Libero environment¶

Run Libero tasks in plugrl-env-client via an optional dependency group.

Install¶

Install env client with Libero extras.

pip install -e ".[libero]"

Quickstart¶

Inspect the CLI config.

plugrl-run-env-client libero-v1 --help

Run a few episodes.

plugrl-run-env-client libero-v1 --num-episodes 10 --server-host 127.0.0.1 --server-port 8000

As E11 ran it¶

E11 evaluated a pi05_libero checkpoint on LIBERO through a PlugRL server: one client process per task, ten tasks at once, with each task's initial states taken in order rather than sampled.

plugrl-run-env-client libero-v1 \
  --server-host 127.0.0.1 --server-port 8000 \
  --num-envs 1 --num-procs 10 --num-episodes 10 \
  --env.task-suite-name libero_spatial \
  --env.no-randomize-initial-state \
  --runner.pass-proc-id \
  --runner.replan-steps 5 \
  --runner.seed 7 \
  --recorder.no-thread0-only \
  --exp-name my_eval

--runner.pass-proc-id gives process i task i, so ten processes cover a ten-task suite and --num-episodes counts per task rather than in total. To hold one task fixed instead - which is what the fine-tuning runs did - drop that flag and pass --env.task-id 8 --env.randomize-initial-state.
--recorder.no-thread0-only makes every process write its own summary.json. Without it only process 0 reports, and per-task success rates cannot be recovered afterwards.
Rendering goes through EGL, so the client needs MUJOCO_GL=egl and PYOPENGL_PLATFORM=egl in its environment. Which GPU it renders on follows EGL's own device order, which need not agree with CUDA_VISIBLE_DEVICES.
LIBERO writes its own config file, so point LIBERO_CONFIG_PATH somewhere writable before the first run.

Protocol, results and the recorded environment of both processes: E11.

Verify¶

Env client can create the Libero env.
Episodes run without init errors.

Troubleshooting¶

Multi process init conflicts: try --use-env-lock.