Add `MultiLerobotDataset` for training with multiple `LeRobotDataset`s (#229)

lerobot/common/datasets/compute_stats.py

@@ -16,17 +16,15 @@
 from copy import deepcopy
 from math import ceil
 
-import datasets
 import einops
 import torch
 import tqdm
 from datasets import Image
 
-from lerobot.common.datasets.lerobot_dataset import LeRobotDataset
 from lerobot.common.datasets.video_utils import VideoFrame
 
 
-def get_stats_einops_patterns(dataset: LeRobotDataset | datasets.Dataset, num_workers=0):
+def get_stats_einops_patterns(dataset, num_workers=0):
     """These einops patterns will be used to aggregate batches and compute statistics.
 
     Note: We assume the images are in channel first format
@@ -66,9 +64,8 @@ def get_stats_einops_patterns(dataset: LeRobotDataset | datasets.Dataset, num_wo
     return stats_patterns
 
 
-def compute_stats(
+def compute_stats(dataset, batch_size=32, num_workers=16, max_num_samples=None):
-    dataset: LeRobotDataset | datasets.Dataset, batch_size=32, num_workers=16, max_num_samples=None
+    """Compute mean/std and min/max statistics of all data keys in a LeRobotDataset."""
-):
     if max_num_samples is None:
         max_num_samples = len(dataset)
 
@@ -159,3 +156,54 @@ def compute_stats(
             "min": min[key],
         }
     return stats
+
+
+def aggregate_stats(ls_datasets) -> dict[str, torch.Tensor]:
+    """Aggregate stats of multiple LeRobot datasets into one set of stats without recomputing from scratch.
+
+    The final stats will have the union of all data keys from each of the datasets.
+
+    The final stats will have the union of all data keys from each of the datasets. For instance:
+    - new_max = max(max_dataset_0, max_dataset_1, ...)
+    - new_min = min(min_dataset_0, min_dataset_1, ...)
+    - new_mean = (mean of all data)
+    - new_std = (std of all data)
+    """
+    data_keys = set()
+    for dataset in ls_datasets:
+        data_keys.update(dataset.stats.keys())
+    stats = {k: {} for k in data_keys}
+    for data_key in data_keys:
+        for stat_key in ["min", "max"]:
+            # compute `max(dataset_0["max"], dataset_1["max"], ...)`
+            stats[data_key][stat_key] = einops.reduce(
+                torch.stack([d.stats[data_key][stat_key] for d in ls_datasets if data_key in d.stats], dim=0),
+                "n ... -> ...",
+                stat_key,
+            )
+        total_samples = sum(d.num_samples for d in ls_datasets if data_key in d.stats)
+        # Compute the "sum" statistic by multiplying each mean by the number of samples in the respective
+        # dataset, then divide by total_samples to get the overall "mean".
+        # NOTE: the brackets around (d.num_samples / total_samples) are needed tor minimize the risk of
+        # numerical overflow!
+        stats[data_key]["mean"] = sum(
+            d.stats[data_key]["mean"] * (d.num_samples / total_samples)
+            for d in ls_datasets
+            if data_key in d.stats
+        )
+        # The derivation for standard deviation is a little more involved but is much in the same spirit as
+        # the computation of the mean.
+        # Given two sets of data where the statistics are known:
+        # σ_combined = sqrt[ (n1 * (σ1^2 + d1^2) + n2 * (σ2^2 + d2^2)) / (n1 + n2) ]
+        # where d1 = μ1 - μ_combined, d2 = μ2 - μ_combined
+        # NOTE: the brackets around (d.num_samples / total_samples) are needed tor minimize the risk of
+        # numerical overflow!
+        stats[data_key]["std"] = torch.sqrt(
+            sum(
+                (d.stats[data_key]["std"] ** 2 + (d.stats[data_key]["mean"] - stats[data_key]["mean"]) ** 2)
+                * (d.num_samples / total_samples)
+                for d in ls_datasets
+                if data_key in d.stats
+            )
+        )
+    return stats

lerobot/common/datasets/factory.py

@@ -16,9 +16,9 @@
 import logging
 
 import torch
-from omegaconf import OmegaConf
+from omegaconf import ListConfig, OmegaConf
 
