From 0ac610e02977992aca432a3b2e52e82fbed39f63 Mon Sep 17 00:00:00 2001 From: Pepijn Date: Wed, 15 Apr 2026 10:01:17 +0200 Subject: [PATCH] docs(robomme): drop prototype-branch note and move dataset to lerobot/robomme MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit - Remove the "Related work" block referencing the prototype branch feat/robomme-integration; the PR stands on its own. - Point all dataset references at lerobot/robomme (docs, env module docstring, RoboMMEEnvConfig docstring) — this is the canonical HF location once the dataset is mirrored. Co-Authored-By: Claude Opus 4.6 (1M context) --- docs/source/robomme.mdx | 13 ++++--------- src/lerobot/envs/configs.py | 2 +- src/lerobot/envs/robomme.py | 2 +- 3 files changed, 6 insertions(+), 11 deletions(-) diff --git a/docs/source/robomme.mdx b/docs/source/robomme.mdx index 8f990379c..2f66fbba8 100644 --- a/docs/source/robomme.mdx +++ b/docs/source/robomme.mdx @@ -4,14 +4,9 @@ - **16 tasks** across 4 memory-skill suites - **1,600 training demos** (100 per task, 50 val, 50 test) -- **Dataset**: [`pepijn223/robomme_data_lerobot`](https://huggingface.co/datasets/pepijn223/robomme_data_lerobot) — LeRobot v3.0, 768K frames at 10 fps +- **Dataset**: [`lerobot/robomme`](https://huggingface.co/datasets/lerobot/robomme) — LeRobot v3.0, 768K frames at 10 fps - **Simulator**: ManiSkill / SAPIEN, Panda arm, Linux only -> **Related work**: The initial integration of RoboMME into LeRobot was prototyped in -> [`feat/robomme-integration`](https://github.com/huggingface/lerobot/tree/feat/robomme-integration). -> This PR rebases that work onto `feat/async-vector-env`, adopts the `create_envs()` dispatch -> pattern, and adds the `LazyVectorEnv` helper for large task-count benchmarks. - ## Tasks | Suite | Tasks | @@ -69,12 +64,12 @@ lerobot-eval \ ## Dataset -The dataset [`pepijn223/robomme_data_lerobot`](https://huggingface.co/datasets/pepijn223/robomme_data_lerobot) is in **LeRobot v3.0 format** and can be loaded directly: +The dataset [`lerobot/robomme`](https://huggingface.co/datasets/lerobot/robomme) is in **LeRobot v3.0 format** and can be loaded directly: ```python from lerobot.common.datasets.lerobot_dataset import LeRobotDataset -dataset = LeRobotDataset("pepijn223/robomme_data_lerobot") +dataset = LeRobotDataset("lerobot/robomme") ``` ### Dataset features @@ -97,7 +92,7 @@ The env wrapper exposes `front_rgb` and `wrist_rgb` as observation keys. The `fe The dataset uses `image` and `wrist_image` as column names. When fine-tuning from the dataset, you may need to rename columns to match your policy's expected input keys: ```python -dataset = LeRobotDataset("pepijn223/robomme_data_lerobot") +dataset = LeRobotDataset("lerobot/robomme") # rename dataset columns if needed: # dataset.hf_dataset = dataset.hf_dataset.rename_column("image", "front_rgb") ``` diff --git a/src/lerobot/envs/configs.py b/src/lerobot/envs/configs.py index a6e010313..c5980ab68 100644 --- a/src/lerobot/envs/configs.py +++ b/src/lerobot/envs/configs.py @@ -582,7 +582,7 @@ class RoboMMEEnv(EnvConfig): """RoboMME memory-augmented manipulation benchmark (ManiSkill/SAPIEN). 16 tasks across 4 suites: Counting, Permanence, Reference, Imitation. - Dataset: pepijn223/robomme_data_lerobot (LeRobot v3.0, 1,600 episodes). + Dataset: lerobot/robomme (LeRobot v3.0, 1,600 episodes). Benchmark: https://github.com/RoboMME/robomme_benchmark Requires: pip install 'lerobot[robomme]' (Linux only). diff --git a/src/lerobot/envs/robomme.py b/src/lerobot/envs/robomme.py index 4242256d6..b017a3890 100644 --- a/src/lerobot/envs/robomme.py +++ b/src/lerobot/envs/robomme.py @@ -9,7 +9,7 @@ RoboMME tasks: Reference: PickHighlight, VideoRepick, VideoPlaceButton, VideoPlaceOrder Imitation: MoveCube, InsertPeg, PatternLock, RouteStick -Dataset: pepijn223/robomme_data_lerobot (LeRobot v3.0, 1,600 episodes) +Dataset: lerobot/robomme (LeRobot v3.0, 1,600 episodes) Install: pip install 'lerobot[robomme]' (Linux only) Benchmark: https://github.com/RoboMME/robomme_benchmark """