Datasets:

SabaPivot
/

icml26-repro-conditional-coverage-diagnostics

Modalities:

Dataset Preview

Split (1)

train

The full dataset viewer is not available (click to read why). Only showing a preview of the rows.

The dataset generation failed because of a cast error

Error code:   DatasetGenerationCastError
Exception:    DatasetGenerationCastError
Message:      An error occurred while generating the dataset

All the data files must have the same columns, but at some point there are 5 new columns ({'kl_ert.1', 'true_l1.1', 'l2_ert.1', 'coverage.1', 'l1_ert.1'}) and 2 missing columns ({'q', 'rep'}).

This happened while the csv dataset builder was generating data using

hf://datasets/SabaPivot/icml26-repro-conditional-coverage-diagnostics/results/summary.csv (at revision e2caf97d7ccbb93b9ea4b45f230fc089720df5cb), ['hf://datasets/SabaPivot/icml26-repro-conditional-coverage-diagnostics@e2caf97d7ccbb93b9ea4b45f230fc089720df5cb/results/convergence.csv', 'hf://datasets/SabaPivot/icml26-repro-conditional-coverage-diagnostics@e2caf97d7ccbb93b9ea4b45f230fc089720df5cb/results/summary.csv']

Please either edit the data files to have matching columns, or separate them into different configurations (see docs at https://hf.co/docs/hub/datasets-manual-configuration#multiple-configurations)
Traceback:    Traceback (most recent call last):
                File "/usr/local/lib/python3.14/site-packages/datasets/builder.py", line 1837, in _prepare_split_single
                  writer.write_table(table)
                  ~~~~~~~~~~~~~~~~~~^^^^^^^
                File "/usr/local/lib/python3.14/site-packages/datasets/arrow_writer.py", line 765, in write_table
                  self._write_table(pa_table, writer_batch_size=writer_batch_size)
                  ~~~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
                File "/usr/local/lib/python3.14/site-packages/datasets/arrow_writer.py", line 773, in _write_table
                  pa_table = table_cast(pa_table, self._schema)
                File "/usr/local/lib/python3.14/site-packages/datasets/table.py", line 2369, in table_cast
                  return cast_table_to_schema(table, schema)
                File "/usr/local/lib/python3.14/site-packages/datasets/table.py", line 2297, in cast_table_to_schema
                  raise CastError(
                  ...<3 lines>...
                  )
              datasets.table.CastError: Couldn't cast
              n: double
              method: string
              l1_ert: string
              l1_ert.1: string
              l2_ert: string
              l2_ert.1: string
              kl_ert: string
              kl_ert.1: string
              true_l1: string
              true_l1.1: string
              coverage: string
              coverage.1: string
              -- schema metadata --
              pandas: '{"index_columns": [{"kind": "range", "name": null, "start": 0, "' + 1645
              to
              {'n': Value('int64'), 'rep': Value('int64'), 'method': Value('string'), 'coverage': Value('float64'), 'l1_ert': Value('float64'), 'l2_ert': Value('float64'), 'kl_ert': Value('float64'), 'true_l1': Value('float64'), 'q': Value('float64')}
              because column names don't match
              
              During handling of the above exception, another exception occurred:
              
              Traceback (most recent call last):
                File "/src/services/worker/src/worker/job_runners/config/parquet_and_info.py", line 1369, in compute_config_parquet_and_info_response
                  parquet_operations, partial, estimated_dataset_info = stream_convert_to_parquet(
                                                                        ~~~~~~~~~~~~~~~~~~~~~~~~~^
                      builder, max_dataset_size_bytes=max_dataset_size_bytes
                      ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
                  )
                  ^
                File "/src/services/worker/src/worker/job_runners/config/parquet_and_info.py", line 948, in stream_convert_to_parquet
                  builder._prepare_split(split_generator=splits_generators[split], file_format="parquet")
                  ~~~~~~~~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
                File "/usr/local/lib/python3.14/site-packages/datasets/builder.py", line 1683, in _prepare_split
                  for job_id, done, content in self._prepare_split_single(
                                               ~~~~~~~~~~~~~~~~~~~~~~~~~~^
                      gen_kwargs=gen_kwargs, job_id=job_id, **_prepare_split_args
                      ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
                  ):
                  ^
                File "/usr/local/lib/python3.14/site-packages/datasets/builder.py", line 1839, in _prepare_split_single
                  raise DatasetGenerationCastError.from_cast_error(
                  ...<4 lines>...
                  )
              datasets.exceptions.DatasetGenerationCastError: An error occurred while generating the dataset
              
