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
pass@1: double
pass@1_ci95: list<item: double>
child 0, item: double
pass@1_sd_boot: double
pass@1_by_rollout_index: list<item: double>
child 0, item: double
pass@1_sd_rollouts: double
pass@8: double
pass@8_ci95: list<item: double>
child 0, item: double
pass@8_sd_boot: double
pass@16: double
pass@16_ci95: list<item: double>
child 0, item: double
pass@16_sd_boot: double
cap_rate: double
n_problems: int64
n_rollouts: int64
rollouts_per_problem: int64
max_tokens_used: list<item: int64>
child 0, item: int64
prompt_style: list<item: null>
child 0, item: null
median_output_tokens: int64
mean_output_tokens: double
expected_problems: int64
complete: bool
files: list<item: string>
child 0, item: string
per_problem: struct<aime2024_0: struct<rollout_index: list<item: int64>, correct: list<item: int64>, tokens: list (... 8176 chars omitted)
child 0, aime2024_0: struct<rollout_index: list<item: int64>, correct: list<item: int64>, tokens: list<item: int64>, answ (... 161 chars omitted)
child 0, rollout_index: list<item: int64>
child 0, item: int64
child 1, correct: list<item: int64>
child 0, item: int64
child 2, tokens: list<item: int64>
child 0, item: int64
child 3, answer: list<item: string>
child 0, item: string
child 4, hit_cap: list<item: int64>
child 0, item: int64
child 5, answer_end_frac: list<item: double>
child 0, item: double
child 6, answer_marker: list<item: s
...
child 0, k: int64
child 1, expected_tokens: double
child 2, pass_at_k: double
child 3, majority_at_k: double
child 27, 28: struct<k: int64, expected_tokens: double, pass_at_k: double, majority_at_k: double>
child 0, k: int64
child 1, expected_tokens: double
child 2, pass_at_k: double
child 3, majority_at_k: double
child 28, 29: struct<k: int64, expected_tokens: double, pass_at_k: double, majority_at_k: double>
child 0, k: int64
child 1, expected_tokens: double
child 2, pass_at_k: double
child 3, majority_at_k: double
child 29, 30: struct<k: int64, expected_tokens: double, pass_at_k: double, majority_at_k: double>
child 0, k: int64
child 1, expected_tokens: double
child 2, pass_at_k: double
child 3, majority_at_k: double
child 30, 31: struct<k: int64, expected_tokens: double, pass_at_k: double, majority_at_k: double>
child 0, k: int64
child 1, expected_tokens: double
child 2, pass_at_k: double
child 3, majority_at_k: double
child 31, 32: struct<k: int64, expected_tokens: double, pass_at_k: double, majority_at_k: double>
child 0, k: int64
child 1, expected_tokens: double
child 2, pass_at_k: double
child 3, majority_at_k: double
aggregate_by_k_doc: string
committed: bool
problem_key: string
label: string
output_tokens: int64
finish_reason: string
hit_token_cap: bool
completion: string
rollout_index: int64
correct: bool
steer: 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
pass@1: double
pass@1_ci95: list<item: double>
child 0, item: double
pass@1_sd_boot: double
pass@1_by_rollout_index: list<item: double>
child 0, item: double
pass@1_sd_rollouts: double
pass@8: double
pass@8_ci95: list<item: double>
child 0, item: double
pass@8_sd_boot: double
pass@16: double
pass@16_ci95: list<item: double>
child 0, item: double
pass@16_sd_boot: double
cap_rate: double
n_problems: int64
n_rollouts: int64
rollouts_per_problem: int64
max_tokens_used: list<item: int64>
child 0, item: int64
prompt_style: list<item: null>
child 0, item: null
median_output_tokens: int64
mean_output_tokens: double
expected_problems: int64
complete: bool
files: list<item: string>
child 0, item: string
per_problem: struct<aime2024_0: struct<rollout_index: list<item: int64>, correct: list<item: int64>, tokens: list (... 8176 chars omitted)
child 0, aime2024_0: struct<rollout_index: list<item: int64>, correct: list<item: int64>, tokens: list<item: int64>, answ (... 161 chars omitted)
child 0, rollout_index: list<item: int64>
child 0, item: int64
child 1, correct: list<item: int64>
child 0, item: int64
child 2, tokens: list<item: int64>
child 0, item: int64
child 3, answer: list<item: string>
child 0, item: string
child 4, hit_cap: list<item: int64>
child 0, item: int64
child 5, answer_end_frac: list<item: double>
child 0, item: double
child 6, answer_marker: list<item: s
...
