Datasets:

rustambekurokov
/

esm3-distill-hard-contact-benchmark

Languages:

English

Tags:

protein

protein-language-models

contact-prediction

benchmark

License:

Dataset card Data Studio Files Files and versions

xet

Community

Dataset Preview

Duplicate

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

Error code:   DatasetGenerationError
Exception:    ArrowInvalid
Message:      Schema at index 1 was different: 
model_name: string
dtype: string
max_length: int64
max_rss_gb: double
n_manifest: int64
n_scored: int64
n_skipped_too_long: int64
n_skipped_for_memory: int64
skipped_too_long: list<item: null>
skipped_for_memory: list<item: null>
elapsed_seconds: double
seconds_per_protein: double
peak_rss_gb: double
aggregate: struct<shipped: struct<P@L/5: struct<mean: double, n: int64>, P@L/2: struct<mean: double, n: int64>, P@L: struct<mean: double, n: int64>>, heads: struct<stage1_v3_s0: struct<P@L/5: struct<mean: double, n: int64>, P@L/2: struct<mean: double, n: int64>, P@L: struct<mean: double, n: int64>>, stage1_v3_s1: struct<P@L/5: struct<mean: double, n: int64>, P@L/2: struct<mean: double, n: int64>, P@L: struct<mean: double, n: int64>>, stage1_v3_s2: struct<P@L/5: struct<mean: double, n: int64>, P@L/2: struct<mean: double, n: int64>, P@L: struct<mean: double, n: int64>>>, head_mean: struct<P@L/5: struct<mean: double, n: int64>, P@L/2: struct<mean: double, n: int64>, P@L: struct<mean: double, n: int64>>>
rows: list<item: struct<identifier: string, length: int64, shipped: struct<P@L/5: double, P@L/2: double, P@L: double>, heads: struct<stage1_v3_s0: struct<P@L/5: double, P@L/2: double, P@L: double>, stage1_v3_s1: struct<P@L/5: double, P@L/2: double, P@L: double>, stage1_v3_s2: struct<P@L/5: double, P@L/2: double, P@L: double>>, head_mean: struct<P@L/5: double, P@L/2: double, P@L: double>, peak_rss_gb: double>>
vs
model_name: string
manifest: string
contact_dir: string
num_steps: string
temperature: double
cb_mode: string
num_requested: int64
num_scored: int64
num_errors: int64
errors: list<item: null>
aggregate: struct<P@L/5: struct<mean: double, std: double, n: int64>, P@L/2: struct<mean: double, std: double, n: int64>, P@L: struct<mean: double, std: double, n: int64>>
aggregate_ptm_ge_threshold: struct<ptm_threshold: double, n_proteins: int64, P@L/5: struct<mean: double, std: double, n: int64>, P@L/2: struct<mean: double, std: double, n: int64>, P@L: struct<mean: double, std: double, n: int64>>
seconds_per_protein: struct<mean: double, std: double>
total_runtime_seconds: double
rows: list<item: struct<identifier: string, length: int64, seconds: double, ptm: double, mean_plddt: double, cb_mode: string, frac_cb_nan: double, frac_final_coord_nan: double, frac_virtual_cb: double, case_counts: struct<real_cb: int64, virtual_cb: int64, ca_only: int64, unusable: int64>, coords_bos_eos_stripped: bool, P@L/5: double, P@L/2: double, P@L: double>>
Traceback:    Traceback (most recent call last):
                File "/usr/local/lib/python3.14/site-packages/datasets/builder.py", line 1869, in _prepare_split_single
                  num_examples, num_bytes = writer.finalize()
                                            ~~~~~~~~~~~~~~~^^
                File "/usr/local/lib/python3.14/site-packages/datasets/arrow_writer.py", line 781, in finalize
                  self.write_rows_on_file()
                  ~~~~~~~~~~~~~~~~~~~~~~~^^
