Add libero (#1950)

* add libero

* backup

* [pre-commit.ci] auto fixes from pre-commit.com hooks

for more information, see https://pre-commit.ci

* add

* [pre-commit.ci] auto fixes from pre-commit.com hooks

for more information, see https://pre-commit.ci

* add factory

* Add LIBERO as a submodule

* add changes

* [pre-commit.ci] auto fixes from pre-commit.com hooks

for more information, see https://pre-commit.ci

* add multitask

* remove photos

* [pre-commit.ci] auto fixes from pre-commit.com hooks

for more information, see https://pre-commit.ci

* bug remove

* [pre-commit.ci] auto fixes from pre-commit.com hooks

for more information, see https://pre-commit.ci

* fix video paths and train.py

* [pre-commit.ci] auto fixes from pre-commit.com hooks

for more information, see https://pre-commit.ci

* fix renaming issues with cams

* [pre-commit.ci] auto fixes from pre-commit.com hooks

for more information, see https://pre-commit.ci

* Update .gitignore

* final refactor/fix

* [pre-commit.ci] auto fixes from pre-commit.com hooks

for more information, see https://pre-commit.ci

* add safethread support

* ad

* Add dep (#4)

* Add 'libero' dependencies to pyproject.toml

* Add Git dependencies for egl_probe and LIBERO

* Update libero-requirements.txt

* add future dep

* update bash

* quick fix

* remove step1

* cleanup (#5)

* [pre-commit.ci] auto fixes from pre-commit.com hooks

for more information, see https://pre-commit.ci

* clean

* [pre-commit.ci] auto fixes from pre-commit.com hooks

for more information, see https://pre-commit.ci

* cleanup 2

* improve install

* Delete libero-requirements.txt

* iterate on review

* [pre-commit.ci] auto fixes from pre-commit.com hooks

for more information, see https://pre-commit.ci

* add docs for eval

* [pre-commit.ci] auto fixes from pre-commit.com hooks

for more information, see https://pre-commit.ci

* doc

* [pre-commit.ci] auto fixes from pre-commit.com hooks

for more information, see https://pre-commit.ci

* update doc

* [pre-commit.ci] auto fixes from pre-commit.com hooks

for more information, see https://pre-commit.ci

* remove brkpt

* fix docs

* update docs/script

* update doc

* skip test warning

* [pre-commit.ci] auto fixes from pre-commit.com hooks

for more information, see https://pre-commit.ci

* hotfix: flip actions

* add train

* new things

* More things

* add new changes

* iterate on review

* [pre-commit.ci] auto fixes from pre-commit.com hooks

for more information, see https://pre-commit.ci

* remove files

* [pre-commit.ci] auto fixes from pre-commit.com hooks

for more information, see https://pre-commit.ci

* remove unces

* [pre-commit.ci] auto fixes from pre-commit.com hooks

for more information, see https://pre-commit.ci

* remove

* [pre-commit.ci] auto fixes from pre-commit.com hooks

for more information, see https://pre-commit.ci

* new changes

* iterate on review

* remove files

* factor

* update installation

* doc title

* make it reproducible

* [pre-commit.ci] auto fixes from pre-commit.com hooks

for more information, see https://pre-commit.ci

* iterate on review

* [pre-commit.ci] auto fixes from pre-commit.com hooks

for more information, see https://pre-commit.ci

* remove files

* update tests

* update doc

* [pre-commit.ci] auto fixes from pre-commit.com hooks

for more information, see https://pre-commit.ci

* remove sh files

* add gym

---------

Signed-off-by: Jade Choghari <chogharijade@gmail.com>
Co-authored-by: pre-commit-ci[bot] <66853113+pre-commit-ci[bot]@users.noreply.github.com>
Co-authored-by: Jade Choghari (jchoghar) <chogharijade@gmai.com>
Co-authored-by: Adil Zouitine <adilzouitinegm@gmail.com>

tests/envs/test_envs.py

@@ -46,7 +46,10 @@ def test_env(env_name, env_task, obs_type):
 @require_env
 def test_factory(env_name):
     cfg = make_env_config(env_name)
-    env = make_env(cfg, n_envs=1)
+    envs = make_env(cfg, n_envs=1)
+    suite_name = next(iter(envs))
+    task_id = next(iter(envs[suite_name]))
+    env = envs[suite_name][task_id]
     obs, _ = env.reset()
     obs = preprocess_observation(obs)
 

tests/policies/test_policies.py

@@ -158,7 +158,7 @@ def test_policy(ds_repo_id, env_name, env_kwargs, policy_name, policy_kwargs):
     assert isinstance(policy, PreTrainedPolicy)
 
     # Check that we run select_actions and get the appropriate output.
-    env = make_env(train_cfg.env, n_envs=2)
+    envs = make_env(train_cfg.env, n_envs=2)
 
     dataloader = torch.utils.data.DataLoader(
         dataset,
@@ -187,6 +187,13 @@ def test_policy(ds_repo_id, env_name, env_kwargs, policy_name, policy_kwargs):
 
     # reset the policy and environment
     policy.reset()
+    # For testing purposes, we only need a single environment instance.
+    # So here we unwrap the first suite_name and first task_id to grab
+    # the actual env object (SyncVectorEnv) that exposes `.reset()`.
+    suite_name = next(iter(envs))
+    task_id = next(iter(envs[suite_name]))
+    env = envs[suite_name][task_id]
+
     observation, _ = env.reset(seed=train_cfg.seed)
 
     # apply transform to normalize the observations