Dataset Viewer
Auto-converted to Parquet Duplicate
context
listlengths
1
1
gen_kwargs
dict
repeat_idx
int64
0
9
request_idx
int64
0
514
task_name
stringclasses
9 values
metadata
dict
model_outputs
stringlengths
897
278k
[ { "content": "Generate an executable Python function generated from the given prompt. The function should take stdin as input and print the output. Simply call the function after the definition.There are N competitive programmers numbered person 1, person 2, \\ldots, and person N.\nThere is a relation called su...
{ "do_sample": false, "max_new_tokens": 32768, "seed": 0, "temperature": 0.7 }
0
256
LiveCodeBench
{ "expected_answer": "", "problem_id": "", "reference_solution": "", "subject": null, "type": null }
<think> Okay, I need to solve this problem where I have to determine if there's a unique strongest programmer based on the given information. Let's think about how to approach this. First, the problem states that the superiority relation is a total order because for any two distinct programmers, one is stronger than t...
[ { "content": "In this problem, the final will be a numeric value. Give the numerical answer correct upto the 2nd decimal digit. Give a detailed solution and end the solution with the final answer. Mark your solution with \\boxed\nAnswer:\n\nProblem: 2 mol of $\\mathrm{Hg}(g)$ is combusted in a fixed volume bomb...
{ "do_sample": false, "max_new_tokens": 32768, "seed": 2, "temperature": 0.7 }
2
400
JEEBench
{ "expected_answer": "90.39", "problem_id": "37", "reference_solution": null, "subject": "chem", "type": "Numeric" }
<think> Alright, so I need to find the standard molar enthalpy of formation of HgO(s) at 298 K. The problem gives me some information about a combustion reaction in a bomb calorimeter, including the temperature change, the heat capacity of the bomb, the enthalpy of formation of Hg(g), and the gas constant. Let me try t...
[ { "content": "You are a coding expert. Given a competition-level coding problem, you need to write a Python program to solve it. You may start by outlining your thought process. In the end, please provide the complete code in a code block enclosed with ``` ```. The code should take stdin as input and print the ...
{ "do_sample": false, "max_new_tokens": 32768, "seed": 1, "temperature": 0.7 }
1
363
CodeForces
{ "expected_answer": "", "problem_id": "", "reference_solution": "", "subject": null, "type": null }
<think> Okay, let's see. I need to find the maximum number of points Kevin can earn by rearranging the array. Hmm, the problem says that each time Kevin adds an a_i to s, and if the new s is even, he can earn a point by dividing by 2 until it's odd. But he can only get one point per operation. So the key is to maximize...
[ { "content": "Return your final response within \\boxed{} and only include the letter choice (A, B, C, or D) as your final response.\nProblem: How many of the stars listed below would be detectable using the ESPRESSO spectrograph, when it is coupled with one of the 8m VLT telescopes at the Paranal Observatory? ...
{ "do_sample": true, "max_new_tokens": 32768, "seed": 0, "temperature": 0.7 }
0
186
GPQADiamond
{ "expected_answer": "", "problem_id": "", "reference_solution": "", "subject": null, "type": null }
<think> Okay, so I need to figure out how many of these stars are detectable by the ESPRESSO spectrograph when using one of the 8m VLT telescopes. The question specifies that a signal-to-noise ratio (S/N) of at least 10 per binned pixel during a 1-hour exposure is required. The options are A) 4, B) 2, C)5, D)3. First...
[{"content":"You are a coding expert. Given a competition-level coding problem, you need to write a (...TRUNCATED)
{ "do_sample": false, "max_new_tokens": 32768, "seed": 0, "temperature": 0.7 }
0
321
CodeForces
{ "expected_answer": "", "problem_id": "", "reference_solution": "", "subject": null, "type": null }
"<think> Okay, I need to solve this problem where we have a deck of cards and a joker's position. Th(...TRUNCATED)
[{"content":"Generate an executable Python function generated from the given prompt. Return the func(...TRUNCATED)
{ "do_sample": false, "max_new_tokens": 32768, "seed": 2, "temperature": 0.7 }
2
167
LiveCodeBench
{ "expected_answer": "", "problem_id": "", "reference_solution": "", "subject": null, "type": null }
"<think> Okay, so I need to find the total number of incremovable subarrays in the given array. Hmm.(...TRUNCATED)
[{"content":"\"You are a coding expert. Given a competition-level coding problem, you need to write (...TRUNCATED)
{ "do_sample": false, "max_new_tokens": 32768, "seed": 2, "temperature": 0.7 }
2
152
CodeElo
{ "expected_answer": "", "problem_id": "", "reference_solution": "", "subject": null, "type": null }
"<think> Okay, so I need to solve this problem where we have to find the earliest time when all the (...TRUNCATED)
[{"content":"\"You are a coding expert. Given a competition-level coding problem, you need to write (...TRUNCATED)
{ "do_sample": false, "max_new_tokens": 32768, "seed": 0, "temperature": 0.7 }
0
31
CodeElo
{ "expected_answer": "", "problem_id": "", "reference_solution": "", "subject": null, "type": null }
"<think> Okay, I need to solve this problem where I have a tree and perform a series of toggle queri(...TRUNCATED)
[{"content":"Generate an executable Python function generated from the given prompt. The function sh(...TRUNCATED)
{ "do_sample": false, "max_new_tokens": 32768, "seed": 1, "temperature": 0.7 }
1
359
LiveCodeBench
{ "expected_answer": "", "problem_id": "", "reference_solution": "", "subject": null, "type": null }
"<think> Okay, so I need to write a Python function that generates a valid arrangement of Takahashi (...TRUNCATED)
[{"content":"Problem: Rectangle $ABCD$ has center $O$ and $AB/AD=k$. A point is randomly chosen from(...TRUNCATED)
{ "do_sample": false, "max_new_tokens": 32768, "seed": 0, "temperature": 0.7 }
0
444
MATH500
{ "expected_answer": "", "problem_id": "", "reference_solution": "", "subject": null, "type": null }
"<think> Okay, so I have this problem about a rectangle ABCD with center O, and the ratio of AB to A(...TRUNCATED)
End of preview. Expand in Data Studio