                File "/usr/local/lib/python3.14/site-packages/datasets/arrow_writer.py", line 662, in write_rows_on_file
                  table = pa.concat_tables(self.current_rows)
                File "pyarrow/table.pxi", line 6321, in pyarrow.lib.concat_tables
                File "pyarrow/error.pxi", line 155, in pyarrow.lib.pyarrow_internal_check_status
                File "pyarrow/error.pxi", line 92, in pyarrow.lib.check_status
                  raise convert_status(status)
              pyarrow.lib.ArrowInvalid: Schema at index 1 was different: 
              model_name: string
              dtype: string
              max_length: int64
              max_rss_gb: double
              n_manifest: int64
              n_scored: int64
              n_skipped_too_long: int64
              n_skipped_for_memory: int64
              skipped_too_long: list<item: null>
              skipped_for_memory: list<item: null>
              elapsed_seconds: double
              seconds_per_protein: double
              peak_rss_gb: double
              aggregate: struct<shipped: struct<P@L/5: struct<mean: double, n: int64>, P@L/2: struct<mean: double, n: int64>, P@L: struct<mean: double, n: int64>>, heads: struct<stage1_v3_s0: struct<P@L/5: struct<mean: double, n: int64>, P@L/2: struct<mean: double, n: int64>, P@L: struct<mean: double, n: int64>>, stage1_v3_s1: struct<P@L/5: struct<mean: double, n: int64>, P@L/2: struct<mean: double, n: int64>, P@L: struct<mean: double, n: int64>>, stage1_v3_s2: struct<P@L/5: struct<mean: double, n: int64>, P@L/2: struct<mean: double, n: int64>, P@L: struct<mean: double, n: int64>>>, head_mean: struct<P@L/5: struct<mean: double, n: int64>, P@L/2: struct<mean: double, n: int64>, P@L: struct<mean: double, n: int64>>>
              rows: list<item: struct<identifier: string, length: int64, shipped: struct<P@L/5: double, P@L/2: double, P@L: double>, heads: struct<stage1_v3_s0: struct<P@L/5: double, P@L/2: double, P@L: double>, stage1_v3_s1: struct<P@L/5: double, P@L/2: double, P@L: double>, stage1_v3_s2: struct<P@L/5: double, P@L/2: double, P@L: double>>, head_mean: struct<P@L/5: double, P@L/2: double, P@L: double>, peak_rss_gb: double>>
              vs
              model_name: string
              manifest: string
              contact_dir: string
              num_steps: string
              temperature: double
              cb_mode: string
              num_requested: int64
              num_scored: int64
              num_errors: int64
              errors: list<item: null>
              aggregate: struct<P@L/5: struct<mean: double, std: double, n: int64>, P@L/2: struct<mean: double, std: double, n: int64>, P@L: struct<mean: double, std: double, n: int64>>
              aggregate_ptm_ge_threshold: struct<ptm_threshold: double, n_proteins: int64, P@L/5: struct<mean: double, std: double, n: int64>, P@L/2: struct<mean: double, std: double, n: int64>, P@L: struct<mean: double, std: double, n: int64>>
              seconds_per_protein: struct<mean: double, std: double>
              total_runtime_seconds: double
              rows: list<item: struct<identifier: string, length: int64, seconds: double, ptm: double, mean_plddt: double, cb_mode: string, frac_cb_nan: double, frac_final_coord_nan: double, frac_virtual_cb: double, case_counts: struct<real_cb: int64, virtual_cb: int64, ca_only: int64, unusable: int64>, coords_bos_eos_stripped: bool, P@L/5: double, P@L/2: double, P@L: double>>
              
