Dataset · openai
Grade school math word problems
openai/gsm8k
8.5K diverse grade school math problems with step-by-step solutions for reasoning evaluation.
What's inside
GSM8K is a set of 8.5K high-quality, linguistically diverse English grade school math word problems. The problems require 2 to 8 reasoning steps and are solved using a sequence of basic arithmetic operations. The dataset is designed for question answering on basic mathematics with multi-step reasoning.
What you can do with it
- Fine-tuning models for mathematical reasoning
- Benchmark for LLMs
- Evaluating chain-of-thought
- Tasks
- Text generation
- Languages
- English
- Size
- 10K–100K rows
- Downloads this month
- 1.3M
- Likes
- 2.1K
Models trained on it
- vinhnx90/vt-gwen-2.5-3b-Q4_k_m-gguf 6.4K
- mradermacher/Next-3-Pro-i1-GGUF 4.3K
- cybertruck32489/LFM3-1.7B-RDText generation 3.9K
- PeterAM4/Qwen3-Embedding-0.6B-GGUFSentence similarity 3.7K
- mradermacher/EryriLabs-cymraeg-12B-i1-GGUF 3.6K
- mradermacher/EryriLabs-cymraeg-4B-i1-GGUF 3.6K
- jaredpalmer/kev-27bText classification 3.5K
- mradermacher/Aitana-7B-S-Instruct-i1-GGUFText generation 2.8K
How to load it
from datasets import load_dataset
ds = load_dataset("openai/gsm8k")