Skip to content

Dataset · TeraflopAI

43 billion tokens of financial reports from SEC EDGAR

TeraflopAI/SEC-EDGAR

A massive corpus of company documents and reports from SEC EDGAR for training and evaluating financial language models.

What's inside

The dataset is 590 GB and includes 8 million samples with a total of 43 billion tokens in English. It includes all major documents from the US regulatory database SEC EDGAR, collected using the datamule-python library and official API. The release was prepared jointly by Datamule, Teraflop AI, and Eventual.

What you can do with it

  • Pre-training and fine-tuning specialized financial LLMs
  • Training data extraction and classification models from corporate reports
  • Building RAG systems for analyzing public companies
Tasks
Text generation, Text classification
Languages
English
Size
1M–10M rows
Downloads this month
16.4K
Likes
52

How to load it

from datasets import load_dataset

ds = load_dataset("TeraflopAI/SEC-EDGAR")