Datasets:
Dataset Preview
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: JSON parse error: Missing a colon after a name of object member. in row 116
Traceback: Traceback (most recent call last):
File "/usr/local/lib/python3.12/site-packages/datasets/packaged_modules/json/json.py", line 276, in _generate_tables
df = pandas_read_json(f)
^^^^^^^^^^^^^^^^^^^
File "/usr/local/lib/python3.12/site-packages/datasets/packaged_modules/json/json.py", line 34, in pandas_read_json
return pd.read_json(path_or_buf, **kwargs)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/usr/local/lib/python3.12/site-packages/pandas/io/json/_json.py", line 815, in read_json
return json_reader.read()
^^^^^^^^^^^^^^^^^^
File "/usr/local/lib/python3.12/site-packages/pandas/io/json/_json.py", line 1014, in read
obj = self._get_object_parser(self.data)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/usr/local/lib/python3.12/site-packages/pandas/io/json/_json.py", line 1040, in _get_object_parser
obj = FrameParser(json, **kwargs).parse()
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/usr/local/lib/python3.12/site-packages/pandas/io/json/_json.py", line 1176, in parse
self._parse()
File "/usr/local/lib/python3.12/site-packages/pandas/io/json/_json.py", line 1392, in _parse
ujson_loads(json, precise_float=self.precise_float), dtype=None
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
ValueError: Trailing data
During handling of the above exception, another exception occurred:
Traceback (most recent call last):
File "/usr/local/lib/python3.12/site-packages/datasets/builder.py", line 1872, in _prepare_split_single
for key, table in generator:
^^^^^^^^^
File "/usr/local/lib/python3.12/site-packages/datasets/packaged_modules/json/json.py", line 279, in _generate_tables
raise e
File "/usr/local/lib/python3.12/site-packages/datasets/packaged_modules/json/json.py", line 242, in _generate_tables
pa_table = paj.read_json(
^^^^^^^^^^^^^^
File "pyarrow/_json.pyx", line 342, in pyarrow._json.read_json
File "pyarrow/error.pxi", line 155, in pyarrow.lib.pyarrow_internal_check_status
File "pyarrow/error.pxi", line 92, in pyarrow.lib.check_status
pyarrow.lib.ArrowInvalid: JSON parse error: Missing a colon after a name of object member. in row 116
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 1347, in compute_config_parquet_and_info_response
parquet_operations = convert_to_parquet(builder)
^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/src/services/worker/src/worker/job_runners/config/parquet_and_info.py", line 980, in convert_to_parquet
builder.download_and_prepare(
File "/usr/local/lib/python3.12/site-packages/datasets/builder.py", line 884, in download_and_prepare
self._download_and_prepare(
File "/usr/local/lib/python3.12/site-packages/datasets/builder.py", line 947, in _download_and_prepare
self._prepare_split(split_generator, **prepare_split_kwargs)
File "/usr/local/lib/python3.12/site-packages/datasets/builder.py", line 1739, in _prepare_split
for job_id, done, content in self._prepare_split_single(
^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/usr/local/lib/python3.12/site-packages/datasets/builder.py", line 1925, in _prepare_split_single
raise DatasetGenerationError("An error occurred while generating the dataset") from e
datasets.exceptions.DatasetGenerationError: An error occurred while generating the datasetNeed help to make the dataset viewer work? Make sure to review how to configure the dataset viewer, and open a discussion for direct support.
