每日 · ML · AI · CSDaily · ML · AI · CS

每日 arXiv 摘要,
翻成研究者能快速判断的中文。
The daily arXiv digest,
translated to plain English.

每个工作日都有 200+ 篇 ML 新论文,大多数只是噪声。PaperTLDR 逐篇读取,标注 novelty 与 incremental,写出点明方法差异的三句 TLDR,再把匹配分类和关键词的子集发给你。200+ new ML papers land every weekday. Most are noise. PaperTLDR reads them all, tags novelty vs incremental, writes a three-sentence TLDR that names the methodological delta, and emails you the curated subset that matches your categories and keywords.

实时样例Live sample

试用实时 digestTry a live digest

输入 arXiv 分类,在页面里预览 3 条种子 TLDR。Enter an arXiv category and preview three seeded TLDRs inline.

GPT-5.4 论文阅读器GPT-5.4 paper reader

粘贴论文,拿到研究简报Paste a paper and get a research brief

OpenFlowX GPT-5.4 返回 TL;DR、关键主张、方法、局限和 BibTeX。免费 demo 按 IP 每小时 5 次。OpenFlowX GPT-5.4 returns a TL;DR, key claims, method, limitations, and BibTeX. Free demo limit: 5 summaries per hour per IP.

Free includes 3 summaries per day. This demo endpoint enforces 5 per hour per IP.

昨天的 arXiv digestYesterday's arXiv digest

Clearly labeled example · no saved seed digest found

[DEMO] cs.LG digest preview

Three seeded papers are grouped by category, ranked by impact, and rewritten as plain-English TLDRs before the email is sent.

订阅Subscribe

免费版每天 3 篇摘要,直接生效。不需要信用卡,一键退订。Free is 3 summaries per day and activates immediately. No credit card. Unsubscribe is one click.

归档Archive

Past digests stay browseable by category.

浏览归档Browse archive

标注新意Novelty-tagged

Every paper tagged method, dataset, benchmark, survey, position, or other — so you can skim the benchmarks and read the methods.

写方法差异,不复述摘要Methodological deltas, not abstracts

TLDR's second sentence is structurally required to name what's new vs prior work. No "we achieve state-of-the-art on…" filler.

处处保留来源链接Source links everywhere

Every TLDR links back to the arXiv PDF. We read papers so you can read them too — not so you can skip them.

价格Pricing

Pro 和 Lab 当前先登记候补,免费版今天就能用。Pro and Lab are currently waitlist-only. Free is available today.

免费Free

$0

  • 每天 3 篇论文摘要3 paper summaries per day
  • 1 个 arXiv 分类1 arXiv category
  • 每日邮件 digestDaily email digest
  • 网页归档Web archive of past issues

ProPro

$9/mo

  • 更高每日摘要额度Higher daily summary allowance
  • 不限分类Unlimited categories
  • 关键词过滤Keyword filters
  • RSS feed 输出RSS feed output
  • 清晨投递Early morning delivery

实验室版Lab

$49/mo

  • 10 个席位10 seats
  • API 访问API access
  • Slack webhookSlack webhook
  • 优先支持Priority support

常见问题FAQ

Is PaperTLDR generated by an LLM?

Yes - GPT-5.4 through OpenFlowX, tightly scoped with schema validation. We reject any summary that doesn't cite a methodological delta. If the model is uncertain, the TLDR says so rather than hallucinate.

Does this replace reading papers?

No. We surface what's worth your actual reading time. The TLDR is a filter, not a substitute.

What categories are supported?

All arXiv categories with active daily submissions — cs.*, stat.ML, math.*, q-bio.*, eess.*. See /categories.

When does the digest arrive?

Free tier: next business day morning, in your local timezone. Pro timing is planned for within 2 hours of arXiv's nightly cutoff once billing is live.

Can I cancel anytime?

Yes. Free has one-click unsubscribe in every email. Pro and Lab are waitlist-only until billing exists.