ydy0615/lerobot-clone
1
0
Fork 0
mirror of https://github.com/huggingface/lerobot.git synced 2026-05-31 19:01:28 +00:00
Code Issues Packages Projects Releases Wiki Activity

Added logging for interventions to monitor the rate of interventions through time

Browse Source 
Added an s keyboard command to force success in the case the reward classifier fails

Co-authored-by: Adil Zouitine <adilzouitinegm@gmail.com>
This commit is contained in:
Michel Aractingi
2025-02-13 11:04:49 +01:00
parent 5d6879d93a
commit ee820859d3
5 changed files with 45 additions and 18 deletions
Download Patch File Download Diff File Expand all files Collapse all files

4
lerobot/scripts/server/actor_server.py
View File

@@ -223,6 +223,7 @@ def act_with_policy(cfg: DictConfig, robot: Robot, reward_classifier: nn.Module)
sum_reward_episode = 0
list_transition_to_send_to_learner = []
list_policy_time = []
episode_intervention = False
for interaction_step in range(cfg.training.online_steps):
if interaction_step >= cfg.training.online_step_before_learning:
@@ -252,6 +253,7 @@ def act_with_policy(cfg: DictConfig, robot: Robot, reward_classifier: nn.Module)
if info["is_intervention"]:
# TODO: Check the shape
action = info["action_intervention"]
episode_intervention = True
# Check for NaN values in observations
for key, tensor in obs.items():
@@ -295,11 +297,13 @@ def act_with_policy(cfg: DictConfig, robot: Robot, reward_classifier: nn.Module)
interaction_message={
"Episodic reward": sum_reward_episode,
"Interaction step": interaction_step,
"Episode intervention": int(episode_intervention),
**stats,
}
)
)
sum_reward_episode = 0.0
episode_intervention = False
obs, info = online_env.reset()