filename stringclasses 10
values | token_count int64 140 1B | size_bytes int64 560 4B |
|---|---|---|
fineweb_00.npy | 1,000,000,000 | 4,000,000,000 |
fineweb_01.npy | 1,000,000,000 | 4,000,000,000 |
fineweb_02.npy | 1,000,000,000 | 4,000,000,000 |
fineweb_03.npy | 1,000,000,000 | 4,000,000,000 |
fineweb_04.npy | 1,000,000,000 | 4,000,000,000 |
fineweb_05.npy | 1,000,000,000 | 4,000,000,000 |
fineweb_06.npy | 1,000,000,000 | 4,000,000,000 |
fineweb_07.npy | 1,000,000,000 | 4,000,000,000 |
fineweb_08.npy | 140 | 560 |
held_out_web.npy | 1,000,000 | 4,000,000 |
prepretraining-web-v1
FineWeb (unfiltered) tokenized web data for pre-pretraining experiments. Split into 1B-token .npy chunks. GPT-NeoX tokenizer, uint32 format.
Dataset Info
- Rows: 10
- Columns: 3
Columns
| Column | Type | Description |
|---|---|---|
| filename | Value('string') | Name of the .npy file |
| token_count | Value('int64') | Number of uint32 token IDs in the file |
| size_bytes | Value('int64') | File size in bytes |
Generation Parameters
{
"script_name": "data/upload_data.py",
"model": "N/A (pre-tokenized training data, not model outputs)",
"description": "FineWeb (unfiltered) tokenized web data for pre-pretraining experiments. Split into 1B-token .npy chunks. GPT-NeoX tokenizer, uint32 format.",
"tokenizer": "allenai/gpt-neox-olmo-dolma-v1_5",
"format": "Pre-tokenized uint32 .npy memmap arrays for OLMo-core",
"source": "HuggingFaceFW/fineweb",
"subset": "sample-10BT",
"vocab_size": 50280,
"eos_token_id": 50279,
"total_train_tokens": 8000000140,
"held_out_tokens": 1000000,
"chunk_size": 1000000000,
"num_chunks": 9,
"seed": 42,
"hyperparameters": {},
"input_datasets": []
}
Experiment Documentation
For complete experiment details, see https://github.com/Zayne-sprague/SC-Research-Notes/tree/main/experiments/prepretraining
Usage
from datasets import load_dataset
dataset = load_dataset("reasoning-degeneration-dev/prepretraining-web-v1", split="train")
print(f"Loaded {len(dataset)} rows")
This dataset is tracked in reasoning-degeneration-dev/PROJECT-MANIFEST
- Downloads last month
- 20