-from lerobot.common.datasets.lerobot_dataset import LeRobotDataset
+from lerobot.common.datasets.lerobot_dataset import LeRobotDataset, MultiLeRobotDataset
 
 
 def resolve_delta_timestamps(cfg):
@@ -35,11 +35,27 @@ def resolve_delta_timestamps(cfg):
                 cfg.training.delta_timestamps[key] = eval(delta_timestamps[key])
 
 
-def make_dataset(
+def make_dataset(cfg, split: str = "train") -> LeRobotDataset | MultiLeRobotDataset:
-    cfg,
+    """
-    split="train",
+    Args:
-):
+        cfg: A Hydra config as per the LeRobot config scheme.
-    if cfg.env.name not in cfg.dataset_repo_id:
+        split: Select the data subset used to create an instance of LeRobotDataset.
+            All datasets hosted on [lerobot](https://huggingface.co/lerobot) contain only one subset: "train".
+            Thus, by default, `split="train"` selects all the available data. `split` aims to work like the
+            slicer in the hugging face datasets:
+            https://huggingface.co/docs/datasets/v2.19.0/loading#slice-splits
+            As of now, it only supports `split="train[:n]"` to load the first n frames of the dataset or
+            `split="train[n:]"` to load the last n frames. For instance `split="train[:1000]"`.
+    Returns:
+        The LeRobotDataset.
+    """
+    if not isinstance(cfg.dataset_repo_id, (str, ListConfig)):
+        raise ValueError(
+            "Expected cfg.dataset_repo_id to be either a single string to load one dataset or a list of "
+            "strings to load multiple datasets."
+        )
+
+    if isinstance(cfg.dataset_repo_id, str) and cfg.env.name not in cfg.dataset_repo_id:
         logging.warning(
             f"There might be a mismatch between your training dataset ({cfg.dataset_repo_id=}) and your "
             f"environment ({cfg.env.name=})."
@@ -49,11 +65,16 @@ def make_dataset(
 
     # TODO(rcadene): add data augmentations
 
+    if isinstance(cfg.dataset_repo_id, str):
         dataset = LeRobotDataset(
             cfg.dataset_repo_id,
             split=split,
             delta_timestamps=cfg.training.get("delta_timestamps"),
         )
+    else:
+        dataset = MultiLeRobotDataset(
+            cfg.dataset_repo_id, split=split, delta_timestamps=cfg.training.get("delta_timestamps")
+        )
 
     if cfg.get("override_dataset_stats"):
         for key, stats_dict in cfg.override_dataset_stats.items():

lerobot/common/datasets/lerobot_dataset.py

@@ -13,12 +13,16 @@
 # 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 logging
 import os
 from pathlib import Path
+from typing import Callable
 
