🇬🇧 English | 🇷🇺 Русский
Daily signal from the ML noise.
We automatically track, score, and publish the best papers, repositories, and datasets from the ML world — every day.
Selected, not collected.
Our pipeline scans arXiv, GitHub, Hugging Face, and Kaggle daily, scores each item, and publishes the top picks. No hype, no noise. Just signal.
| Type | Sources |
|---|---|
| Papers | arXiv |
| Repos | GitHub |
| Datasets | Hugging Face, Kaggle |
- 📄 Daily feed: @papersdatacode
- 📈 Digests: @papersdatacode_digests
- 🐦 Twitter: @papersdatacode
- 🌐 Russian: @papersdatacode_ru · @papersdatacode_digests_ru
PDC combines automated scoring with quality filters and ranks each work across weekly, monthly, quarterly, and yearly horizons — so you see what actually holds momentum, not just what's new.
Each work is ranked by an internal score across multiple horizons. Top picks at one level pull into the next: weekly winners pool into monthly, monthly into quarterly, quarterly into yearly.
digests/[year]/weekly/— weekly Top 3 (papers, repos, datasets) and recap with TL;DR and Trendsdigests/[year]/monthly/— monthly rankings, recap, and tracking metrics for top picksdigests/[year]/quarterly/— quarterly rankings and recapdigests/[year]/yearly/— year-end rankings and recapdigests/ru/[year]/...— Russian digests in the same structure
Each period has paired files: main digest with top 3 by category and a recap with TL;DR and Trends.
File naming: date-first for chronological sorting (e.g. 2026-05-10-W19.md).
MIT for metadata. Original papers and code retain their respective licenses.