              All the data files must have the same columns, but at some point there are 5 new columns ({'kl_ert.1', 'true_l1.1', 'l2_ert.1', 'coverage.1', 'l1_ert.1'}) and 2 missing columns ({'q', 'rep'}).
              
              This happened while the csv dataset builder was generating data using
              
              hf://datasets/SabaPivot/icml26-repro-conditional-coverage-diagnostics/results/summary.csv (at revision e2caf97d7ccbb93b9ea4b45f230fc089720df5cb), ['hf://datasets/SabaPivot/icml26-repro-conditional-coverage-diagnostics@e2caf97d7ccbb93b9ea4b45f230fc089720df5cb/results/convergence.csv', 'hf://datasets/SabaPivot/icml26-repro-conditional-coverage-diagnostics@e2caf97d7ccbb93b9ea4b45f230fc089720df5cb/results/summary.csv']
              
              Please either edit the data files to have matching columns, or separate them into different configurations (see docs at https://hf.co/docs/hub/datasets-manual-configuration#multiple-configurations)

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.

n int64	rep int64	method string	coverage float64	l1_ert float64	l2_ert float64	kl_ert float64	true_l1 float64	q float64
250	0	lightgbm	0.908	0.04	-0.010331	-0.062974	0.088224	2.375508
250	0	partitionwise	0.908	0.0188	-0.001668	-0.084235	0.088224	2.375508
250	1	lightgbm	0.92	0.0176	-0.017284	-0.077464	0.099398	2.323073
250	1	partitionwise	0.92	0.0244	-0.00236	-0.212973	0.099398	2.323073
250	2	lightgbm	0.904	0.0144	-0.007883	-0.059872	0.092814	2.380816
250	2	partitionwise	0.904	0.0056	-0.003426	-0.173506	0.092814	2.380816
250	3	lightgbm	0.888	0.032	-0.013788	-0.055321	0.100633	2.226168
250	3	partitionwise	0.888	0.0132	-0.003528	-0.099178	0.100633	2.226168
250	4	lightgbm	0.876	0.0488	-0.003141	-0.000811	0.096031	2.400625
250	4	partitionwise	0.876	0.0048	-0.004871	-0.057188	0.096031	2.400625
250	5	lightgbm	0.892	0.052	-0.009912	-0.008932	0.094117	2.463117
250	5	partitionwise	0.892	-0.0176	-0.007019	-0.149004	0.094117	2.463117
250	6	lightgbm	0.9	0.048	-0.006498	-0.041168	0.087982	2.541049
250	6	partitionwise	0.9	0.0092	-0.005467	-0.220233	0.087982	2.541049
250	7	lightgbm	0.932	0.032	-0.006474	-0.043567	0.092613	2.37756
250	7	partitionwise	0.932	0.0076	-0.002385	-0.121827	0.092613	2.37756
500	0	lightgbm	0.888	0.0676	-0.001364	-0.018113	0.095558	2.368974
500	0	partitionwise	0.888	0.0068	-0.001586	-0.008964	0.095558	2.368974
500	1	lightgbm	0.884	0.0644	0.000061	-0.01726	0.091644	2.369893
500	1	partitionwise	0.884	-0.0004	-0.002531	-0.094924	0.091644	2.369893
500	2	lightgbm	0.904	0.018	-0.016711	-0.08575	0.091081	2.372447
500	2	partitionwise	0.904	-0.0014	-0.002846	-0.056611	0.091081	2.372447
500	3	lightgbm	0.926	0.05	-0.000765	-0.025957	0.088051	2.466052
500	3	partitionwise	0.926	-0.0072	-0.001369	-0.009399	0.088051	2.466052