 import datasets
 import torch
+import torch.utils
 
+from lerobot.common.datasets.compute_stats import aggregate_stats
 from lerobot.common.datasets.utils import (
     calculate_episode_data_index,
     load_episode_data_index,
@@ -42,7 +46,7 @@ class LeRobotDataset(torch.utils.data.Dataset):
         version: str | None = CODEBASE_VERSION,
         root: Path | None = DATA_DIR,
         split: str = "train",
-        transform: callable = None,
+        transform: Callable | None = None,
         delta_timestamps: dict[list[float]] | None = None,
     ):
         super().__init__()
@@ -171,7 +175,7 @@ class LeRobotDataset(torch.utils.data.Dataset):
     @classmethod
     def from_preloaded(
         cls,
-        repo_id: str,
+        repo_id: str = "from_preloaded",
         version: str | None = CODEBASE_VERSION,
         root: Path | None = None,
         split: str = "train",
@@ -183,7 +187,15 @@ class LeRobotDataset(torch.utils.data.Dataset):
         stats=None,
         info=None,
         videos_dir=None,
-    ):
+    ) -> "LeRobotDataset":
+        """Create a LeRobot Dataset from existing data and attributes instead of loading from the filesystem.
+
+        It is especially useful when converting raw data into LeRobotDataset before saving the dataset
+        on the filesystem or uploading to the hub.
+
+        Note: Meta-data attributes like `repo_id`, `version`, `root`, etc are optional and potentially
+        meaningless depending on the downstream usage of the return dataset.
+        """
         # create an empty object of type LeRobotDataset
         obj = cls.__new__(cls)
         obj.repo_id = repo_id
@@ -195,6 +207,192 @@ class LeRobotDataset(torch.utils.data.Dataset):
         obj.hf_dataset = hf_dataset
         obj.episode_data_index = episode_data_index
         obj.stats = stats
-        obj.info = info
+        obj.info = info if info is not None else {}
         obj.videos_dir = videos_dir
         return obj
+
+
+class MultiLeRobotDataset(torch.utils.data.Dataset):
+    """A dataset consisting of multiple underlying `LeRobotDataset`s.
+
+    The underlying `LeRobotDataset`s are effectively concatenated, and this class adopts much of the API
+    structure of `LeRobotDataset`.
+    """
+
+    def __init__(
+        self,
+        repo_ids: list[str],
+        version: str | None = CODEBASE_VERSION,
+        root: Path | None = DATA_DIR,
+        split: str = "train",
+        transform: Callable | None = None,
+        delta_timestamps: dict[list[float]] | None = None,
+    ):
+        super().__init__()
+        self.repo_ids = repo_ids
+        # Construct the underlying datasets passing everything but `transform` and `delta_timestamps` which
+        # are handled by this class.
+        self._datasets = [
+            LeRobotDataset(
+                repo_id,
+                version=version,
+                root=root,
+                split=split,
+                delta_timestamps=delta_timestamps,
+                transform=transform,
+            )
+            for repo_id in repo_ids
+        ]
+        # Check that some properties are consistent across datasets. Note: We may relax some of these
+        # consistency requirements in future iterations of this class.
+        for repo_id, dataset in zip(self.repo_ids, self._datasets, strict=True):
+            if dataset.info != self._datasets[0].info:
+                raise ValueError(
+                    f"Detected a mismatch in dataset info between {self.repo_ids[0]} and {repo_id}. This is "
+                    "not yet supported."
+                )
+        # Disable any data keys that are not common across all of the datasets. Note: we may relax this
+        # restriction in future iterations of this class. For now, this is necessary at least for being able
+        # to use PyTorch's default DataLoader collate function.
+        self.disabled_data_keys = set()
+        intersection_data_keys = set(self._datasets[0].hf_dataset.features)
+        for dataset in self._datasets:
+            intersection_data_keys.intersection_update(dataset.hf_dataset.features)
+        if len(intersection_data_keys) == 0:
+            raise RuntimeError(
+                "Multiple datasets were provided but they had no keys common to all of them. The "
+                "multi-dataset functionality currently only keeps common keys."
+            )
+        for repo_id, dataset in zip(self.repo_ids, self._datasets, strict=True):
+            extra_keys = set(dataset.hf_dataset.features).difference(intersection_data_keys)
+            logging.warning(
+                f"keys {extra_keys} of {repo_id} were disabled as they are not contained in all the "
+                "other datasets."
+            )
+            self.disabled_data_keys.update(extra_keys)
+
+        self.version = version