              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/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 1694, 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 1880, in _prepare_split_single
                  raise DatasetGenerationError("An error occurred while generating the dataset") from e
              datasets.exceptions.DatasetGenerationError: An error occurred while generating the dataset

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.

cluster string	identifier string	length int64	release_date timestamp[us]	resolution float64
line:186664	10YV	275	2026-05-06T00:00:00	2.15
singleton:11CP	11CP	184	2026-09-09T00:00:00	2.7
line:5154	11FC	620	2026-04-29T00:00:00	2.1
line:21723	11GZ	220	2026-05-20T00:00:00	2.44
line:842	12HG	79	2026-08-26T00:00:00	2.35
line:262165	22OC	193	2026-03-18T00:00:00	2
line:267640	23AA	104	2026-08-12T00:00:00	2.03
line:6818	24XH	311	2026-06-24T00:00:00	3
line:63061	25QC	292	2026-08-05T00:00:00	2.2
line:1784	25WN	139	2026-05-27T00:00:00	2.6
line:17306	26TC	278	2026-05-27T00:00:00	2
singleton:27JO	27JO	127	2026-09-02T00:00:00	2.08
line:53939	28OJ	88	2026-04-08T00:00:00	2.12
line:5643	28SV	288	2026-07-22T00:00:00	2.21
line:6335	29TA	181	2026-04-15T00:00:00	2.8
line:3097	43CQ	328	2026-07-29T00:00:00	2.9
line:5654	9CME	683	2026-01-14T00:00:00	2.7
line:8747	9D3H	340	2025-12-10T00:00:00	2.28
line:1004	9E37	622	2025-10-29T00:00:00	2
line:3481	9E58	348	2026-02-04T00:00:00	2
line:9233	9EDJ	279	2026-01-21T00:00:00	2.6
line:10840	9GED	186	2026-02-11T00:00:00	2.484
line:729	9GEX	151	2026-02-18T00:00:00	2.6
line:7121	9H29	291	2025-10-29T00:00:00	2.17
line:2548	9HHU	600	2026-06-03T00:00:00	2.16
line:122061	9HQ8	105	2025-12-24T00:00:00	2
line:30979	9HRJ	347	2025-11-12T00:00:00	2.14
line:3312	9IG5	323	2026-03-04T00:00:00	2.4
line:23494	9JB6	92	2026-03-11T00:00:00	2.49
line:5788	9JH2	162	2025-12-24T00:00:00	2.3
line:33433	9JX4	371	2026-04-15T00:00:00	2.2
line:90396	9KCS	494	2025-11-05T00:00:00	2.13
line:7067	9KCY	323	2026-05-06T00:00:00	2.429
line:6503	9KTP	328	2026-06-10T00:00:00	2.06
line:208918	9KVN	126	2026-03-18T00:00:00	2.702
line:80855	9L10	227	2025-11-19T00:00:00	2.61
line:13121	9L1H	244	2025-12-17T00:00:00	2.1
line:196817	9L4Q	214	2025-10-29T00:00:00	2.27
line:72103	9L5M	180	2025-11-05T00:00:00	2
line:217546	9L75	281	2025-12-31T00:00:00	2.72
line:10162	9LUZ	541	2026-02-11T00:00:00	2.3
line:188118	9LXM	289	2026-08-26T00:00:00	2.13
line:6422	9M5F	267	2026-03-18T00:00:00	2.35
line:40528	9M71	113	2026-03-11T00:00:00	2.39
line:44545	9M8T	333	2026-03-04T00:00:00	2.66
line:53427	9MCD	112	2026-04-08T00:00:00	2.6
line:28832	9MXM	343	2026-06-24T00:00:00	2.06
line:68876	9N1C	631	2026-01-28T00:00:00	2
line:254981	9N30	120	2026-07-29T00:00:00	2.2
line:33476	9NP7	248	2026-06-10T00:00:00	3.4
line:6519	9NY9	349	2026-04-08T00:00:00	2.74
line:49141	9OR2	212	2025-11-26T00:00:00	2.07
line:5240	9PSR	255	2026-08-05T00:00:00	2.21
line:48575	9PW1	104	2026-08-05T00:00:00	2.91
line:1263	9PYU	320	2026-05-13T00:00:00	2.44
