Dataset · tokyotech-llm
Large English corpus for mathematical generation
tokyotech-llm/swallow-math
Million-scale set of mathematical texts from Tokyo Tech for training generation models.
What's inside
Contains English text data for the text-generation task, with a volume ranging from 1 to 10 million records. The dataset was created by researchers from Tokyo Tech (rioyokotalab) using specialized text rewriting pipelines. The card also indicates that an updated version, SwallowMath-v2, has already been released.
What you can do with it
- Fine-tuning language models for solving mathematical problems and reasoning
- Pre-training or SFT tuning of specialized mathematical LLMs
- Tasks
- Text generation
- Languages
- English
- Size
- 1M–10M rows
- Downloads this month
- 1K
- Likes
- 50
Models trained on it
- tokyotech-llm/Llama-3.1-Swallow-8B-v0.5 2.1K
- mradermacher/Llama-3.1-Swallow-8B-v0.5-i1-GGUF 738
- mradermacher/Llama-3.1-Swallow-8B-v0.5-GGUF 179
- tokyotech-llm/Llama-3.1-8B-math-ablation-exp2-LR2.5e-5-WD0.1-iter0002500 82
- tokyotech-llm/Llama-3.1-8B-math-ablation-exp2-LR2.5e-5-WD0.1-iter0007500 51
- tokyotech-llm/Llama-3.1-8B-math-ablation-exp2-LR2.5e-5-WD0.1-iter0010000 47
- tokyotech-llm/Llama-3.1-8B-math-ablation-exp2-LR2.5e-5-WD0.1-iter0005000 46
- tokyotech-llm/Llama-3.1-8B-math-ablation-exp2-LR2.5e-5-WD0.1-iter0012500 39
How to load it
from datasets import load_dataset
ds = load_dataset("tokyotech-llm/swallow-math")