+        self.root = root
+        self.split = split
+        self.transform = transform
+        self.delta_timestamps = delta_timestamps
+        self.stats = aggregate_stats(self._datasets)
+
+    @property
+    def repo_id_to_index(self):
+        """Return a mapping from dataset repo_id to a dataset index automatically created by this class.
+
+        This index is incorporated as a data key in the dictionary returned by `__getitem__`.
+        """
+        return {repo_id: i for i, repo_id in enumerate(self.repo_ids)}
+
+    @property
+    def repo_index_to_id(self):
+        """Return the inverse mapping if repo_id_to_index."""
+        return {v: k for k, v in self.repo_id_to_index}
+
+    @property
+    def fps(self) -> int:
+        """Frames per second used during data collection.
+
+        NOTE: Fow now, this relies on a check in __init__ to make sure all sub-datasets have the same info.
+        """
+        return self._datasets[0].info["fps"]
+
+    @property
+    def video(self) -> bool:
+        """Returns True if this dataset loads video frames from mp4 files.
+
+        Returns False if it only loads images from png files.
+
+        NOTE: Fow now, this relies on a check in __init__ to make sure all sub-datasets have the same info.
+        """
+        return self._datasets[0].info.get("video", False)
+
+    @property
+    def features(self) -> datasets.Features:
+        features = {}
+        for dataset in self._datasets:
+            features.update({k: v for k, v in dataset.features.items() if k not in self.disabled_data_keys})
+        return features
+
+    @property
+    def camera_keys(self) -> list[str]:
+        """Keys to access image and video stream from cameras."""
+        keys = []
+        for key, feats in self.features.items():
+            if isinstance(feats, (datasets.Image, VideoFrame)):
+                keys.append(key)
+        return keys
+
+    @property
+    def video_frame_keys(self) -> list[str]:
+        """Keys to access video frames that requires to be decoded into images.
+
+        Note: It is empty if the dataset contains images only,
+        or equal to `self.cameras` if the dataset contains videos only,
+        or can even be a subset of `self.cameras` in a case of a mixed image/video dataset.
+        """
+        video_frame_keys = []
+        for key, feats in self.features.items():
+            if isinstance(feats, VideoFrame):
+                video_frame_keys.append(key)
+        return video_frame_keys
+
+    @property
+    def num_samples(self) -> int:
+        """Number of samples/frames."""
+        return sum(d.num_samples for d in self._datasets)
+
+    @property
+    def num_episodes(self) -> int:
+        """Number of episodes."""
+        return sum(d.num_episodes for d in self._datasets)
+
+    @property
+    def tolerance_s(self) -> float:
+        """Tolerance in seconds used to discard loaded frames when their timestamps
+        are not close enough from the requested frames. It is only used when `delta_timestamps`
+        is provided or when loading video frames from mp4 files.
+        """
+        # 1e-4 to account for possible numerical error
+        return 1 / self.fps - 1e-4
+
+    def __len__(self):
+        return self.num_samples
+
+    def __getitem__(self, idx: int) -> dict[str, torch.Tensor]:
+        if idx >= len(self):
+            raise IndexError(f"Index {idx} out of bounds.")
+        # Determine which dataset to get an item from based on the index.
+        start_idx = 0
+        dataset_idx = 0
+        for dataset in self._datasets:
+            if idx >= start_idx + dataset.num_samples:
+                start_idx += dataset.num_samples
+                dataset_idx += 1
+            break
+        else:
+            raise AssertionError("We expect the loop to break out as long as the index is within bounds.")
+        item = self._datasets[dataset_idx][idx - start_idx]
+        item["dataset_index"] = torch.tensor(dataset_idx)
+        for data_key in self.disabled_data_keys:
+            if data_key in item:
+                del item[data_key]
+        return item
+
+    def __repr__(self):
+        return (
+            f"{self.__class__.__name__}(\n"
+            f"  Repository IDs: '{self.repo_ids}',\n"
+            f"  Version: '{self.version}',\n"
+            f"  Split: '{self.split}',\n"
+            f"  Number of Samples: {self.num_samples},\n"
+            f"  Number of Episodes: {self.num_episodes},\n"
+            f"  Type: {'video (.mp4)' if self.video else 'image (.png)'},\n"
+            f"  Recorded Frames per Second: {self.fps},\n"
+            f"  Camera Keys: {self.camera_keys},\n"
+            f"  Video Frame Keys: {self.video_frame_keys if self.video else 'N/A'},\n"
+            f"  Transformations: {self.transform},\n"
+            f")"
+        )

