|
Download README.md from DataMuncher-Labs/LSReasoning-125000: direct link, hf CLI and curl.
- Browser
- Download file 886 Bytes
-
https://huggingface.co/datasets/DataMuncher-Labs/LSReasoning-125000/resolve/main/README.md
- Command line
-
hf download hf://datasets/DataMuncher-Labs/LSReasoning-125000/README.md
-
curl -L -o README.md https://huggingface.co/datasets/DataMuncher-Labs/LSReasoning-125000/resolve/main/README.md
886 Bytes
metadata
language:
- en
tags:
- math
- algebra
- reasoning
- equations
- fractions
- word-problems
- logic
- llm-training
- finetuning
- Math
- Algebra
- Reasoning
- Equations
- Fractions
- Equations
- Logic
- Finetuning
pretty_name: LSReasoning-125000
size_categories:
- 100K<n<1M
license: mit
Large-Scale Mathematic Reasoning-125000
Built via a python script
Contents
- Addition
- Subtraction
- Multiplication
- Division
- Linear equations
- Fractional equations
- Two step equations
- Algebra word problems Synthetic algebra-heavy math reasoning dataset.
Each row:
- question
- problem
- how_to_solve
- answer
Useage
Licensed under MIT Free to use for everyone ##Notes It is
- Better to use smaller variants of this dataset for finetuning
- Better to use very small variants of this dataset for evaluation
- Better to use large variants of this dataset for pre-training