YAML Metadata Warning:empty or missing yaml metadata in repo card

Check out the documentation for more information.

mlfoundations-dev/openthoughts3_herorun_ckpt01000_eval_636d

Precomputed model outputs for evaluation.

Evaluation Results

Summary

Metric AIME24 AMC23 MATH500 MMLUPro JEEBench GPQADiamond LiveCodeBench CodeElo CodeForces
Accuracy 41.7 82.2 85.2 0.4 50.9 39.1 40.8 12.7 16.3

AIME24

  • Average Accuracy: 41.67% ± 1.51%
  • Number of Runs: 10
Run Accuracy Questions Solved Total Questions
1 36.67% 11 30
2 46.67% 14 30
3 43.33% 13 30
4 33.33% 10 30
5 43.33% 13 30
6 43.33% 13 30
7 36.67% 11 30
8 40.00% 12 30
9 43.33% 13 30
10 50.00% 15 30

AMC23

  • Average Accuracy: 82.25% ± 1.03%
  • Number of Runs: 10
Run Accuracy Questions Solved Total Questions
1 85.00% 34 40
2 87.50% 35 40
3 77.50% 31 40
4 80.00% 32 40
5 80.00% 32 40
6 82.50% 33 40
7 85.00% 34 40
8 82.50% 33 40
9 85.00% 34 40
10 77.50% 31 40

MATH500

  • Accuracy: 85.20%
    Accuracy Questions Solved Total Questions
    85.20% 426 500

MMLUPro

  • Accuracy: 0.38%
    Accuracy Questions Solved Total Questions
    0.38% N/A N/A

JEEBench

  • Average Accuracy: 50.94% ± 1.29%
  • Number of Runs: 3
Run Accuracy Questions Solved Total Questions
1 48.25% 248.5 515
2 50.83% 261.75 515
3 53.74% 276.75 515

GPQADiamond

  • Average Accuracy: 39.06% ± 0.50%
  • Number of Runs: 3
Run Accuracy Questions Solved Total Questions
1 39.39% 78 198
2 39.90% 79 198
3 37.88% 75 198

LiveCodeBench

  • Average Accuracy: 40.77% ± 0.53%
  • Number of Runs: 3
Run Accuracy Questions Solved Total Questions
1 41.49% 212 511
2 41.10% 210 511
3 39.73% 203 511

CodeElo

  • Average Accuracy: 12.70% ± 0.60%
  • Number of Runs: 3
Run Accuracy Questions Solved Total Questions
1 13.30% 52 391
2 13.30% 52 391
3 11.51% 45 391

CodeForces

  • Average Accuracy: 16.26% ± 0.07%
  • Number of Runs: 3
Run Accuracy Questions Solved Total Questions
1 16.34% 74 453
2 16.34% 74 453
3 16.11% 73 453
Downloads last month
3