lerobot/common/datasets/utils.py

@@ -59,7 +59,7 @@ def unflatten_dict(d, sep="/"):
     return outdict
 
 
-def hf_transform_to_torch(items_dict):
+def hf_transform_to_torch(items_dict: dict[torch.Tensor | None]):
     """Get a transform function that convert items from Hugging Face dataset (pyarrow)
     to torch tensors. Importantly, images are converted from PIL, which corresponds to
     a channel last representation (h w c) of uint8 type, to a torch image representation
@@ -73,6 +73,8 @@ def hf_transform_to_torch(items_dict):
         elif isinstance(first_item, dict) and "path" in first_item and "timestamp" in first_item:
             # video frame will be processed downstream
             pass
+        elif first_item is None:
+            pass
         else:
             items_dict[key] = [torch.tensor(x) for x in items_dict[key]]
     return items_dict
@@ -318,8 +320,7 @@ def calculate_episode_data_index(hf_dataset: datasets.Dataset) -> Dict[str, torc
 
 
 def reset_episode_index(hf_dataset: datasets.Dataset) -> datasets.Dataset:
-    """
+    """Reset the `episode_index` of the provided HuggingFace Dataset.
-    Reset the `episode_index` of the provided HuggingFace Dataset.
 
     `episode_data_index` (and related functionality such as `load_previous_and_future_frames`) requires the
     `episode_index` to be sorted, continuous (1,1,1 and not 1,2,1) and start at 0.
@@ -338,6 +339,7 @@ def reset_episode_index(hf_dataset: datasets.Dataset) -> datasets.Dataset:
         return example
 
     hf_dataset = hf_dataset.map(modify_ep_idx_func)
+
     return hf_dataset
 
 

lerobot/configs/default.yaml

@@ -23,6 +23,10 @@ use_amp: false
 # `seed` is used for training (eg: model initialization, dataset shuffling)
 # AND for the evaluation environments.
 seed: ???
+# You may provide a list of datasets here. `train.py` creates them all and concatenates them. Note: only data
+# keys common between the datasets are kept. Each dataset gets and additional transform that inserts the
+# "dataset_index" into the returned item. The index mapping is made according to the order in which the
+# datsets are provided.
 dataset_repo_id: lerobot/pusht
 
 training:

lerobot/scripts/push_dataset_to_hub.py

@@ -71,9 +71,9 @@ import torch
 from huggingface_hub import HfApi
 from safetensors.torch import save_file
 
+from lerobot.common.datasets.compute_stats import compute_stats
 from lerobot.common.datasets.lerobot_dataset import CODEBASE_VERSION, LeRobotDataset
 from lerobot.common.datasets.push_dataset_to_hub._download_raw import download_raw
-from lerobot.common.datasets.push_dataset_to_hub.compute_stats import compute_stats
 from lerobot.common.datasets.utils import flatten_dict
 
 

lerobot/scripts/train.py

@@ -16,7 +16,6 @@
 import logging
 import time
 from contextlib import nullcontext
-from copy import deepcopy
 from pathlib import Path
 from pprint import pformat
 