child 0, k: int64
child 1, expected_tokens: double
child 2, pass_at_k: double
child 3, majority_at_k: double
child 27, 28: struct<k: int64, expected_tokens: double, pass_at_k: double, majority_at_k: double>
child 0, k: int64
child 1, expected_tokens: double
child 2, pass_at_k: double
child 3, majority_at_k: double
child 28, 29: struct<k: int64, expected_tokens: double, pass_at_k: double, majority_at_k: double>
child 0, k: int64
child 1, expected_tokens: double
child 2, pass_at_k: double
child 3, majority_at_k: double
child 29, 30: struct<k: int64, expected_tokens: double, pass_at_k: double, majority_at_k: double>
child 0, k: int64
child 1, expected_tokens: double
child 2, pass_at_k: double
child 3, majority_at_k: double
child 30, 31: struct<k: int64, expected_tokens: double, pass_at_k: double, majority_at_k: double>
child 0, k: int64
child 1, expected_tokens: double
child 2, pass_at_k: double
child 3, majority_at_k: double
child 31, 32: struct<k: int64, expected_tokens: double, pass_at_k: double, majority_at_k: double>
child 0, k: int64
child 1, expected_tokens: double
child 2, pass_at_k: double
child 3, majority_at_k: double
aggregate_by_k_doc: string
committed: bool
problem_key: string
label: string
output_tokens: int64
finish_reason: string
hit_token_cap: bool
completion: string
rollout_index: int64
correct: bool
steer: 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 |
|---|---|---|---|---|---|---|---|---|---|---|---|
aime2024_0 | 0 | none | base | run | true | false | false | stop | 888 | 31,744 | Let's start by analyzing the given information.
When Aya walks at a constant speed of $s$ km/h, the total time taken for the walk, including the time spent in the coffee shop, is 4 hours and $t$ minutes. Since $t$ minutes is a fraction of an hour, we convert it to hours: $t/60$ hours. Therefore, the total time is $4... |
aime2024_0 | 1 | none | base | run | true | false | false | stop | 775 | 31,744 | Let's denote Aya's speed as $s$ km/h and the time she spends in the coffee shop as $t$ minutes. From the problem, we have two scenarios:
1. When she walks at speed $s$ km/h, the total time is 4 hours (240 minutes) including $t$ minutes in the coffee shop. Therefore:
\[ \frac{9}{s} = 240 + \frac{t}{60} \]
\[ \frac{9}... |
aime2024_0 | 2 | none | base | run | true | true | false | stop | 1,214 | 31,744 | To solve this problem, we can start by using the information about Aya's walking times to set up equations.
1. When Aya walks at a constant speed of $s$ kilometers per hour, the walk takes her 4 hours, including $t$ minutes spent in the coffee shop. The distance covered in this case is $9$ kilometers.
The time taken... |
aime2024_0 | 3 | none | base | run | true | true | false | stop | 1,185 | 31,744 | Let's start by understanding the given information and setting up equations based on the problem.
