Papers With Code — Guide

Papers With Code — User Guide

Papers with code (SOTA).

Visit website VPN may be required Free

Strengths

There is a one-to-one correspondence between the paper and the code, and you can directly find the implementation.
Track SOTA (best results) rankings for each task
Completely free, no registration required
Unified indexing of datasets and models
Deep integration with arXiv

Best for

Find the code implementation corresponding to the paper
Understand the current best approach to a task
Discover high-quality open source AI projects
Track progress in a specific research area
Find benchmark data sets

Find the code corresponding to the paper

The core function of Papers With Code is to match papers with open source code implementations.

Scenario

Find the code implementation of the paper

Prompt example

Method 1: Search through papers


1. Search the paper title at paperswithcode.com


2. Click on the paper and view the "Code" tab


3. List all related code repositories


4. Sort by number of stars and choose the most popular ones




Method 2: Extension through arXiv


1. Install Papers With Code browser extension


2. Automatically display code links on arXiv paper pages




Method 3: Search directly on GitHub


Search on GitHub:


"paper title" + "implementation"

Output / what to expect

Find the official or community implementation of the paper,

Directly run the code to verify the paper results,

Significantly saves reproducibility time.

Tips

Give priority to the official implementation (provided by the author), and secondly choose the community implementation with a large number of stars.

Scenario

View task SOTA rankings

Prompt example

Check it out at paperswithcode.com/sota:




Popular tasks ranking:


- Image Classification


- Object Detection


- Machine Translation


- Question Answering


- Text Generation




Each task displays:


- Current best methods and results


- Historical progress curve


- Corresponding papers and codes


- Dataset used

Output / what to expect

Understand the current best approach to a task,

Judge the gap between your own method and SOTA,

Find the baseline method that is most worth reproducing.

Tips

SOTA rankings are the fastest way to understand the current status of a certain research direction.

Dataset discovery

Papers With Code contains a large collection of standard datasets used in AI research.

Scenario

Find a suitable training data set

Prompt example

Search paperswithcode.com/datasets for:

1. Filter by task type (such as Image Classification)
2. View data set details:
   - Data volume (number of samples)
   - Annotation type
   - License (whether it can be used commercially)
   - Download method
3. View papers using this dataset
4. Understand the SOTA results on this dataset

Output / what to expect

Find a standard dataset suitable for the task,

Understand the size and quality of the data set,

Judge the difficulty of the data set through SOTA results.

Tips

Select widely used standard data sets to facilitate fair comparisons with other methods.

Compared with similar tools

Tool	Strength	Best for	Pricing
Papers With Code This tool	Paper code correspondence, SOTA tracking, completely free	AI researchers need to find code to implement and track SOTA	completely free
Hugging Face	Model and dataset hosting, the community is more active	Use pre-trained models directly	Free / Pro $9/month
arXiv	Latest papers, fastest published	Follow the latest research progress	completely free
Semantic Scholar	Citation Analysis, AI Summary	Literature discovery and understanding	completely free

Sources & references:

Papers With Code Official Website (2025-03)