domain string | seq_len int64 | sequence_label int64 | sequence_avg_max_lookback float64 | sequence_avg_median_lookback float64 | sequence_distance_variance int64 |
|---|---|---|---|---|---|
longform | 131,072 | 1 | 65,444.992188 | 47,183.210938 | 484,923,328 |
longform | 131,072 | 1 | 65,185.449219 | 36,120.71875 | 475,834,720 |
longform | 131,072 | 1 | 65,403.285156 | 45,414.660156 | 494,662,464 |
longform | 131,072 | 1 | 65,378.902344 | 53,252.171875 | 501,026,336 |
longform | 131,072 | 1 | 63,980.757813 | 56,531.203125 | 497,180,096 |
longform | 131,072 | 1 | 54,352.382813 | 48,643.535156 | 472,596,928 |
longform | 131,072 | 1 | 65,421.210938 | 48,786.882813 | 499,100,960 |
longform | 131,072 | 1 | 65,439.671875 | 54,728.507813 | 497,982,208 |
longform | 131,072 | 1 | 65,333.523438 | 46,186.316406 | 491,902,720 |
longform | 131,072 | 1 | 64,777.484375 | 44,015.726563 | 490,259,744 |
longform | 131,072 | 1 | 64,867.027344 | 44,197.257813 | 489,181,344 |
longform | 131,072 | 1 | 64,084.539063 | 54,996.34375 | 494,354,912 |
longform | 131,072 | 1 | 65,446.0625 | 50,780.710938 | 496,387,584 |
longform | 131,072 | 1 | 64,659.722656 | 41,571.648438 | 486,747,840 |
longform | 131,072 | 1 | 65,458.0625 | 46,051.515625 | 485,487,712 |
longform | 131,072 | 1 | 65,420.695313 | 43,593.140625 | 491,559,296 |
longform | 131,072 | 1 | 65,409.890625 | 47,705.894531 | 499,155,296 |
longform | 131,072 | 1 | 65,457.957031 | 47,305.828125 | 491,976,064 |
longform | 131,072 | 1 | 65,469.75 | 52,673.527344 | 502,921,664 |
longform | 131,072 | 1 | 64,831.742188 | 47,454.222656 | 487,884,864 |
longform | 131,072 | 1 | 65,351.179688 | 42,861.429688 | 491,470,048 |
longform | 131,072 | 1 | 65,340.878906 | 54,500.703125 | 497,167,040 |
longform | 131,072 | 1 | 65,369.464844 | 52,511.808594 | 495,321,472 |
longform | 131,072 | 1 | 55,142.523438 | 48,244.132813 | 491,507,104 |
longform | 131,072 | 1 | 64,948.371094 | 47,020.808594 | 487,823,680 |
longform | 131,072 | 1 | 64,731.738281 | 43,813.460938 | 487,006,144 |
longform | 131,072 | 1 | 65,279.953125 | 39,381.671875 | 482,496,704 |
longform | 131,072 | 1 | 65,059.625 | 47,835.140625 | 489,936,288 |
longform | 131,072 | 1 | 64,408.796875 | 42,792.105469 | 485,746,048 |
longform | 131,072 | 1 | 65,459.585938 | 50,037.320313 | 500,974,592 |
longform | 131,072 | 1 | 65,339.460938 | 50,998.390625 | 498,826,752 |
longform | 131,072 | 1 | 65,459.945313 | 51,851.332031 | 501,810,880 |
longform | 131,072 | 1 | 65,469.640625 | 49,209.226563 | 501,357,888 |
longform | 131,072 | 1 | 65,469.5625 | 53,533.203125 | 498,904,640 |
longform | 131,072 | 1 | 65,459.808594 | 44,854.15625 | 494,205,248 |
longform | 131,072 | 1 | 65,175.539063 | 50,755.058594 | 487,366,016 |
longform | 131,072 | 1 | 65,418.101563 | 46,694.875 | 499,567,936 |
longform | 131,072 | 1 | 64,818.484375 | 41,234.671875 | 492,013,600 |
longform | 131,072 | 1 | 65,108.710938 | 44,369.335938 | 486,116,672 |
longform | 131,072 | 1 | 65,434.71875 | 43,402.304688 | 497,855,328 |
longform | 131,072 | 1 | 65,406.953125 | 54,905.90625 | 496,131,200 |
longform | 131,072 | 1 | 65,432.789063 | 49,177.730469 | 504,028,288 |
