Datasets:

Modalities:
Text
Formats:
parquet
Languages:
Arabic
ArXiv:
License:
SultanR commited on
Commit
953d474
·
verified ·
1 Parent(s): 9ec0e1f

Update README.md

Browse files
Files changed (1) hide show
  1. README.md +3 -5
README.md CHANGED
@@ -25,9 +25,7 @@ default: minhash_deduped
25
 
26
  AraMix ([https://arxiv.org/abs/2512.18834](https://arxiv.org/abs/2512.18834)) is an Arabic pretraining corpus containing 178 billion tokens across 179 million documents (in the minhash subset). Rather than scraping the web again, AraMix combines seven publicly available Arabic datasets, applies Arabic-specific quality filtering, and performs cross-dataset deduplication.
27
 
28
- We train a 1.4B parameter language model through nanotron on 30 billion tokens to show that the `consensus` subset of AraMix outperforms the previous state-of-the-art, [arabicweb24](https://huggingface.co/datasets/lightonai/ArabicWeb24) (see [Appendix A9 in the Fineweb-2 paper](https://arxiv.org/pdf/2506.20920)) while having more total tokens. Furthermore, the `minhash_deduped` subset performs on-par with nearly 5 times the total number of tokens.
29
-
30
- In this setup, we remove all samples in `consensus` with more than 5 duplicates.
31
 
32
  ## Subsets
33
 
@@ -35,9 +33,9 @@ In this setup, we remove all samples in `consensus` with more than 5 duplicates.
35
  |--------|-----------|--------|-------------|
36
  | `sentence_deduped` | 167.6M | 158.8B | MinHash + sentence-level deduplication |
37
  | `minhash_deduped` | 178.9M | 177.8B | Document-level MinHash deduplication only |
38
- | `consensus` | 47.9M | 54.1B | Documents appearing in 2+ source datasets |
39
 
40
- The consensus subset uses cross-dataset agreement as a signal for quality.
41
 
42
  ## Usage
43
 
 
25
 
26
  AraMix ([https://arxiv.org/abs/2512.18834](https://arxiv.org/abs/2512.18834)) is an Arabic pretraining corpus containing 178 billion tokens across 179 million documents (in the minhash subset). Rather than scraping the web again, AraMix combines seven publicly available Arabic datasets, applies Arabic-specific quality filtering, and performs cross-dataset deduplication.
27
 
28
+ We train a 1.4B parameter language model through nanotron on 30 billion tokens to show that the `matched` subset of AraMix, outperforms the previous state-of-the-art model-free approach, [arabicweb24](https://huggingface.co/datasets/lightonai/ArabicWeb24) (see [Appendix A9 in the Fineweb-2 paper](https://arxiv.org/pdf/2506.20920)) and the previous state-of-the-art model-based approach, [FineWeb2-HQ](https://huggingface.co/datasets/epfml/FineWeb2-HQ) while having more total tokens. Furthermore, the `minhash_deduped` subset performs on-par with nearly 5 times the total number of tokens.
 
 
29
 
30
  ## Subsets
31
 
 
33
  |--------|-----------|--------|-------------|
34
  | `sentence_deduped` | 167.6M | 158.8B | MinHash + sentence-level deduplication |
35
  | `minhash_deduped` | 178.9M | 177.8B | Document-level MinHash deduplication only |
36
+ | `matched` | 47.9M | 54.1B | Documents appearing in 2+ source datasets |
37
 
38
+ The matched subset uses cross-dataset agreement as a signal for quality.
39
 
40
  ## Usage
41