haowu89/math-ai-bench-sources-high
收藏Hugging Face2026-04-15 更新2026-04-26 收录
下载链接：
https://hf-mirror.com/datasets/haowu89/math-ai-bench-sources-high
下载链接
链接失效反馈官方服务：
资源简介：
---
configs:
- config_name: aime25
data_files:
- split: train
path: aime25/train-*
- config_name: aime26
data_files:
- split: train
path: aime26/train-*
- config_name: apex_2025
data_files:
- split: train
path: apex_2025/train-*
- config_name: arxivmath
data_files:
- split: train
path: arxivmath/train-*
- config_name: cmimc_2025
data_files:
- split: train
path: cmimc_2025/train-*
- config_name: gpqa_diamond
data_files:
- split: train
path: gpqa_diamond/train-*
- config_name: hmmt_feb_2026
data_files:
- split: train
path: hmmt_feb_2026/train-*
- config_name: hmmt_nov_2025
data_files:
- split: train
path: hmmt_nov_2025/train-*
- config_name: imobench
data_files:
- split: train
path: imobench/train-*
- config_name: olympiadbench
data_files:
- split: train
path: olympiadbench/train-*
- config_name: theoremqa
data_files:
- split: train
path: theoremqa/train-*
---
# math-ai-bench-sources-high
Question-level subsets built from `haowu89/math-ai-bench-sources-final`.
Each row keeps one benchmark question and 8 selected reasoning trajectories.
Fields:
- `problem`, `original_solution`, `answer`, `source`, `index`
- `model`: mixed quality bucket identifier
- `generated_solutions`: 8 selected trajectories
- `count`: always 8
- `context_metadata`: token counts, correctness labels, selected source models, and selected trajectory indices
Available configs:
- `aime25`
- `aime26`
- `apex_2025`
- `arxivmath`
- `cmimc_2025`
- `gpqa_diamond`
- `hmmt_feb_2026`
- `hmmt_nov_2025`
- `imobench`
- `olympiadbench`
- `theoremqa`
提供机构：
haowu89
