Back to datasets
Dataset assetOpen Source CommunityMathematical Problem SolvingMachine Learning Training
MATH-500_L3_beam_N128_B5_D40_T1e-05_0-105
This dataset is used to train models to solve specific problems. It includes problem descriptions, solutions, search traces with values, search methods, ground‑truth answers, and token counts for inputs and outputs. The dataset is split into a training set containing 10 samples. Download size is 57 372 bytes and the total dataset size is 92 301 bytes.
Source
huggingface
Created
Dec 22, 2024
Updated
Dec 23, 2024
Signals
194 views
Availability
Linked source ready
Overview
Dataset description and usage context
Dataset Overview
Dataset Information
-
Features:
- problem: string, the problem statement.
- solution: string, the solution.
- search_trace_with_values: string, the search trace with values.
- search_method: string, the search method.
- ground_truth: string, the ground truth.
- search_input_tokens: int64, number of input tokens for the search.
- search_output_tokens: int64, number of output tokens for the search.
- solution_input_tokens: int64, number of input tokens for the solution.
- solution_output_tokens: int64, number of output tokens for the solution.
-
Data Split:
- train: 10 samples, occupying 92 301 bytes.
-
Download Size: 57 372 bytes
-
Dataset Size: 92 301 bytes
Configuration
- Config Name: default
- Data Files:
- split: train
- path: data/train-*
- Data Files:
Need downstream help?
Pair the dataset with AI analysis and content workflows.
Once the source passes your review, move straight into summarization, transformation, report drafting, or presentation generation with the JuheAI toolchain.