Dataset Viewer
The dataset viewer is not available because its heuristics could not detect any supported data files. You can try uploading some data files, or configuring the data files location manually.
Berna R7 v3 — Tokenized Training Data
Pre-tokenized training data for Berna R7 v3 model.
Contents
| File | Tokens | Size |
|---|---|---|
tokenized/eng.npy |
936,493,453 | 3.75 GB |
tokenized/math.npy |
1,106,090,267 | 4.42 GB |
tokenized/code.npy |
621,688,132 | 2.49 GB |
| Total | 2,664,271,852 | 10.7 GB |
- Format: NumPy
.npyarrays of token IDs (uint32) - Tokenizer: Custom BPE, vocab=32000
- Sequence length: 2048
Usage
import numpy as np
# Memory-mapped load
data = np.load("tokenized/eng.npy", mmap_mode="r")
print(f"Tokens: {len(data):,}")
Note: The Dataset Viewer does not support .npy files. Use hf download to fetch them.
Related
Model: muhamedkamil/berna-r7-v3
Theory (R5): https://doi.org/10.5281/zenodo.23015308
Author: Muhamed Kamil (ORCID)
License
CC BY-NC 4.0 — research and non-commercial use only.
Full text: https://creativecommons.org/licenses/by-nc/4.0/legalcode
- Downloads last month
- 53