500	4	lightgbm	0.906	0.0348	-0.009201	-0.057181	0.091755	2.37889
500	4	partitionwise	0.906	0.0086	-0.00087	-0.006589	0.091755	2.37889
500	5	lightgbm	0.9	0.0648	0.001399	-0.01927	0.095018	2.413164
500	5	partitionwise	0.9	-0.0248	-0.002517	-0.07263	0.095018	2.413164
500	6	lightgbm	0.882	0.0244	-0.010966	-0.050888	0.093573	2.361924
500	6	partitionwise	0.882	0.0172	-0.000624	-0.001729	0.093573	2.361924
500	7	lightgbm	0.926	0.0476	-0.005361	-0.018511	0.089308	2.433761
500	7	partitionwise	0.926	0.0172	-0.000566	-0.041959	0.089308	2.433761
1,000	0	lightgbm	0.908	0.0422	-0.006181	-0.052446	0.093714	2.422211
1,000	0	partitionwise	0.908	0.0111	-0.00084	-0.006446	0.093714	2.422211
1,000	1	lightgbm	0.897	0.04	-0.001885	-0.031776	0.092538	2.438946
1,000	1	partitionwise	0.897	0.0012	-0.001411	-0.009535	0.092538	2.438946
1,000	2	lightgbm	0.9	0.0312	-0.004502	-0.044968	0.098063	2.354222
1,000	2	partitionwise	0.9	0.0128	-0.000747	-0.006151	0.098063	2.354222
1,000	3	lightgbm	0.898	0.0512	-0.000552	-0.010542	0.092025	2.432617
1,000	3	partitionwise	0.898	-0.0103	-0.002052	-0.013174	0.092025	2.432617
1,000	4	lightgbm	0.91	0.0392	-0.003348	-0.024315	0.091419	2.462242
1,000	4	partitionwise	0.91	0.005	-0.001028	-0.007127	0.091419	2.462242
1,000	5	lightgbm	0.891	0.042	-0.005071	-0.039109	0.095402	2.374227
1,000	5	partitionwise	0.891	0.001	-0.001345	-0.008849	0.095402	2.374227
1,000	6	lightgbm	0.89	0.049	-0.001951	-0.035276	0.095194	2.367235
1,000	6	partitionwise	0.89	-0.0083	-0.002011	-0.012072	0.095194	2.367235
1,000	7	lightgbm	0.889	0.0662	-0.003055	-0.00255	0.099369	2.311349
1,000	7	partitionwise	0.889	0.0062	-0.001836	-0.010211	0.099369	2.311349
2,000	0	lightgbm	0.9025	0.0675	0.000116	0.012311	0.092955	2.412973
2,000	0	partitionwise	0.9025	0.0097	-0.000019	-0.000408	0.092955	2.412973
2,000	1	lightgbm	0.9005	0.0505	-0.002126	-0.003819	0.089679	2.460187
2,000	1	partitionwise	0.9005	0.0071	-0.000467	-0.002984	0.089679	2.460187
2,000	2	lightgbm	0.902	0.0443	-0.003346	-0.011893	0.091157	2.445941
2,000	2	partitionwise	0.902	0.0084	-0.000626	-0.003961	0.091157	2.445941
2,000	3	lightgbm	0.8945	0.0608	0.004411	0.016042	0.092548	2.420959
2,000	3	partitionwise	0.8945	-0.00165	-0.000631	-0.003662	0.092548	2.420959
2,000	4	lightgbm	0.9	0.0582	0.000058	0.001006	0.093666	2.413068
2,000	4	partitionwise	0.9	0.003	-0.000471	-0.002553	0.093666	2.413068
2,000	5	lightgbm	0.909	0.0425	-0.002353	-0.009153	0.092179	2.439728
2,000	5	partitionwise	0.909	0.01705	-0.000028	-0.000547	0.092179	2.439728
2,000	6	lightgbm	0.9055	0.0445	-0.003359	-0.014525	0.09408	2.336744
2,000	6	partitionwise	0.9055	0.0203	0.000106	-0.00062	0.09408	2.336744
2,000	7	lightgbm	0.886	0.0615	0.002388	0.008869	0.096375	2.368648
2,000	7	partitionwise	0.886	0.018	0.000216	0.000998	0.096375	2.368648
5,000	0	lightgbm	0.8934	0.07588	0.005166	0.03854	0.095079	2.359781