longform | 131,072 | 1 | 65,400.257813 | 38,756.660156 | 497,841,472 |
longform | 131,072 | 1 | 65,382.3125 | 47,476.835938 | 489,260,128 |
longform | 131,072 | 1 | 64,816.246094 | 48,461.796875 | 481,698,240 |
longform | 131,072 | 1 | 64,068.113281 | 46,042.570313 | 495,460,032 |
longform | 131,072 | 1 | 65,408.1875 | 48,825.074219 | 503,340,096 |
longform | 131,072 | 1 | 65,424.101563 | 53,956.257813 | 502,491,904 |
longform | 131,072 | 1 | 65,413.835938 | 42,497.773438 | 492,530,752 |
longform | 131,072 | 1 | 65,387.515625 | 37,821.121094 | 489,757,344 |
longform | 131,072 | 1 | 65,430.46875 | 51,453.699219 | 491,084,544 |
longform | 131,072 | 1 | 65,444.9375 | 58,286.34375 | 494,680,672 |
longform | 131,072 | 1 | 65,382.703125 | 48,266.734375 | 495,356,448 |
longform | 131,072 | 1 | 65,450.042969 | 51,665.328125 | 498,239,872 |
longform | 131,072 | 1 | 65,390.25 | 48,587.425781 | 496,202,240 |
longform | 131,072 | 1 | 64,976.453125 | 48,680.804688 | 496,839,232 |
longform | 131,072 | 1 | 65,471.351563 | 47,524.128906 | 499,121,216 |
longform | 131,072 | 1 | 65,206.710938 | 47,917.851563 | 497,388,032 |
longform | 131,072 | 1 | 65,376.148438 | 48,295.546875 | 497,066,816 |
longform | 131,072 | 1 | 64,975.363281 | 35,751.984375 | 481,160,192 |
longform | 131,072 | 1 | 65,404.203125 | 46,849.898438 | 487,786,816 |
longform | 131,072 | 1 | 65,483.226563 | 46,002.804688 | 497,685,760 |
longform | 131,072 | 1 | 64,692.109375 | 36,985.941406 | 490,582,016 |
longform | 131,072 | 1 | 65,437.699219 | 46,488.839844 | 495,479,616 |
longform | 131,072 | 1 | 56,271.539063 | 34,485.570313 | 471,073,344 |
longform | 131,072 | 1 | 65,445.117188 | 50,382.539063 | 498,829,184 |
longform | 131,072 | 1 | 65,426.960938 | 49,276.023438 | 503,022,528 |
longform | 131,072 | 1 | 65,265.101563 | 42,334.625 | 487,711,680 |
longform | 131,072 | 1 | 65,148.335938 | 50,181.671875 | 498,518,496 |
longform | 131,072 | 1 | 65,393.519531 | 45,041.261719 | 492,472,896 |
longform | 131,072 | 1 | 65,434.1875 | 42,516.304688 | 498,003,904 |
longform | 131,072 | 1 | 65,448.886719 | 49,956.210938 | 499,652,640 |
longform | 131,072 | 1 | 64,791.898438 | 46,260.265625 | 486,328,448 |
longform | 131,072 | 1 | 65,457.945313 | 46,095.40625 | 492,821,568 |
longform | 131,072 | 1 | 65,089.902344 | 47,904.058594 | 494,721,408 |
longform | 131,072 | 1 | 65,022.40625 | 46,060.085938 | 492,260,480 |
longform | 131,072 | 1 | 65,273.425781 | 45,554.574219 | 496,541,856 |
longform | 131,072 | 1 | 53,737.59375 | 44,689.632813 | 482,560,960 |
longform | 131,072 | 1 | 65,436.855469 | 54,457.839844 | 497,479,680 |
longform | 131,072 | 1 | 65,136.476563 | 41,106.84375 | 480,606,016 |
longform | 131,072 | 1 | 65,441.296875 | 58,195.742188 | 500,990,688 |
longform | 131,072 | 1 | 65,008.058594 | 52,146.855469 | 496,672,128 |
longform | 131,072 | 1 | 63,855.695313 | 38,848.507813 | 482,777,088 |
longform | 131,072 | 1 | 64,711.664063 | 35,034.652344 | 490,540,576 |
longform | 131,072 | 1 | 65,483.128906 | 58,999.0625 | 498,054,784 |
longform | 131,072 | 1 | 65,441.957031 | 53,373.015625 | 492,842,016 |
