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: CastError
Message: Couldn't cast
_doc: string
rows: list<item: struct<rev: string, stage: int64, step: int64, arm: string, pass1: double, pass1_ci95: li (... 221 chars omitted)
child 0, item: struct<rev: string, stage: int64, step: int64, arm: string, pass1: double, pass1_ci95: list<item: do (... 209 chars omitted)
child 0, rev: string
child 1, stage: int64
child 2, step: int64
child 3, arm: string
child 4, pass1: double
child 5, pass1_ci95: list<item: double>
child 0, item: double
child 6, pass1_sd_boot: double
child 7, pass1_sd_rollouts: double
child 8, pass1_by_rollout_index: list<item: double>
child 0, item: double
child 9, cap_rate: double
child 10, median_tokens: double
child 11, gain: double
child 12, n_problems: int64
child 13, n_rollouts: int64
child 14, source: string
correct: bool
output_tokens: int64
completion: string
max_tokens_used: int64
committed: bool
rollout_index: int64
hit_token_cap: bool
arm: string
steer: string
finish_reason: string
problem_key: string
label: string
to
{'problem_key': Value('string'), 'rollout_index': Value('int64'), 'arm': Value('string'), 'steer': Value('string'), 'label': Value('string'), 'committed': Value('bool'), 'correct': Value('bool'), 'hit_token_cap': Value('bool'), 'finish_reason': Value('string'), 'output_tokens': Value('int64'), 'max_tokens_used': Value('int64'), 'completion': Value('string')}
because column names don't match
Traceback: Traceback (most recent call last):
File "/usr/local/lib/python3.14/site-packages/datasets/builder.py", line 1827, in _prepare_split_single
for key, table in generator:
^^^^^^^^^
File "/src/services/worker/src/worker/job_runners/config/parquet_and_info.py", line 613, in wrapped
for item in generator(*args, **kwargs):
~~~~~~~~~^^^^^^^^^^^^^^^^^
File "/usr/local/lib/python3.14/site-packages/datasets/packaged_modules/json/json.py", line 343, in _generate_tables
self._cast_table(pa_table, json_field_paths=json_field_paths),
~~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/usr/local/lib/python3.14/site-packages/datasets/packaged_modules/json/json.py", line 132, in _cast_table
pa_table = table_cast(pa_table, self.info.features.arrow_schema)
File "/usr/local/lib/python3.14/site-packages/datasets/table.py", line 2378, in table_cast
return cast_table_to_schema(table, schema)
File "/usr/local/lib/python3.14/site-packages/datasets/table.py", line 2306, in cast_table_to_schema
raise CastError(
...<3 lines>...
)
datasets.table.CastError: Couldn't cast
_doc: string
rows: list<item: struct<rev: string, stage: int64, step: int64, arm: string, pass1: double, pass1_ci95: li (... 221 chars omitted)
child 0, item: struct<rev: string, stage: int64, step: int64, arm: string, pass1: double, pass1_ci95: list<item: do (... 209 chars omitted)
child 0, rev: string
child 1, stage: int64
child 2, step: int64
child 3, arm: string
child 4, pass1: double
child 5, pass1_ci95: list<item: double>
child 0, item: double
child 6, pass1_sd_boot: double
child 7, pass1_sd_rollouts: double
child 8, pass1_by_rollout_index: list<item: double>
child 0, item: double
child 9, cap_rate: double
child 10, median_tokens: double
child 11, gain: double
child 12, n_problems: int64
child 13, n_rollouts: int64
child 14, source: string
correct: bool
output_tokens: int64
completion: string
max_tokens_used: int64
committed: bool
rollout_index: int64
hit_token_cap: bool
arm: string
steer: string
finish_reason: string
problem_key: string
label: string
to
{'problem_key': Value('string'), 'rollout_index': Value('int64'), 'arm': Value('string'), 'steer': Value('string'), 'label': Value('string'), 'committed': Value('bool'), 'correct': Value('bool'), 'hit_token_cap': Value('bool'), 'finish_reason': Value('string'), 'output_tokens': Value('int64'), 'max_tokens_used': Value('int64'), 'completion': Value('string')}
because column names don't match
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 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.
