Popular Papers

Select a paper to start implementing.

2012CVCNN

ImageNet Classification with Deep Convolutional Neural Networks

Alex Krizhevsky, Ilya Sutskever, Geoffrey E. Hinton

The AlexNet paper that revolutionized computer vision with deep CNNs. Implementing core CNN components from scratch: convolution, ReLU activation, max pooling, Local Response Normalization, and more.

1986NLPRNN

Recurrent Neural Networks

Various

Implementing core RNN components from scratch: RNN cells, forward/backward passes, BPTT, initialization, padding utilities, gradient clipping, embeddings, and full RNN blocks.

1995MLClassificationKernel Methods

Support Vector Machines

Cortes, Vapnik

Margin-based learning algorithm for classification, implementing the soft-margin SVM formulation and kernel tricks for non-linear decision boundaries.

2013CVVAE

Auto-Encoding Variational Bayes

Diederik P Kingma, Max Welling

Introduces the Variational Autoencoder (VAE), a deep generative model that learns a latent variable space using the reparameterization trick.

2018NLPTransformerPre-training

BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding

Devlin et al.

BERT is a bidirectional encoder representation from Transformers that uses masked language modeling and next sentence prediction for pre-training.

2022Computer VisionObject DetectionVision Transformer

OWL-ViT: Open-Vocabulary Object Detection using Vision Transformers

Minderer et al.

Open-vocabulary object detection using vision transformers with text prompts.

2017NLPTransformer

Attention Is All You Need

Vaswani et al.

The foundational Transformer architecture that introduced self-attention as a replacement for recurrence, enabling parallel processing and improved long-range dependency modeling.

2014Deep LearningGAN

Generative Adversarial Networks

Ian Goodfellow et al.

The original GAN paper introducing adversarial training between a generator and discriminator network for generative modeling.

2015CVCNNDeep Learning

Deep Residual Learning for Image Recognition

Kaiming He, Xiangyu Zhang, Shaoqing Ren, Jian Sun

Introduces residual connections that enable training of very deep networks by learning residual functions with reference to layer inputs.

2013NLPEmbeddings

Efficient Estimation of Word Representations in Vector Space

Tomas Mikolov et al.

Word2Vec introduces efficient methods for learning distributed word representations using Skip-gram and CBOW architectures.

2015Deep LearningOptimization

Batch Normalization: Accelerating Deep Network Training

Sergey Ioffe, Christian Szegedy

Batch Normalization normalizes layer inputs to accelerate training and allow higher learning rates, reducing sensitivity to initialization.

2013RLReinforcement LearningDeep Learning

Playing Atari with Deep Reinforcement Learning

Mnih et al.

DQN combines deep neural networks with Q-learning, introducing experience replay and target networks for stable training on Atari games.