Dataset Viewer
Auto-converted to Parquet Duplicate
problem
string
expected_answer
string
original_expected_answer
string
changed_answer_to_majority
bool
pass_bucket
int64
answer_tier
int64
data_source
string
license
string
url
string
user_name
string
user_url
string
metadata
string
traces_high_no_tool
list
n_traces
int64
Find the limit $\lim_{x\to\infty}(\ln(a+x)-\ln(a-x))$.
-\,i\pi
null
true
4
2
stackflow
cc-by-sa-4.0
https://math.stackexchange.com/questions/2828085/limit-of-logarithm
jahly
https://math.stackexchange.com/users/376538/jahly
{"reason_low_with_tool": {"count": 8, "pass": 1, "accuracy": 0.125}, "reason_low_no_tool": {"count": 8, "pass": 1, "accuracy": 0.125}, "reason_medium_with_tool": {"count": 8, "pass": 3, "accuracy": 0.375}, "reason_medium_no_tool": {"count": 8, "pass": 0, "accuracy": 0.0}, "reason_high_with_tool": {"count": 8, "pass": 3...
[ "[{\"role\": \"user\", \"content\": \"Solve the following math problem. Make sure to put the answer (and only answer) inside \\\\boxed{}.\\n\\nFind the limit $\\\\lim_{x\\\\to\\\\infty}(\\\\ln(a+x)-\\\\ln(a-x))$.\", \"reasoning_content\": \"\", \"tool_calls\": [], \"tool_call_id\": \"\", \"name\": \"\"}, {\"role\":...
1
"100 blocks are selected from a crate containing 33 blocks of each of the following dimensions: \\( (...TRUNCATED)
595
null
true
7
1
aops
cc-by-4.0
null
null
null
"{\"reason_low_with_tool\": {\"count\": 8, \"pass\": 1, \"accuracy\": 0.125}, \"reason_low_no_tool\"(...TRUNCATED)
["[{\"role\": \"user\", \"content\": \"Solve the following math problem. Make sure to put the answer(...TRUNCATED)
6
"Find the least $C$ such that, for any initial ordering of 200 people (100 couples), they can be rea(...TRUNCATED)
1
99
true
4
1
aops
cc-by-4.0
null
null
null
"{\"reason_low_with_tool\": {\"count\": 8, \"pass\": 1, \"accuracy\": 0.125}, \"reason_low_no_tool\"(...TRUNCATED)
["[{\"role\": \"user\", \"content\": \"Solve the following math problem. Make sure to put the answer(...TRUNCATED)
2
"What is the probability that the sum of four randomly chosen numbers (with repetitions allowed) fro(...TRUNCATED)
\(\frac{1378}{397953}\)
\(\frac{1378}{397953}\)
false
5
2
stackflow
cc-by-sa-4.0
https://math.stackexchange.com/questions/2978148/hello-small-probability-related-question
user610402
https://math.stackexchange.com/users/610402/user610402
"{\"reason_low_with_tool\": {\"count\": 8, \"pass\": 8, \"accuracy\": 1.0}, \"reason_low_no_tool\": (...TRUNCATED)
["[{\"role\": \"user\", \"content\": \"Solve the following math problem. Make sure to put the answer(...TRUNCATED)
1
"How many different colorings can be obtained by coloring a circle divided into 60 pieces using 4 di(...TRUNCATED)
23713472717216429668046350036152.
23713472717216429668046350036152.
false
6
2
stackflow
cc-by-sa-4.0
https://math.stackexchange.com/questions/3508271/polya-enumeration
M-S-R
https://math.stackexchange.com/users/603093/m-s-r
"{\"reason_low_with_tool\": {\"count\": 8, \"pass\": 6, \"accuracy\": 0.75}, \"reason_low_no_tool\":(...TRUNCATED)
["[{\"role\": \"user\", \"content\": \"Solve the following math problem. Make sure to put the answer(...TRUNCATED)
1
"One-inch squares are cut from the corners of a 5-inch square. What is the area in square inches of (...TRUNCATED)
15
15
false
4
1
aops
cc-by-4.0
null
null
null
"{\"reason_low_with_tool\": {\"count\": 8, \"pass\": 0, \"accuracy\": 0.0}, \"reason_low_no_tool\": (...TRUNCATED)
["[{\"role\": \"user\", \"content\": \"Solve the following math problem. Make sure to put the answer(...TRUNCATED)
4
"Model a ski jump curve that satisfies the following conditions:\n- The curve passes through the poi(...TRUNCATED)
y = 0.00353545\,x^{2}+0.185339\,x
null
true
4
2
aops
cc-by-4.0
null
null
null
"{\"reason_low_with_tool\": {\"count\": 8, \"pass\": 3, \"accuracy\": 0.375}, \"reason_low_no_tool\"(...TRUNCATED)
["[{\"role\": \"user\", \"content\": \"Solve the following math problem. Make sure to put the answer(...TRUNCATED)
4
"If there is some \\( K \\in \\mathbb{Z} \\) such that for all \\( M \\) with \\( |M| = K \\), \\( P(...TRUNCATED)
0
null
true
4
1
stackflow
cc-by-sa-4.0
https://math.stackexchange.com/questions/4636867/are-there-always-finitely-many-sinks-in-a-lattice-with-a-piece-metric
Teddy Astor
https://math.stackexchange.com/users/1128783/teddy-astor
"{\"reason_low_with_tool\": {\"count\": 8, \"pass\": 0, \"accuracy\": 0.0}, \"reason_low_no_tool\": (...TRUNCATED)
["[{\"role\": \"user\", \"content\": \"Solve the following math problem. Make sure to put the answer(...TRUNCATED)
1
"A box contains 100 tickets, labeled from 1 to 100. You are given up to three times to randomly pick(...TRUNCATED)
70.03
70.03
false
5
1
stackflow
cc-by-sa-4.0
https://math.stackexchange.com/questions/3864020/picking-a-number-out-of-1-to-100-three-times
cic team4
https://math.stackexchange.com/users/836855/cic-team4
"{\"reason_low_with_tool\": {\"count\": 8, \"pass\": 4, \"accuracy\": 0.5}, \"reason_low_no_tool\": (...TRUNCATED)
["[{\"role\": \"user\", \"content\": \"Solve the following math problem. Make sure to put the answer(...TRUNCATED)
1
"Compute the Laplace Transform of the function \\( f_3(t) = t - \\left\\lfloor t \\right\\rfloor \\)(...TRUNCATED)
\(\frac{1+s-e^s}{s^2(1-e^s)}\)
\(\frac{1+s-e^s}{s^2(1-e^s)}\)
false
6
2
stackflow
cc-by-sa-4.0
https://math.stackexchange.com/questions/2763571/how-to-compute-the-laplace-transforms-of-summation
user490308
null
"{\"reason_low_with_tool\": {\"count\": 8, \"pass\": 3, \"accuracy\": 0.375}, \"reason_low_no_tool\"(...TRUNCATED)
["[{\"role\": \"user\", \"content\": \"Solve the following math problem. Make sure to put the answer(...TRUNCATED)
1
End of preview. Expand in Data Studio

