Previous RLLab/eval-set revision: ac33a25597b4951ecf7c821c9266639adf9ee585
aime24: 30 -> 30 questions; 0 prompts changed; all gold strings and row order retained.
aime25: 30 -> 30 questions; 0 prompts changed; all gold strings and row order retained.
gpqa-d: 198 -> 198 questions; 0 added; of the 198 retained questions, 198 are byte-identical, 198 identical up to whitespace, and 198 keep their gold letter.
mmlu-pro-subset: 10681 -> 10681 questions; 0 added; of the 10681 retained questions, 10681 are byte-identical, 10681 identical up to whitespace, and 10681 keep their gold letter.
minervamath: 272 -> 272 questions; problems and order unchanged; gold notation rewritten: -3./2 -> -\frac{3}{2}.
olympiadbench: 674 -> 674 questions; problems and order unchanged; gold rewrites already applied.
Sources:
- [MathArena/aime_2024_I](https://huggingface.co/datasets/MathArena/aime_2024_I/tree/ea5b061c3e8039dc9858defaafc407d04b995e9f)
- [MathArena/aime_2024_II](https://huggingface.co/datasets/MathA…
repair minervamath, olympiadbench (see description)
Previous RLLab/eval-set revision: acbeda874c86be821a4927e4c1f824f63980630a
aime24: 30 -> 30 questions; 0 prompts changed; all gold strings and row order retained.
aime25: 30 -> 30 questions; 0 prompts changed; all gold strings and row order retained.
gpqa-d: 197 -> 198 questions; 1 added; of the 197 retained questions, 8 are byte-identical, 8 identical up to whitespace, and 49 keep their gold letter.
mmlu-pro-subset: 10681 -> 10681 questions; 0 added; of the 10681 retained questions, 10344 are byte-identical, 10681 identical up to whitespace, and 10681 keep their gold letter.
Sources:
- [MathArena/aime_2024_I](https://huggingface.co/datasets/MathArena/aime_2024_I/tree/ea5b061c3e8039dc9858defaafc407d04b995e9f)
- [MathArena/aime_2024_II](https://huggingface.co/datasets/MathArena/aime_2024_II/tree/29d5d31e9b46e215fc24d9b2a3047506823dd101)
- [MathArena/aime_2025](https://huggingface.co/datasets/MathArena/aime_2025/tree/c94da77eb22bbd6439e62a323bec18493a421302)
Preprocessing: remove Asymp…
AIME24/25: MathArena text without diagram code; answers unchanged
Previous RLLab/eval-set revision: c1694e2825191597d289fd15110685ae6b106cbe
aime24: 30 -> 30 questions; 0 prompts changed; all gold strings and row order retained.
aime25: 30 -> 30 questions; 0 prompts changed; all gold strings and row order retained.
Sources:
- [MathArena/aime_2024_I](https://huggingface.co/datasets/MathArena/aime_2024_I/tree/ea5b061c3e8039dc9858defaafc407d04b995e9f)
- [MathArena/aime_2024_II](https://huggingface.co/datasets/MathArena/aime_2024_II/tree/29d5d31e9b46e215fc24d9b2a3047506823dd101)
- [MathArena/aime_2025](https://huggingface.co/datasets/MathArena/aime_2025/tree/c94da77eb22bbd6439e62a323bec18493a421302)
Preprocessing: remove Asymptote blocks (`[asy]`, `<asy>` and the LaTeX `asy` environment) and TikZ `tikzpicture` environments. Preserve all prose, mathematical formulas and tables outside those blocks. Collapse excess blank lines, trim outer whitespace, and append the step-by-step / boxed-answer instruction. Question order and gold formatting are unchanged.…
AIME24/25: MathArena text without diagram code; answers unchanged
Previous RLLab/eval-set revision: c774e009046b9173716066989a392d0851ea9e4a
aime24: 30 -> 30 questions; 25 prompts changed; all gold strings and row order retained.
aime25: 30 -> 30 questions; 18 prompts changed; all gold strings and row order retained.
Sources:
- [MathArena/aime_2024_I](https://huggingface.co/datasets/MathArena/aime_2024_I/tree/ea5b061c3e8039dc9858defaafc407d04b995e9f)
- [MathArena/aime_2024_II](https://huggingface.co/datasets/MathArena/aime_2024_II/tree/29d5d31e9b46e215fc24d9b2a3047506823dd101)
- [MathArena/aime_2025](https://huggingface.co/datasets/MathArena/aime_2025/tree/c94da77eb22bbd6439e62a323bec18493a421302)
Preprocessing: remove Asymptote blocks (`[asy]`, `<asy>` and the LaTeX `asy` environment) and TikZ `tikzpicture` environments. Preserve all prose, mathematical formulas and tables outside those blocks. Collapse excess blank lines, trim outer whitespace, and append the step-by-step / boxed-answer instruction. Question order and gold formatting are unchange…
minervamath: rewrite gold notation that math-verify misreads (values unchanged)
Previous RLLab/eval-set revision: b0c72eb18091847ae39aa73de37af1c44af4c6a3
Problems are untouched. Only the contents of \boxed{...} in `solution` change, and only in notation:
- e-notation golds: math-verify parses 4.5e33 as 4.5*e*33 (Euler's number), so a correct answer written as 4.5 \times 10^{33} was graded wrong. Every rewritten value is exactly equal to the original (checked with Decimal).
- symbolic golds whose trailing one-letter factor math-verify dropped as a unit (\frac{1}{m} t -> 1/m), or that used numpy syntax.
Diff against the previous revision (272 -> 272 rows):
- rows unchanged: 210
- rows whose gold was rewritten (same problem): 62
- problems added: 0
- problems removed: 0
- order of the remaining rows kept: yes
Gold rewrites (contents of \boxed{...}):
- 4.5e33 -> 4.5 \times 10^{33}
- np.arcsin(10/13) -> \arcsin(10/13)
- 3.83e35 -> 3.83 \times 10^{35}
- 8.7e8 -> 8.7 \times 10^{8}
- 4e33 -> 4 \times 10^{33}
- 3.3e12 -> 3.3 \times 10^{12}
- 3e6 -> 3 \times 10^{6}
- 7e…
olympiadbench: rebuild from official Hothan/OlympiadBench
Source: Hothan/OlympiadBench @ 91184b52131e7fc9455fef848035173aea8cc01a, config OE_TO_maths_en_COMP.
Previous RLLab/eval-set revision: b0c72eb18091847ae39aa73de37af1c44af4c6a3
The previous rows were a copy of knoveleng/OlympiadBench, an older snapshot:
- 15 ARML relay problems read "Let $T=T N Y W R$" (the number you will receive) and could not be solved. Upstream now states T; the final answers are unchanged.
- 1 problem (official id 1965, a prose answer that cannot be graded automatically) was removed upstream and is dropped here.
- 1 gold follows upstream, which moved the degree sign to its `unit` field.
Format is unchanged: `problem` = official question + the step-by-step / \boxed{} instruction, `solution` = "\boxed{" + final_answer without "$" + "}".
Diff against the previous revision (675 -> 674 rows):
- rows unchanged: 658
- rows whose gold was rewritten (same problem): 1
- problems added: 15
- problems removed: 16
- order of the remaining rows kept: yes
Gold rewrites (conte…