Let $s$ be the constant speed at which Aya walks. When she walks at this speed, the total time taken for the 9-kilometer walk, including the time spent in the coffee shop, is 4 hours. The time spent walking is $\frac{9}{... |
aime2024_0 | 4 | none | base | run | true | false | false | stop | 1,118 | 31,744 | Let's first analyze the given information:
1. When walking at speed $s$ km/h, the distance covered is 9 km, and the time taken is 4 hours, including $t$ minutes in the coffee shop. This gives us the equation:
\[ 9 = s \times 4 + \frac{t}{60} \]
\[ 9 = 4s + \frac{t}{60} \]
\[ 360 = 240s + t \]
\[ t = 360 - 240s \quad ... |
aime2024_0 | 5 | none | base | run | true | false | false | stop | 3,462 | 31,744 | Let's denote the time Aya walks at $s$ km/h as $t_1$ hours and the time at $s+2$ km/h as $t_2$ hours. Given that she walks 9 km in total:
\[ t_1 \cdot s = 9 \]
\[ t_2 \cdot (s+2) = 9 \]
Also, the time spent in the coffee shop is the same in both cases, so:
\[ t = 4 - t_1 = 2\frac{2}{5} - t_2 \]
From the first equati... |
aime2024_0 | 6 | none | base | run | true | false | false | stop | 3,386 | 31,744 | Let us first determine the value of $s$. When she walks at a constant speed of $s$ kilometers per hour, the walk takes her 4 hours, including $t$ minutes spent in the coffee shop. Thus, the total time taken is 4 hours and $t$ minutes, which is equal to $\frac{240 + t}{60}$ hours. The distance covered in this case is $... |
aime2024_0 | 7 | none | base | run | false | false | false | stop | 15,198 | 31,744 | Let's first convert the times into hours.
For the first case:
\[ \text{Time} = \frac{9}{s} + \frac{t}{60} = 4 \]
\[ \frac{9}{s} = 4 - \frac{t}{60} \]
\[ \frac{9}{s} = \frac{240 - t}{60} \]
\[ 540 = s(240 - t) \]
\[ 540 = 240s - st \]
\[ 540 = 240s - st \quad (1) \]
For the second case:
\[ \text{Time} = \frac{9}{s+2}... |
aime2024_0 | 8 | none | base | run | true | false | false | stop | 1,827 | 31,744 | Let's denote the time Aya spends walking at $s$ km/h as $t_1$ hours and at $s+2$ km/h as $t_2$ hours. We know that the total time for each case is given, including the time spent in the coffee shop.
For the first case, walking at $s$ km/h:
\[ \text{Time} = \frac{\text{Distance}}{\text{Speed}} = \frac{9}{s} \text{ hou... |
aime2024_0 | 9 | none | base | run | true | true | false | stop | 1,070 | 31,744 | Let's break down the information given:
1. **First Scenario (Constant Speed $s$):**
- Total distance = 9 km
- Time taken = 4 hours = 240 minutes
- Time spent in the coffee shop = $t$ minutes
- Therefore, the actual walking time = 240 - $t$ minutes
- Walking speed = $s$ km/h, so walking time = $\frac{9}... |
aime2024_0 | 10 | none | base | run | true | false | false | stop | 2,419 | 31,744 | Let's denote the time Aya takes for the walk as $t_1$ when walking at $s$ km/h and $t_2$ when walking at $s+2$ km/h. We know that $t_1 = 4$ hours and $t_2 = 2 + \frac{24}{60} = 2.4$ hours.
First, let's express the time taken for each scenario using the formula $time = \frac{distance}{speed}$.
For the first scenario,... |
aime2024_0 | 11 | none | base | run | true | false | false | stop | 11,345 | 31,744 | Let's denote the time spent walking at $s$ km/h as $t_1$ hours and at $s+2$ km/h as $t_2$ hours. We know that:
1. For $s$ km/h, the total time including the coffee shop is 4 hours, so $t_1 + t_{coffee} = 4$. Given $t_{coffee} = t$ minutes, $t_1 = 4 - \frac{t}{60}$ hours.
