Datasets:
license: cc
pretty_name: DCLM-Deduped
task_categories:
- text-generation
language:
- en
size_categories:
- n>1T
DCLM-Deduped
DCLM is a recently released high quality dataset that uses model-based quality filtering to filter a large subset of common-crawl for similarity to OpenHermes and other instruction-tuning datasets. For reference see the DCLM paper
The original authors of DCLM did not deduplicate their dataset and claimed that deduplication did not improve performance. Nevertheless, when performing our own deduplication of DCLM for Zyda-2, we noticed that DCLM contained a large fraction of duplicates. Specifically, the dataset appears to consist of approximately 80% duplicates.
This lack of impact on downstream performance given this large duplication proportion is perplexing. However, in our own ablations we also replicated this fact. It seems that performing, on average, 5 epochs over the DCLM 'core dataset' is not harmful to language modelling. Nevertheless, the full impacts of this level of duplication on language models are not clear beyond evaluation scores.
As such, we release a fully deduplicated version of DCLM in case it is of interest to the community. DCLM-deduped consists of approximately 750B tokens. If you are planning to pretrain on less than this amount of DCLM tokens it is perhaps safer to use this version than the original DCLM.
How to download
-/ TODO YURY
Breakdown by component
| Dataset | Documents (millions) | gpt-neox tokens (billions) |
|---|---|---|
| DCLM baseline | 2949.3 | 3854.9 |
| DCLM full-deduped | 615.2 | 750.3 |
Dataset Description
- Curated by: Zyphra (deduplicated from DCLM)
- Language(s) (NLP): Primarily English
- License: CC-BY-4
Licensing Information
We are releasing this dataset under the terms of cc-by-4, the same license as the original DCLM dataset.