problem_key string | rollout_index int64 | arm string | steer string | label string | committed bool | correct bool | hit_token_cap bool | finish_reason string | output_tokens int64 | max_tokens_used int64 | completion string |
|---|---|---|---|---|---|---|---|---|---|---|---|
precalculus__8ee539 | 0 | nn_okay | base | math500 | false | false | true | length | 7,168 | 7,168 |
Okay, so can you have a solution to this problem? The problem is that it is not a solution to this problem, it is a solution to this problem. The solution is to use the following expression, which will give you a solution to this problem. The solution is to use the following expression. The solution is to use the f... |
precalculus__8ee539 | 1 | nn_okay | base | math500 | false | false | true | length | 7,168 | 7,168 |
Okay, so that's the problem. For example, if you want to use the same line of the problem as in the previous example, you can use the same line of the problem and it will be the same as the same line of the problem. For example, if you want to use the same line of the problem as in the previous example, the problem... |
precalculus__8ee539 | 2 | nn_okay | base | math500 | false | false | true | length | 7,168 | 7,168 |
Okay, so that it can be shown that $\theta$ is the solution of the problem. See the solution for the problem. The problem is that $\theta$ is the solution of the problem. Note that $\theta$ is the solution of the problem. Note that the solution of the problem is to find the solution of the problem. The solution ... |
precalculus__8ee539 | 3 | nn_okay | base | math500 | false | false | true | length | 7,168 | 7,168 |
Okay, that's why the problem is not the problem. The problem is that it's not the problem. The problem is that if $r \in \mathbb{R}^n$, then $r = 0$ and $r = 0$. The problem is that if $r = 0$, then $r = 0$ and $r = 0$. If $r = 0$, then $r = 0$. The problem is that if $r = 0$, then $r = 0$, and $r = 0$, then $r = 0$.... |
intermediate_algebra__f4e837 | 0 | nn_okay | base | math500 | false | false | true | length | 7,168 | 7,168 |
Okay, that's a good way of explaining the solution. First of all, let's try this on its own line. Second, let's try this on its own line. Let's say $p = 1$ and $p = 1$. Let's say $p = 1$. Let's see if $p = 1$ and $p = 1$ and $p = 1$. Let's see if $p = 1$. Let's see if $p = 1$. Let's see if $p = 1$. Let's see if $p = ... |
intermediate_algebra__f4e837 | 1 | nn_okay | base | math500 | false | false | true | length | 7,168 | 7,168 |
Okay, so the line of the problem is the same as the line of the problem. The problem is that the line of the problem is not the same as the line of the problem. The line of the problem is the line of the problem, where the line of the problem is the line of the problem. The line of the problem is the line of the prob... |
intermediate_algebra__f4e837 | 2 | nn_okay | base | math500 | false | false | true | length | 7,168 | 7,168 |
Okay, so here are the lines of the problem. For example, if we have a question that is not the answer to the problem, we have a question that will be the answer to the problem. For example, if we have a question that is not the answer to the problem, it will be the answer to the problem. For example, if we have a que... |
intermediate_algebra__f4e837 | 3 | nn_okay | base | math500 | false | false | true | length | 7,168 | 7,168 | "\n\nOkay, now that it is not the answer. It is not the answer to the problem that it does not know (...TRUNCATED) |
algebra__96e5cb | 0 | nn_okay | base | math500 | false | false | true | length | 7,168 | 7,168 | "\n\nOkay, so it can be more complicated than this.\nThis should be very confusing since it is not a(...TRUNCATED) |
algebra__96e5cb | 1 | nn_okay | base | math500 | false | false | true | length | 7,168 | 7,168 | "\n\nOkay, so, so, so, so, so, so, so, so, so, so, so, so, so, so, so, so, so, so, so, so, so, so, s(...TRUNCATED) |
End of preview.
rollouts-olmo7b-ladder-csail
Model: allenai/Olmo-3-1025-7B intermediate checkpoints (olmo_ladder stage1-step{1000,20000,150000,500000,1000000,1413814}, stage2-step{1000,3000,7000,14000,25000,47684}); top20_stage1 = stage1-step1413814. Tokenizer: allenai/Olmo-3-1025-7B (snapshot a81bae42). Protocol: RL-Zero prompt, MATH-500 x 4, budget 7,168, T 0.6, top-p 0.95, seed 20260819; none + nn_okay per checkpoint; top20_stage1: none + the top-20 beam nominees on the pretraining-only checkpoint.
Rollouts generated on the CSAIL cluster for the reasoning-registers paper (Sophie Wang, MIT), uploaded 2026-09-13; sister repositories of the same organisation hold the cells run by Rulin Shao (see HF_MANIFEST.md there). Layout: each top-level folder is one local store with its date prefix dropped, keeping the store's own layout (rollouts_base_<arm>_p<start>.jsonl, one line per rollout with problem_key, rollout_index, arm, completion, correct, output_tokens, hit_token_cap; *_b2 = the second seeded batch; *_execgraded/ = the execution-graded copy for code). summaries/ holds the cell summaries the paper's tables read (scripts/eval/summarize_cell.py), tables/ the derived tables. Folders: ladder_rlzero, top20_stage1, tables. Files are plain JSON Lines, not compressed.
- Downloads last month
- 25