2. For $s+2$ km/h, the total time including t... |
aime2024_0 | 12 | none | base | run | true | false | false | stop | 15,210 | 31,744 | Let's start by setting up the equations based on the information given. When Aya walks at a constant speed of $s$ kilometers per hour, the time taken for the walk is 4 hours, and she spends $t$ minutes in the coffee shop. The total time including the coffee shop is 4 hours and $t$ minutes, which is 4 hours and $\frac{... |
aime2024_0 | 13 | none | base | run | false | false | false | stop | 6,828 | 31,744 | First, we can find the time it takes for the first scenario to understand the relationship between speed and time. When Aya walks at a constant speed $s$ kilometers per hour, the walk takes her 4 hours. Therefore, the distance covered is $9$ kilometers, so we can write:
\[ 9 = 4s \Rightarrow s = \frac{9}{4} = 2.25 \te... |
aime2024_0 | 14 | none | base | run | true | false | false | stop | 966 | 31,744 | Let's denote the time spent in the coffee shop as $t$ minutes. The total time for the walk at speed $s$ is $4$ hours, which is $240$ minutes, so:
\[ \frac{9}{s} \times 60 = 240 - t \]
\[ \frac{540}{s} = 240 - t \]
\[ 540 = s(240 - t) \]
Similarly, for the walk at speed $s+2$:
\[ \frac{9}{s+2} \times 60 = 2 \times 60 +... |
aime2024_0 | 15 | none | base | run | true | false | false | stop | 1,898 | 31,744 |
First, we can find the value of $s$ by using the information about the two speeds. Let's denote the distance as $d = 9$ kilometers and the time for the first speed as $t_1$ and for the second speed as $t_2$.
For the first speed ($s$ km/h):
\[ t_1 = \frac{d}{s} + \frac{t}{60} \]
Given that $t_1 = 4$ hours, we have:
\... |
aime2024_1 | 0 | none | base | run | true | false | false | stop | 1,617 | 31,744 | We know that $BD=DC=8$ and $AD=15$ by the Two Tangent Theorem. We can use power of a point to find $BP=CP=20$.
Let $AP=x$. By Stewart's Theorem, we have
\[ 15^2\cdot 10 + 8^2\cdot 20 = x^2\cdot 9 + 20^2\cdot 15 \]
\[ 2250 + 1280 = 9x^2 + 6000 \]
\[ 9x^2 = -2650 \]
\[ x^2 = -\frac{2650}{9} \]
\[ x = \sqrt{-\frac{2650}... |
End of preview.
rollouts-smollm3-rl — evaluation rollouts (boxed prompt)
Model: reasoning-cues/smollm3-rl @ 97bda61 — a GRPO LoRA adapter (r=64, alpha=128, all attention and
MLP projections) on HuggingFaceTB/SmolLM3-3B-Base @ d78a42f, merged with peft merge_and_unload
(bf16) and served with the base model's tokenizer files (the adapter repo's tokenizer_config.json
declares TokenizersBackend; its tokenizer.json is byte-identical to the base's). Protocol: the paper's
released-model grid — 32 rollouts per problem (two seeded batches of 16, seeds 20260819 / 20260820),
temperature 0.6, top-p 0.95, 31,744 generated tokens in a 32,768 window, full problem set of each
benchmark. Prompt: boxed (data/bank). Arms: none and the cue of record sp_nn_the = " \n\nThe"
(SmolLM3's own boxed-prompt opening from the label-free beam search). Prompts and grader: the paper's
repository (sophicle/reason); requested in collab/rulin/2026-09-18_smollm3_rl_boxed.md.
Rollouts generated on a GB300 cluster with vLLM 0.29.0 (SmolLM3 through vLLM's Transformers backend;
the pinned 0.27.1 has no wheel for that architecture); all other pins as in the repo. HumanEval is
execution-graded (run_execgraded/). Rollout jsonl files are kept as written under
boxed/<benchmark>/run/; cell summaries (with answer_end_tokens from the model tokenizer and
aggregate_by_k) are at boxed/<benchmark>/summary/<arm>.json — summaries for HumanEval live under
boxed/humaneval_execgraded/summary/. The same summaries are committed on the results-rulin branch
under results/rulin/smollm3_rl/boxed/.
| benchmark | none |
sp_nn_the |
|---|---|---|
| MATH-500 | 0.544 | 0.608 |
| GSM8K | 0.557 | 0.732 |
| AMC 23 | 0.290 | 0.331 |
| AIME 2024 | 0.031 | 0.056 |
| AIME 2025 | 0.027 | 0.045 |
| OlympiadBench | 0.262 | 0.283 |
| GPQA Diamond (MCQ) | 0.121 | 0.138 |
| HumanEval (exec) | 0.535 | 0.542 |
pass@1 over 500 (or the full set) × 32; every cell complete: true.
- Downloads last month
- 27