5,000	0	partitionwise	0.8934	0.00664	-0.000187	-0.000983	0.095079	2.359781
5,000	1	lightgbm	0.908	0.06752	0.004421	0.037016	0.093079	2.395296
5,000	1	partitionwise	0.908	0.00604	0.000063	0.00029	0.093079	2.395296
5,000	2	lightgbm	0.9052	0.06972	0.004841	0.040406	0.091379	2.446973
5,000	2	partitionwise	0.9052	0.00448	-0.000193	-0.001152	0.091379	2.446973
5,000	3	lightgbm	0.9128	0.06716	0.003403	0.030676	0.091409	2.438814
5,000	3	partitionwise	0.9128	0.01168	0.000057	0.000232	0.091409	2.438814
5,000	4	lightgbm	0.8898	0.07864	0.008842	0.050751	0.095965	2.328892
5,000	4	partitionwise	0.8898	0.01176	0.000045	0.000367	0.095965	2.328892
5,000	5	lightgbm	0.9042	0.07768	0.007441	0.052178	0.092792	2.427159
5,000	5	partitionwise	0.9042	0.0062	-0.000171	-0.001044	0.092792	2.427159
5,000	6	lightgbm	0.8998	0.08004	0.007109	0.04985	0.09392	2.398474
5,000	6	partitionwise	0.8998	0.00352	-0.000238	-0.001198	0.09392	2.398474
5,000	7	lightgbm	0.9122	0.05884	0.002965	0.029196	0.088434	2.496596
5,000	7	partitionwise	0.9122	0.00812	-0.000036	-0.00043	0.088434	2.496596
null	null	null	mean	mean	mean	mean	mean	null
250	null	lightgbm	0.9025000000000001	0.03559999999999999	-0.009413980958760955	-0.04376382716230404	0.09397664560904574	null
250	null	partitionwise	0.9025000000000001	0.00825	-0.003840472815945898	-0.13976801070346667	0.09397664560904574	null
500	null	lightgbm	0.902	0.04645	-0.005363400077936773	-0.036616079509397645	0.09199853085725315	null
500	null	partitionwise	0.902	0.0019999999999999996	-0.0016136982930391307	-0.036600624475474776	0.09199853085725315	null
1,000	null	lightgbm	0.897875	0.045125	-0.0033181898753882073	-0.03012281872741975	0.09471546880685786	null
1,000	null	partitionwise	0.897875	0.0023375000000000006	-0.001408709592638226	-0.00919565522005159	0.09471546880685786	null
2,000	null	lightgbm	0.9	0.053725	-0.000526496592940818	-0.0001451744353569073	0.0928298126700978	null
2,000	null	partitionwise	0.9	0.0102375	-0.00024000269806431245	-0.0017170962963282974	0.0928298126700978	null
5,000	null	lightgbm	0.9031750000000001	0.071935	0.005523568988568581	0.04107671704266958	0.09275719576857498	null
5,000	null	partitionwise	0.9031750000000001	0.007304999999999998	-8.249022035821679e-05	-0.0004897223434163751	0.09275719576857498	null

YAML Metadata Warning:empty or missing yaml metadata in repo card

Check out the documentation for more information.

Conditional Coverage Diagnostics reproduction

Independent scaled checks for ERT definitions, classifier power, convergence, the over/under decomposition, and the k-fold estimator. Run with:

uv run --with-requirements requirements.txt python reproduce.py

The author_code/ checkout is retained for provenance. reproduce.py does not import it.

Downloads last month: 7

Total file size:

1.16 MB