Datasets:

THULab
/

so101_eval2_merged_25prompts

Exception:    SplitsNotFoundError
Message:      The split names could not be parsed from the dataset config.
Traceback:    Traceback (most recent call last):
                File "tsfile/tsfile_py_cpp.pyx", line 567, in tsfile.tsfile_py_cpp.tsfile_reader_new_c
              tsfile.exceptions.FileOpenError: 28: 
              
              The above exception was the direct cause of the following exception:
              
              Traceback (most recent call last):
                File "/usr/local/lib/python3.14/site-packages/datasets/inspect.py", line 286, in get_dataset_config_info
                  for split_generator in builder._split_generators(
                                         ~~~~~~~~~~~~~~~~~~~~~~~~~^
                      StreamingDownloadManager(base_path=builder.base_path, download_config=download_config)
                      ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
                  )
                  ^
                File "/usr/local/lib/python3.14/site-packages/datasets/packaged_modules/tsfile/tsfile.py", line 271, in _split_generators
                  scan = self._scan_metadata(all_files)
                File "/usr/local/lib/python3.14/site-packages/datasets/packaged_modules/tsfile/tsfile.py", line 318, in _scan_metadata
                  with self._open_reader(file) as reader:
                       ~~~~~~~~~~~~~~~~~^^^^^^
                File "/usr/local/lib/python3.14/site-packages/datasets/packaged_modules/tsfile/tsfile.py", line 742, in _open_reader
                  return TsFileReader(file)
                File "tsfile/tsfile_reader.pyx", line 323, in tsfile.tsfile_reader.TsFileReaderPy.__init__
              SystemError: <class '_weakrefset.WeakSet'> returned a result with an exception set
              
              The above exception was the direct cause of the following exception:
              
              Traceback (most recent call last):
                File "/src/services/worker/src/worker/job_runners/config/split_names.py", line 68, in compute_split_names_from_streaming_response
                  for split in get_dataset_split_names(
                               ~~~~~~~~~~~~~~~~~~~~~~~^
                      path=dataset,
                      ^^^^^^^^^^^^^
                      config_name=config,
                      ^^^^^^^^^^^^^^^^^^^
                      token=hf_token,
                      ^^^^^^^^^^^^^^^
                  )
                  ^
                File "/usr/local/lib/python3.14/site-packages/datasets/inspect.py", line 340, in get_dataset_split_names
                  info = get_dataset_config_info(
                      path,
                  ...<6 lines>...
                      **config_kwargs,
                  )
                File "/usr/local/lib/python3.14/site-packages/datasets/inspect.py", line 291, in get_dataset_config_info
                  raise SplitsNotFoundError("The split names could not be parsed from the dataset config.") from err
              datasets.inspect.SplitsNotFoundError: The split names could not be parsed from the dataset config.

Need help to make the dataset viewer work? Make sure to review how to configure the dataset viewer, and open a discussion for direct support.

SO101-eval2-merged (TsFile)

Apache TsFile version of danilodjor/SO101-eval2-merged-25prompts, a single LeRobot v3.0 dataset built by merging the six per-permutation csacan/SO101-eval2-{gbr,rgb,bgr,brg,grb,rbg} recordings and relabelling every episode with a 27-prompt vocabulary.

Overview

The goal is to give Eval 2's compositional banana-in-bowl task both halves of its problem in one place:

Visual colour grounding — every prompt that mentions a colour is represented across all six bowl orderings, so the model cannot memorise "blue = slot 0".
Language coverage — every recorded trajectory is paired with every prompt phrasing that resolves to the slot it actually targets, so the model sees the same motion grounded in many surface forms ("the red bowl", "the 2nd from the left", "not green and not blue", "right of the blue bowl", …).
Robot: SO-101 follower (so_follower)
Scale: 1,518 episodes, 671,081 frames, 27 tasks (prompts), 30 fps
Split: train

Schema (TsFile structure)

The TsFile table is named so101_eval2_merged_25prompts.

Role	Columns
Time	`Time`, INT64 milliseconds
TAG	`episode_index`, `task_index`
FIELD	`frame_index`, `sample_index`
FIELD	`action_0` … `action_5`, FLOAT
FIELD	`observation_state_0` … `observation_state_5`, FLOAT

The original 6-element action and observation.state vectors are flattened into scalar FLOAT measurements. The source index column is retained as sample_index.

Conversion notes

Time = round(timestamp * 1000) with millisecond precision; time restarts inside each episode, while episode_index and task_index identify the TsFile device.
The original timestamp field is omitted because it is exactly represented by Time / 1000.
The front camera stream (observation.images.front) is video and is NOT included in this repository; it remains in the source videos tree.
meta/ is mirrored from the source (including tasks.parquet with the 27-prompt vocabulary). Aside from the redundant timestamp column and the excluded videos, no source rows or numeric fields are dropped.

Read example

from tsfile import TsFileReader

with TsFileReader("data/so101_eval2_merged_25prompts.tsfile") as reader:
    print(reader.get_all_table_schemas()["so101_eval2_merged_25prompts"])

Source & license

Original dataset: https://huggingface.co/datasets/danilodjor/SO101-eval2-merged-25prompts
Author: danilodjor
License: apache-2.0

Downloads last month: 87

Total file size:

14.3 MB