stop
02:24:02	`act_PhubR_50k_hub_prior`	3.0	0.50	16.1	10.6	—	—	no	35.9 / 38.6	dashboard stop
02:26:28	`act_PhubR_aug_50k_best_placement_v6`	3.0	0.50	2.1	—	—	—	no	46.4 / 0.0	3 consecutive chunks rejected
02:27:02	`act_PhubR_aug_50k_best_placement_v6`	3.0	0.50	8.5	—	—	—	no	44.9 / 0.0	3 consecutive chunks rejected
02:27:52	`act_PhubR_c30_50k_best_reach_v6`	1.0	0.50	12.6	—	—	—	no	11.4 / 0.0	dashboard stop
02:28:53	`act_PhubR_50k_hub_prior`	1.0	0.50	2.1	—	—	—	no	27.6 / 0.0	dashboard stop
02:29:41	`act_PhubR_aug_50k_best_placement_v6`	3.0	0.50	3.8	—	—	—	no	48.5 / 0.0	dashboard stop
02:31:33	`act_PhubR_aug_50k_best_placement_v6`	3.0	0.50	28.2	5.8	13.0	13.8	yes	50.2 / 53.3	dashboard stop
02:37:15	`act_PhubR_aug_50k_best_placement_v6`	6.0	0.50	14.8	4.3	11.6	12.9	yes	77.4 / 91.6	dashboard stop
02:41:06	`act_PhubR_aug_50k_best_placement_v6`	6.0	0.50	7.8	5.0	—	—	no	79.3 / 85.9	dashboard stop
02:44:11	`act_PhubR_aug_50k_best_placement_v6`	6.0	0.60	11.6	4.8	—	—	no	78.3 / 71.9	dashboard stop
02:45:31	`act_PhubR_aug_50k_best_placement_v6`	3.0	0.50	19.1	5.0	12.9	13.0	yes	51.9 / 48.9	dashboard stop
02:46:51	`act_PhubR_aug_50k_best_placement_v6`	4.0	0.50	19.5	5.0	—	—	no	58.1 / 82.4	dashboard stop
02:47:52	`act_PhubR_aug_50k_best_placement_v6`	4.0	0.50	18.5	5.1	11.6	12.3	yes	52.8 / 65.6	dashboard stop
02:51:32	`act_PhubR_aug_50k_best_placement_v6`	5.0	0.50	8.0	5.0	—	—	no	64.1 / 68.9	dashboard stop
02:52:32	`act_PhubR_aug_50k_best_placement_v6`	4.5	0.50	14.9	4.3	11.5	12.9	yes	60.2 / 60.6	dashboard stop
02:53:11	`act_PhubR_aug_50k_best_placement_v6`	4.0	0.50	13.4	4.3	11.4	—	no	100.7 / 60.0	dashboard stop
02:54:40	`act_PhubR_aug_50k_best_placement_v6`	4.0	0.50	26.9	5.1	—	—	no	55.6 / 59.7	dashboard stop
02:56:34	`act_PhubR_c30_50k_best_reach_v6`	4.0	0.50	12.8	—	—	—	no	56.4 / 62.3	dashboard stop
03:00:26	`act_PhubR_c30_50k_best_reach_v6`	4.0	0.10	27.1	11.9	—	—	no	47.2 / 52.7	dashboard stop
03:02:00	`act_PhubR_c30_50k_best_reach_v6`	5.0	0.10	16.7	4.7	12.7	13.3	yes	67.5 / 62.3	dashboard stop
03:05:58	`act_PhubR_c30_50k_best_reach_v6`	5.0	0.05	14.6	4.3	12.1	12.7	yes	69.4 / 63.8	dashboard stop
03:07:32	`act_PhubR_c30_50k_best_reach_v6`	5.0	0.05	16.9	4.1	—	—	no	61.8 / 63.4	dashboard stop
03:12:32	`smolvla_PhubR30_b_20k`	2.0	0.10	5.4	—	—	—	no	88.9 / 43.1	dashboard stop
03:13:42	`act_PhubR_c30_50k_best_reach_v6`	6.0	0.05	11.3	3.6	8.4	8.9	yes	73.3 / 92.0	dashboard stop
03:16:39	`act_PhubR_c30_50k_best_reach_v6`	6.0	0.05	12.3	3.5	9.2	9.7	yes	85.2 / 114.8	dashboard stop

Built by episode_pipeline/export_policy_runs.py from BigQuery + GCS; repo PranayTest/piper-speedup-evals.