@@ -28,6 +27,7 @@ from termcolor import colored
 from torch.cuda.amp import GradScaler
 
 from lerobot.common.datasets.factory import make_dataset, resolve_delta_timestamps
+from lerobot.common.datasets.lerobot_dataset import MultiLeRobotDataset
 from lerobot.common.datasets.utils import cycle
 from lerobot.common.envs.factory import make_env
 from lerobot.common.logger import Logger, log_output_dir
@@ -280,6 +280,11 @@ def train(cfg: DictConfig, out_dir: str | None = None, job_name: str | None = No
 
     logging.info("make_dataset")
     offline_dataset = make_dataset(cfg)
+    if isinstance(offline_dataset, MultiLeRobotDataset):
+        logging.info(
+            "Multiple datasets were provided. Applied the following index mapping to the provided datasets: "
+            f"{pformat(offline_dataset.repo_id_to_index , indent=2)}"
+        )
 
     # Create environment used for evaluating checkpoints during training on simulation data.
     # On real-world data, no need to create an environment as evaluations are done outside train.py,
@@ -330,7 +335,7 @@ def train(cfg: DictConfig, out_dir: str | None = None, job_name: str | None = No
                     max_episodes_rendered=4,
                     start_seed=cfg.seed,
                 )
-            log_eval_info(logger, eval_info["aggregated"], step, cfg, offline_dataset, is_offline)
+            log_eval_info(logger, eval_info["aggregated"], step, cfg, offline_dataset, is_offline=True)
             if cfg.wandb.enable:
                 logger.log_video(eval_info["video_paths"][0], step, mode="eval")
             logging.info("Resume training")
@@ -362,7 +367,6 @@ def train(cfg: DictConfig, out_dir: str | None = None, job_name: str | None = No
     dl_iter = cycle(dataloader)
 
     policy.train()
-    is_offline = True
     for _ in range(step, cfg.training.offline_steps):
         if step == 0:
             logging.info("Start offline training on a fixed dataset")
@@ -382,7 +386,7 @@ def train(cfg: DictConfig, out_dir: str | None = None, job_name: str | None = No
         )
 
         if step % cfg.training.log_freq == 0:
-            log_train_info(logger, train_info, step, cfg, offline_dataset, is_offline)
+            log_train_info(logger, train_info, step, cfg, offline_dataset, is_offline=True)
 
         # Note: evaluate_and_checkpoint_if_needed happens **after** the `step`th training update has completed,
         # so we pass in step + 1.
@@ -390,41 +394,8 @@ def train(cfg: DictConfig, out_dir: str | None = None, job_name: str | None = No
 
         step += 1
 
-    logging.info("End of offline training")
-
-    if cfg.training.online_steps == 0:
-        if cfg.training.eval_freq > 0:
     eval_env.close()
-        return
+    logging.info("End of training")
-
-    # create an env dedicated to online episodes collection from policy rollout
-    online_training_env = make_env(cfg, n_envs=1)
-
-    # create an empty online dataset similar to offline dataset
-    online_dataset = deepcopy(offline_dataset)
-    online_dataset.hf_dataset = {}
-    online_dataset.episode_data_index = {}
-
-    # create dataloader for online training
-    concat_dataset = torch.utils.data.ConcatDataset([offline_dataset, online_dataset])
-    weights = [1.0] * len(concat_dataset)
-    sampler = torch.utils.data.WeightedRandomSampler(
-        weights, num_samples=len(concat_dataset), replacement=True
-    )
-    dataloader = torch.utils.data.DataLoader(
-        concat_dataset,
-        num_workers=4,
-        batch_size=cfg.training.batch_size,
-        sampler=sampler,
-        pin_memory=device.type != "cpu",
-        drop_last=False,
-    )
-
-    logging.info("End of online training")
-
-    if cfg.training.eval_freq > 0:
-        eval_env.close()
-    online_training_env.close()
 