Nemotron-Math-v2-10k

10,000 machine-verifiable, mid-to-hard math problems subsampled from nvidia/Nemotron-Math-v2, each with all stored gpt-oss-120b high-reasoning, no-tool solution traces. Built for RL training (verifiable final answers) with a teacher-trace column for SFT distillation / analysis.

How it was made (seed 42 throughout)

  1. Dedupe: scanned all 5 splits (high_part00-02, medium, low), deduped rows by md5(problem) → 324,233 unique problems.
  2. Verifiable-answer filter: kept problems whose expected_answer passes tier ≤ 2 of a whitelist filter (answer_filter.py in the scale models repo, tmp/scalar-ppo/docs-ppo/scripts/): tier 1 = plain number, tier 2 = short closed-form (whitelisted LaTeX commands, single-letter variables, ≤ 40 chars, no \text/\sum/ matrices/relations). This removes answer shapes that math_verify-style graders can't check reliably (67% of raw Nemotron answers fail it).
  3. Difficulty buckets: bucketed by metadata.reason_high_no_tool.pass — how many of the 8 gpt-oss-120b attempts (high reasoning, no tools) were verified correct. Kept buckets 4, 5, 6, 7 (mid-to-hard band; excludes both never-solved and saturated problems). Tier≤2 pool sizes: {4: 7796, 5: 8483, 6: 9481, 7: 12744}.
  4. Sample: 2,500 random problems per bucket = 10,000.
  5. Teacher traces: for each problem, collected every stored row from the high_part* splits with tools == [] (high-reasoning no-tool generations) and kept the distinct messages payloads (first-seen order) in traces_high_no_tool. Note the source release heavily duplicates payloads — a problem's no-tool row count tracks its pass statistics, but those rows carry only 1..pass distinct solutions (byte-identical repeats otherwise). Distinct-trace distribution over the 10k: 1 traces: 6818 · 2 traces: 172 · 3 traces: 379 · 4 traces: 738 · 5 traces: 731 · 6 traces: 643 · 7 traces: 458 · 8 traces: 11 · 9 traces: 10 · 10 traces: 9 · 11 traces: 9 · 12 traces: 7 · 13 traces: 4 · 14 traces: 8 · 15 traces: 2 · 18 traces: 1 (9914/10000 problems have n_traces <= pass_bucket).
  6. Split: shuffled, first 200 → validation, remaining 9,800 → train (validation bucket mix: {4: 38, 5: 58, 6: 44, 7: 60}).

Trace caveats

  • Traces were verified by NVIDIA's pipeline, not ours. In a 150-problem spot check with a strict $-wrapped math_verify against expected_answer, ~77% of distinct traces' final \boxed{} answers verify; the rest are mostly verbose boxed formats (\text{...} prose, approximations like n \approx 9.4456) rather than wrong math. Re-filter per-trace if you need strict-format SFT targets.
  • Every trace ends with a \boxed{} answer (0 missing in the spot check).

Columns

column description
problem problem statement (LaTeX)
expected_answer verified final answer (majority-vote corrected where flagged)
original_expected_answer pre-correction answer (differs only when changed_answer_to_majority)
changed_answer_to_majority whether NVIDIA replaced the reference answer by majority vote
pass_bucket gpt-oss-120b high-reasoning no-tool pass count (of 8) — 4..7
answer_tier 1 = plain number, 2 = closed-form expression
data_source aops or stackflow
license per-row source license (cc-by-4.0 for AoPS, cc-by-sa-4.0 for StackExchange)
url, user_name, user_url StackExchange attribution (null for AoPS)
metadata JSON: pass/count/accuracy for all 6 (reasoning × tool) teacher configs
traces_high_no_tool list of JSON message arrays — all distinct stored high-reasoning no-tool teacher traces ([user, assistant]; reasoning_content = thinking, content = final answer with \boxed{})
n_traces len(traces_high_no_tool) (distinct payloads)

License / attribution

Derived from nvidia/Nemotron-Math-v2 (gpt-oss-120b generations). AoPS-sourced rows are CC-BY-4.0; StackExchange-sourced rows are CC-BY-SA-4.0 with per-row attribution in url/user_name/user_url. See the per-row license column.

Downloads last month
42