arXiv link in, readable paper out
Every symbol, linked to the line that defines it.
Paste an arXiv link. You get the paper as a clean HTML page where hovering any math symbol shows where that paper introduced it, and one click jumps you there. It is the same move your editor makes when you jump to a definition, applied to notation.
Papers already indexed are free to open, with no account. So is building one we have not read yet, up to 3 an hour. An account keeps every paper you build in one place.
What a linked symbol looks like
… all sub-layers, as well as the embedding layers, produce outputs of dimension = 512.
Links come from the paper’s own words. When we cannot find where a symbol was defined we say so and leave it unlinked, rather than inventing a definition.
Start here
Free to open, no account, and they do not use up your weekly quota.
-
arXiv:1706.03762Attention Is All You Need
18 symbols linked
-
arXiv:2006.11239Denoising Diffusion Probabilistic Models
10 symbols linked
-
arXiv:1312.6114Auto-Encoding Variational Bayes
33 symbols linked
-
arXiv:1806.07366Neural Ordinary Differential Equations
10 symbols linked
-
arXiv:1810.04805BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding
7 symbols linked
-
arXiv:2012.13711A sparsity result for the Dynamical Mordell-Lang Conjecture in positive characteristic
22 symbols linked
-
arXiv:2012.12137Projected Stochastic Gradient Langevin Algorithms for Constrained Sampling and Non-Convex Learning
43 symbols linked
-
arXiv:1904.12438Effective approximation of heat flow evolution of the Riemann ξ function, and a new upper bound for the de Bruijn-Newman constant
47 symbols linked
-
arXiv:1905.02743Supermembranes and domain walls in 𝒩=1, D=4 SYM
47 symbols linked
-
arXiv:1905.00181Dirichlet forms and polymer models based on stable processes
90 symbols linked
-
arXiv:2012.13167Localizing virtual cycles for Donaldson-Thomas invariants of Calabi-Yau 4-folds
79 symbols linked
Every paper we have read
29 more, newest first. Opening one of these is free and does not use up your weekly quota either.
-
Extremal eigenvalues of critical Erdős-Rényi graphs
arXiv:1905.03243n/a 2026-08-11 -
Unsupervised Representation Learning with Deep Convolutional Generative Adversarial Networks
arXiv:1511.064341 linked of 3 2026-08-11 -
Very Deep Convolutional Networks for Large-Scale Image Recognition
arXiv:1409.15564 linked of 7 2026-08-11 -
Improving neural networks by preventing co-adaptation of feature detectors
arXiv:1207.05804 linked of 33 2026-08-11 -
The Era of 1-bit LLMs: All Large Language Models are in 1.58 Bits
arXiv:2402.177640 linked of 13 2026-08-11 -
Differential Transformer
arXiv:2410.0525822 linked of 74 2026-08-11 -
Model-Agnostic Meta-Learning for Fast Adaptation of Deep Networks
arXiv:1703.0340014 linked of 36 2026-08-11 -
Sentence-BERT: Sentence Embeddings using Siamese BERT-Networks
arXiv:1908.1008411 linked of 19 2026-08-11 -
U-Net: Convolutional Networks for Biomedical Image Segmentation
arXiv:1505.045976 linked of 22 2026-08-11 -
Universal Model Routing for Efficient LLM Inference
arXiv:2502.0877366 linked of 190 2026-08-11 -
An Algorithm for Optimal Partitioning of Data on an Interval
arXiv:math/030928510 linked of 32 2026-08-11 -
Neural Discrete Representation Learning
arXiv:1711.0093711 linked of 33 2026-08-11 -
Chain-of-Thought Prompting Elicits Reasoning in Large Language Models
arXiv:2201.119031 linked of 16 2026-08-11 -
Playing Atari with Deep Reinforcement Learning
arXiv:1312.560211 linked of 40 2026-08-11 -
Densely Connected Convolutional Networks
arXiv:1608.069939 linked of 16 2026-08-11 -
Deep contextualized word representations
arXiv:1802.0536516 linked of 44 2026-08-11 -
Efficient Estimation of Word Representations in Vector Space
arXiv:1301.37818 linked of 33 2026-08-11 -
Layer Normalization
arXiv:1607.0645027 linked of 95 2026-08-11 -
Learning Transferable Visual Models From Natural Language Supervision
arXiv:2103.000202 linked of 31 2026-08-11 -
Model Evaluation, Model Selection, and Algorithm Selection in Machine Learning
arXiv:1811.1280842 linked of 106 2026-08-11 -
Calculation of prompt diphoton production cross sections at Tevatron and LHC energies
arXiv:0704.000158 linked of 282 2026-08-11 -
The entropy formula for the Ricci flow and its geometric applications
arXiv:math/021115948 linked of 210 2026-08-11 -
Batch Normalization: Accelerating Deep Network Training by Reducing Internal Covariate Shift
arXiv:1502.0316716 linked of 65 2026-08-11 -
Neural Machine Translation by Jointly Learning to Align and Translate
arXiv:1409.047354 linked of 103 2026-08-11 -
Distilling the Knowledge in a Neural Network
arXiv:1503.0253115 linked of 39 2026-08-11 -
Language Models are Few-Shot Learners
arXiv:2005.141658 linked of 99 2026-08-11 -
Generative Adversarial Networks
arXiv:1406.26616 linked of 59 2026-08-11 -
Deep Residual Learning for Image Recognition
arXiv:1512.033854 linked of 9 2026-08-11 -
An Image is Worth 16x16 Words: Transformers for Image Recognition at Scale
arXiv:2010.1192913 linked of 58 2026-08-11
How it works
-
We fetch the source, not the PDF
The paper is converted to real HTML with real MathML, so the text reflows, the symbols are selectable, and screen readers can reach them.
-
We find where each symbol is introduced
Every occurrence of a symbol is tied back to the sentence or equation in this paper that defines it, with the section it came from.
-
You navigate
Hover to see the definition, click to jump to it, and see every other place the same symbol appears light up at once.
We show fewer links on purpose. A symbol we are unsure about is listed as undefined with the reason, and you can see the running numbers on the accuracy page.