 
 @hydra.main(version_base="1.2", config_name="default", config_path="../configs")

tests/test_datasets.py

@@ -25,26 +25,34 @@ from datasets import Dataset
 from safetensors.torch import load_file
 
 import lerobot
-from lerobot.common.datasets.factory import make_dataset
+from lerobot.common.datasets.compute_stats import (
-from lerobot.common.datasets.lerobot_dataset import (
+    aggregate_stats,
-    LeRobotDataset,
-)
-from lerobot.common.datasets.push_dataset_to_hub.compute_stats import (
     compute_stats,
     get_stats_einops_patterns,
 )
+from lerobot.common.datasets.factory import make_dataset
+from lerobot.common.datasets.lerobot_dataset import LeRobotDataset
 from lerobot.common.datasets.utils import (
     flatten_dict,
     hf_transform_to_torch,
     load_previous_and_future_frames,
     unflatten_dict,
 )
-from lerobot.common.utils.utils import init_hydra_config
+from lerobot.common.utils.utils import init_hydra_config, seeded_context
 from tests.utils import DEFAULT_CONFIG_PATH, DEVICE
 
 
-@pytest.mark.parametrize("env_name, repo_id, policy_name", lerobot.env_dataset_policy_triplets)
+@pytest.mark.parametrize(
+    "env_name, repo_id, policy_name",
+    lerobot.env_dataset_policy_triplets
+    + [("aloha", ["lerobot/aloha_sim_insertion_human", "lerobot/aloha_sim_transfer_cube_human"], "act")],
+)
 def test_factory(env_name, repo_id, policy_name):
+    """
+    Tests that:
+        - we can create a dataset with the factory.
+        - for a commonly used set of data keys, the data dimensions are correct.
+    """
     cfg = init_hydra_config(
         DEFAULT_CONFIG_PATH,
         overrides=[
@@ -315,3 +323,31 @@ def test_backward_compatibility(repo_id):
     # i = dataset.episode_data_index["to"][-1].item()
     # load_and_compare(i - 2)
     # load_and_compare(i - 1)
+
+
+def test_aggregate_stats():
+    """Makes 3 basic datasets and checks that aggregate stats are computed correctly."""
+    with seeded_context(0):
+        data_a = torch.rand(30, dtype=torch.float32)
+        data_b = torch.rand(20, dtype=torch.float32)
+        data_c = torch.rand(20, dtype=torch.float32)
+
+    hf_dataset_1 = Dataset.from_dict(
+        {"a": data_a[:10], "b": data_b[:10], "c": data_c[:10], "index": torch.arange(10)}
+    )
+    hf_dataset_1.set_transform(hf_transform_to_torch)
+    hf_dataset_2 = Dataset.from_dict({"a": data_a[10:20], "b": data_b[10:], "index": torch.arange(10)})
+    hf_dataset_2.set_transform(hf_transform_to_torch)
+    hf_dataset_3 = Dataset.from_dict({"a": data_a[20:], "c": data_c[10:], "index": torch.arange(10)})
+    hf_dataset_3.set_transform(hf_transform_to_torch)
+    dataset_1 = LeRobotDataset.from_preloaded("d1", hf_dataset=hf_dataset_1)
+    dataset_1.stats = compute_stats(dataset_1, batch_size=len(hf_dataset_1), num_workers=0)
+    dataset_2 = LeRobotDataset.from_preloaded("d2", hf_dataset=hf_dataset_2)
+    dataset_2.stats = compute_stats(dataset_2, batch_size=len(hf_dataset_2), num_workers=0)
+    dataset_3 = LeRobotDataset.from_preloaded("d3", hf_dataset=hf_dataset_3)
+    dataset_3.stats = compute_stats(dataset_3, batch_size=len(hf_dataset_3), num_workers=0)
+    stats = aggregate_stats([dataset_1, dataset_2, dataset_3])
+    for data_key, data in zip(["a", "b", "c"], [data_a, data_b, data_c], strict=True):
+        for agg_fn in ["mean", "min", "max"]:
+            assert torch.allclose(stats[data_key][agg_fn], einops.reduce(data, "n -> 1", agg_fn))
+        assert torch.allclose(stats[data_key]["std"], torch.std(data, correction=0))