line:78836	9Q2G	165	2025-11-05T00:00:00	2
line:64354	9QKN	314	2026-04-08T00:00:00	3.2
line:338	9R3Q	558	2026-05-13T00:00:00	2.12
line:2758	9R44	101	2026-05-20T00:00:00	2.46
line:76807	9RD5	321	2026-01-14T00:00:00	2.5
line:40479	9RK3	117	2025-11-26T00:00:00	2.5
line:29381	9RKP	107	2026-01-14T00:00:00	2.6
line:3050	9RVJ	483	2026-07-22T00:00:00	2.3
line:21204	9S2D	183	2025-12-31T00:00:00	2.35
line:18076	9S4W	207	2026-02-25T00:00:00	2
line:43671	9S8F	532	2026-06-03T00:00:00	2.1
line:44685	9S9I	196	2026-01-28T00:00:00	2.2
line:155401	9SN1	663	2026-02-18T00:00:00	2.04
line:3485	9TCD	341	2026-05-13T00:00:00	2.7
line:12093	9UGK	145	2026-02-11T00:00:00	2.2
line:1302	9UMC	92	2026-04-29T00:00:00	2.6
line:8036	9UML	600	2026-04-29T00:00:00	2.25
line:16226	9UP5	337	2026-03-04T00:00:00	3.44
line:30677	9V0L	359	2025-11-26T00:00:00	2.3
line:1612	9VDQ	229	2026-05-20T00:00:00	3.1
line:64302	9VKJ	128	2026-07-15T00:00:00	2.09
line:4938	9VMU	299	2026-07-08T00:00:00	2.3
line:1145	9VNV	291	2026-02-11T00:00:00	2.5
line:38707	9VSH	312	2026-07-15T00:00:00	2.38
line:80822	9VZO	319	2026-04-22T00:00:00	2.266927
line:25390	9VZU	103	2026-07-08T00:00:00	2.39
line:29849	9W5Z	208	2026-06-10T00:00:00	2.35
line:55918	9W6X	199	2026-07-15T00:00:00	2.38
line:1021	9WAT	440	2025-12-17T00:00:00	2.1
line:4	9WEN	141	2026-08-26T00:00:00	2.3
singleton:9WIY	9WIY	605	2026-09-02T00:00:00	2.43
singleton:9WM5	9WM5	272	2026-09-02T00:00:00	2.4
singleton:9WOO	9WOO	395	2026-09-09T00:00:00	2.47
line:19902	9XUP	195	2026-05-27T00:00:00	2.01
line:15088	9XVW	194	2026-01-07T00:00:00	2.5
singleton:9YOX	9YOX	187	2026-09-09T00:00:00	2.17
singleton:9YOZ	9YOZ	203	2026-09-09T00:00:00	2.19
line:10377	9YZ5	330	2026-02-25T00:00:00	2.58
line:79079	9ZFN	177	2026-02-25T00:00:00	2.1
line:621	9ZMD	583	2025-12-24T00:00:00	2.65
line:16357	9ZN0	118	2026-01-21T00:00:00	2.12
line:16592	9ZW5	141	2026-05-13T00:00:00	2.6
null	null	null	null	null
null	null	null	null	null
null	null	null	null	null

ESM2 hard-stratum contact benchmark results

This lightweight reproducibility companion contains the 97-chain manifest and per-protein evaluation outputs. It deliberately excludes RCSB-derived contact tensors, model caches, and checkpoints pending source-data and model-license review.

benchmark.json: selected chains and structural metadata.
hard_esm2_and_heads.json: shipped ESM2 and three frozen-head outputs.
hard_esm3.json: ESM3 fold-derived contact outputs.
hard_summary.json: macro means, paired bootstrap intervals, and Wilcoxon tests.

Limitation

The set is a 97-chain screened subset, not the originally planned 300-chain benchmark. A separate independent final re-screen was not completed before the compute allocation ended; see the technical report for the exact limitation.

Citation

See paper/main.tex in the source repository until an arXiv identifier exists.

Downloads last month: 48

Total file size:

169 kB

ESM2 hard-stratum contact benchmark results

Contents

Limitation

Citation