admin/lerobot
1
0
Fork 0
mirror of https://github.com/huggingface/lerobot.git synced 2026-07-26 19:26:16 +00:00
Code Issues Packages Projects Releases Wiki Activity

fix(vlabench): drop gripper [-1, 1] remap + remove env unit test

Browse Source 
Addresses PR #3396 review feedback from @s1lent4gnt:

- Remove `_normalize_gripper_action`. The dataset actions[6] column is
  already in [0, 1] (verified against
  VLABench/vlabench_primitive_ft_lerobot_video), and there is no older
  LeRobot VLABench integration to stay compatible with — this is a new
  benchmark. Replace with a plain `np.clip(action[6], 0.0, 1.0)` at the
  single call site.
- Remove tests/envs/test_vlabench_env.py. The smoke test in the CI
  benchmark job already exercises the env end-to-end; the unit test
  only covered the now-removed remap helper.

Made-with: Cursor
This commit is contained in:
Pepijn
2026-04-20 17:41:46 +02:00
parent a9adc66caf
commit 0c08c6ce18
2 changed files with 2 additions and 56 deletions
Download Patch File Download Diff File Expand all files Collapse all files
tests/envs/test_vlabench_env.py
-40
View File
@@ -1,40 +0,0 @@
#!/usr/bin/env python
# Copyright 2025 The HuggingFace Inc. team. All rights reserved.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
import numpy as np
from lerobot.envs.vlabench import ACTION_HIGH, ACTION_LOW, VLABenchEnv
def test_vlabench_action_space_uses_zero_to_one_gripper_channel():
env = VLABenchEnv()
np.testing.assert_array_equal(env.action_space.low, ACTION_LOW)
np.testing.assert_array_equal(env.action_space.high, ACTION_HIGH)
assert env.action_space.shape == (7,)
assert env.action_space.low[-1] == 0.0
assert env.action_space.high[-1] == 1.0
np.testing.assert_array_equal(env.action_space.low[:6], np.full(6, -1.0, dtype=np.float32))
np.testing.assert_array_equal(env.action_space.high[:6], np.full(6, 1.0, dtype=np.float32))
def test_vlabench_gripper_action_normalization_keeps_backward_compatibility():
assert VLABenchEnv._normalize_gripper_action(-1.0) == 0.0
assert VLABenchEnv._normalize_gripper_action(-0.5) == 0.25
assert VLABenchEnv._normalize_gripper_action(0.0) == 0.0
assert VLABenchEnv._normalize_gripper_action(0.5) == 0.5
assert VLABenchEnv._normalize_gripper_action(1.0) == 1.0
assert VLABenchEnv._normalize_gripper_action(1.5) == 1.0