longform | 131,072 | 1 | 65,130.886719 | 57,522.679688 | 492,049,024 |
longform | 131,072 | 1 | 65,464.453125 | 44,576.664063 | 491,746,432 |
longform | 131,072 | 1 | 64,760.84375 | 41,326.476563 | 496,226,624 |
longform | 131,072 | 1 | 65,082.832031 | 46,092.730469 | 490,538,240 |
longform | 131,072 | 1 | 64,939.449219 | 48,664.96875 | 487,148,032 |
longform | 131,072 | 1 | 65,444.542969 | 42,699.601563 | 489,089,984 |
longform | 131,072 | 1 | 65,033.664063 | 50,153.59375 | 493,257,856 |
longform | 131,072 | 1 | 65,470.046875 | 53,211.300781 | 495,583,456 |
longform | 131,072 | 1 | 65,428.085938 | 49,685.890625 | 489,910,816 |
longform | 131,072 | 1 | 65,411.242188 | 46,738.457031 | 493,953,152 |
longform | 131,072 | 1 | 65,477.679688 | 45,294.46875 | 498,904,032 |
longform | 131,072 | 1 | 65,366.011719 | 52,222.007813 | 495,776,832 |
longform | 131,072 | 1 | 65,345.21875 | 51,159.828125 | 490,741,696 |
longform | 131,072 | 1 | 65,456.917969 | 50,277.324219 | 489,280,160 |
End of preview.
Long-Context Attention Labels (64K & 128K)
Attention-based document labels for long-context training data selection.
Overview
Each document is labeled with attention lookback metrics computed by running it through a model and measuring how far back each token attends (via top-k=10 head-averaged attention distances).
| Run | Model | Context | Records |
|---|---|---|---|
| olmo3_64k | OLMo-3-1025-7B (stage2) | 65,536 tokens | ~5,000 |
| olmo3_128k | OLMo-3-1025-7B (stage2) | 131,072 tokens | ~4,600 |
| qwen25_64k | Qwen2.5-7B-Instruct | 65,536 tokens | ~1,900 |
| qwen25_128k | Qwen2.5-7B-Instruct | 131,072 tokens | ~4,800 |
Fields
| Field | Description |
|---|---|
domain |
Source domain (code, books, arxiv, web, govreport) |
seq_len |
Sequence length in tokens |
sequence_label |
Binary label (1=long-range attention, 0=short-range) using fixed 2048-token threshold |
sequence_avg_max_lookback |
Mean of per-token max lookback distance across top-10 attended positions |
sequence_avg_median_lookback |
Mean of per-token median lookback distance (primary metric) |
sequence_distance_variance |
Variance of attention distances |
Labeling Details
- Layers: OLMo-3 [8, 16, 24], Qwen2.5 [7, 14, 21] (spread-3 across model depth)
- Top-k: 10 (top 10 attended positions by head-averaged attention probability)
- Attention sink mitigation: First 32 tokens ignored
- Attention implementation: SDPA with FORCE_CAUSAL_SDPA=1
- OLMo-3 context extension: Linear RoPE scaling (8x for 64K, 16x for 128K)
- Source data: LongMINO-prefiltered pool (code, books, arxiv, web, govreport)
- Platform: OLCF Frontier (AMD MI250X)
Usage
The sequence_label field uses a fixed threshold (2048 tokens) which is not ideal for all model×pool combinations. We recommend percentile-based thresholding on sequence_avg_median_lookback instead — e.g., top 15% as positive.
Note on record counts
Qwen2.5 64K has fewer records (~1,900) because the labeling pipeline applied a neg:pos ratio cap (2:1) which dropped excess negatives. The raw attention metrics were computed for all 10,000 documents but most were not written out. A re-run with the cap disabled is planned.
- Downloads last month
- 39