HuggingFaceFW/finepdfs
Viewer β’ Updated β’ 476M β’ 57.8k β’ 947
We release large pre-training datasets to accelerate open LLM development. Part of the Hugging Face Science team (hf.co/science)
How Can We Synthesize High-Quality Pretraining Data? A Systematic Study of Prompt Design, Generator Model, and Source Data
FineWeb2: One Pipeline to Scale Them All -- Adapting Pre-Training Data Processing to Every Language