For students and early PhD researchers

Get through this week's reading list. Properly.

Papers, lecture readings and journal club picks, broken into cards you swipe through. Tap anything confusing and get it explained in plain English, right where you got stuck.

Not a summariser · 5 papers free in your first 14 days · Works with PDFs and links

The link trick

Reading on arXiv? Type scienze.co.uk/ in front of the link.

The paper opens as cards, straight away. No downloading, no uploading, no faff.

scienze.co.uk/arxiv.org/abs/2509.22202
Try it on a real paper →

The abstract won't get you through the seminar.

A summary tells you what a paper found. Your supervisor will ask how. Scienze keeps every paragraph, equation and figure, and explains the bits that lose you, so you walk in having actually read it.

AI summary

“This paper proposes a novel framework achieving state-of-the-art results across several benchmarks, with implications for future work…”

Thirty seconds later, you couldn't explain the method to a colleague. The gist evaporates because you never touched the reasoning.

Reading with Scienze

We minimise the KL divergence DKL(P ∥ Q) between the target and learned distributions…

Explain

KL divergence measures how different two probability distributions are — think of it as the “cost of pretending Q is P.”

Minimising it here means: make the model's guesses as close as possible to reality.

Upload. Swipe. Tap.

Scienze puts them in order — yours.

Abstract

We propose a framework for aligning learned representations with…

Eq. 3

L = E[log p(x|z)] − β·D(q‖p)

Fig. 2

Reconstruction loss falls sharply after the second phase transition…

§4.1 Methods

Unlike prior approaches, we decouple the encoder from…

Explain

In plain terms: the model learns to compress, then rebuild — and the β knob controls how much detail survives.

Whatever's on the list this week.

Built for the reading that actually gets checked.

Papers

The paper you have to present

Methods, stats and equations explained in place. Know what the authors did, not just what they claim.

Lecture readings

The chapter before Thursday

Upload the PDF from your module page, swipe through it on the bus, and tap the terms your lecturer assumes you know.

Journal club

The paper everyone pretends they read

Read it properly, highlight what bugged you, and turn up with a question that isn't “so… what was the sample size?”

Found a paper? Save it in one click.

Spotted something on a journal site or your library catalogue? Save it from your browser and swipe through it later.

Free · Available on the Chrome Web Store

Get the extension

Questions students ask before they start.

Start reading

Swap one scroll for one paper.

Open the reading you've been putting off, swipe through it, tap anything unclear.