The dataset viewer is not available for this subset.
Exception: SplitsNotFoundError
Message: The split names could not be parsed from the dataset config.
Traceback: 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/webdataset/webdataset.py", line 84, in _split_generators
raise ValueError(
...<2 lines>...
)
ValueError: The TAR archives of the dataset should be in WebDataset format, but the files in the archive don't share the same prefix or the same types.
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.
AIMO Interpretability Challenge: heavy artifacts
Large, non-code artifacts (traces, extracted features, white-box results, labelled datasets, run outputs) behind our submissions to the AIMO Interpretability Challenge (Small Models track).
Code, technical report and the two nominated submissions live on GitHub: https://github.com/RKResident/AIMO-Interp-Challenge
That repo is the entry point. This dataset holds the data its scripts read and write; the GitHub repo's .gitignore lists the
folders that were kept out of git, and every one of them is in the archives below.
Layout
Everything is stored as zstd-compressed tar archives, split into 2 GB parts, under archives/:
archives/<name>.tar.zst.part000, .part001, ...
Each archive is a tar of one top-level folder of the project root (paths inside the tar start with that folder name), so extracting into the root of a clone of the GitHub repo puts the data back where the scripts expect it.
| Archive | Contents (top-level folder(s)) |
|---|---|
misc |
small folders and loose files: attention_grids, casebook, half_completions, olympiad*, patching_results, saved_attention_eigenvalues, saved_trajectories_v2, solution_validity_comp, sonnet_prompts, submission_logs, submitted_solutions, t1_run, extra_dataset, vast_packages, datasets, loose CSV/parquet/zip files |
results_graphs |
results_graphs/ |
analysis |
analysis/ (incl. scripts withheld from GitHub, pickles and CSVs) |
hard_pool |
hard_pool/ |
results_mainv2_t1 |
results_mainv2_t1/ |
results_whitebox |
results_whitebox/ |
saved_trajectories_aggregate_data |
saved_trajectories_aggregate_data/ |
saved_trajectories_v2_data |
saved_trajectories_v2_data/ |
data |
data/ |
extra_deepsek |
extra_deepsek/ |
extraction_out_generations_pearl_more |
extraction_out_generations_pearl_more/ |
extraction_out |
extraction_out/ |
ec2_results |
ec2_results/ |
extraction_out_generations_p2 |
extraction_out_generations_p2/ |
extra_olympiad_problems_dataset |
extra_olympiad_problems_dataset/ |
Not included (kept local only): virtualenv, IDE files, caches, the competition's getting-started/ data, third-party papers,
internal handoff / leaderboard notes, *.log files, and zips that duplicate an uploaded folder.
How to get it back
Needs zstd and hf (pip install -U huggingface_hub).
# 1. clone the code
git clone https://github.com/RKResident/AIMO-Interp-Challenge
cd AIMO-Interp-Challenge
# 2. download the archives (all, or just the ones you need via --include)
hf download Sabyasachi19/aimo-interp-artifacts --repo-type dataset --local-dir ../aimo_hf
# one archive only:
# hf download Sabyasachi19/aimo-interp-artifacts --repo-type dataset --include "archives/results_whitebox.*" --local-dir ../aimo_hf
# 3. extract into the repo root (parts concatenate back into one stream)
for n in results_whitebox analysis data; do # pick names from the table above
cat ../aimo_hf/archives/$n.tar.zst.part* | zstd -d | tar -xf -
done
Extraction needs free disk equal to the uncompressed size (about 38 GB for everything). ec2_results,
extra_olympiad_problems_dataset and extraction_out_generations_p2 are the largest (~4-5 GB raw each).
To list an archive without extracting: cat <name>.tar.zst.part* | zstd -d | tar -tf - | less.
- Downloads last month
- -