## What You Will Learn
In this tutorial, you will submit your first machine learning model implementation to the Sotabench resource. You will:
- 🔎 Find an implementation of a modern image classification model
- 📊 Configure a
benchmark.pyfile to evaluate the model - 🎉 Submit the repository to sotabench and obtain state-of-the-art results!
Sound fun? Let's get started!
To find model implementations, we use Papers With Code, which has over 10,000 implementations and also ranks them according to their performance on popular research benchmarks.
Search for Image Classification:
This will bring you to the Task Page, where you can see various different benchmarks for evaluating the performance of models on image classification:
Let's click on ImageNet since this is now the most famous image classification benchmark:
This is the Evaluation Page. From the graph, we can see the historical progress on the task, and how models have got better over time. Below we can see the leaderboard of the best methods, ranked by performance.
Let's click on the top paper Exploring the Limits of Weakly Supervised Pretraining. This brings us to the paper page where we can see publicly available implementations:
There is an implementation in PyTorch, which we can evaluate using sotabench. Head to the repository: