iFaz/eqm-pusht-seed3-half

EQM Policy — eqm_pusht_seed3

Trained with LeRobot.
Date: 2026-09-03 20:47
Policy type: eqm | Device: cuda

---

📦 Dataset

Parameter	Value
dataset.repo_id	lerobot/pusht

---

🏋️ Training Config

Parameter	Value
steps	140000
batch_size	8
eval_freq	0
save_freq	70000
num_workers	4
seed	3
eval.n_episodes	1
eval.batch_size	1
eval.use_async_envs	True

---

📐 Policy Architecture

All defaults — no overrides applied.

---

🎯 Eval Config

Parameter	Value
env.type	pusht
env.task	PushT-v0
eval.n_episodes	100
eval.batch_size	4
eval.use_async_envs	False
policy.path	/kaggle/working/outputs/train/pusht_seed3/checkpoints/last/pretrained_model
policy.ood_logging_enabled	True
policy.ood_calibration_stats_path	/kaggle/working/eqm_calibration.json
policy.ood_log_path	/kaggle/working/outputs/eqm_ood_log.csv
policy.ood_z_threshold	3.0

---

📊 Eval Results

Metric	Value
Episodes	100
Success rate	24.0%
Avg sum reward	63.94
Avg max reward	0.66
Eval time (s)	562.7

---

Citation

@misc{cadene2024lerobot,
  author = {Cadene, Remi and Alibert, Simon and others},
  title  = {LeRobot},
  year   = {2024},
  url    = {https://github.com/huggingface/lerobot}
}