labml.ai Deep Learning Paper Implementations

This is a collection of simple PyTorch implementations of neural networks and related algorithms. These implementations are documented with explanations,

The website renders these as side-by-side formatted notes. We believe these would help you understand these algorithms better.

We are actively maintaining this repo and adding new implementations almost weekly. for updates.

Paper Implementations

✨ Transformers

• Multi-headed attention
• Transformer building blocks
• Transformer XL
  • Relative multi-headed attention
• Rotary Positional Embeddings
• Attention with Linear Biases (ALiBi)
• RETRO
• Compressive Transformer
• GPT Architecture
• GLU Variants
• kNN-LM: Generalization through Memorization
• Feedback Transformer
• Switch Transformer
• Fast Weights Transformer
• FNet
• Attention Free Transformer
• Masked Language Model
• MLP-Mixer: An all-MLP Architecture for Vision
• Pay Attention to MLPs (gMLP)
• Vision Transformer (ViT)
• Primer EZ
• Hourglass

✨ Eleuther GPT-NeoX

• Generate on a 48GB GPU
• Finetune on two 48GB GPUs
• LLM.int8()

✨ Diffusion models

• Denoising Diffusion Probabilistic Models (DDPM)
• Denoising Diffusion Implicit Models (DDIM)
• Latent Diffusion Models
• Stable Diffusion

✨ Generative Adversarial Networks

• Original GAN
• GAN with deep convolutional network
• Cycle GAN
• Wasserstein GAN
• Wasserstein GAN with Gradient Penalty
• StyleGAN 2

✨ Recurrent Highway Networks

✨ LSTM

✨ HyperNetworks - HyperLSTM

✨ ResNet

✨ ConvMixer

✨ Capsule Networks

✨ U-Net

✨ Sketch RNN

✨ Graph Neural Networks

• Graph Attention Networks (GAT)
• Graph Attention Networks v2 (GATv2)

✨ Counterfactual Regret Minimization (CFR)

Solving games with incomplete information such as poker with CFR.

• Kuhn Poker

✨ Reinforcement Learning

• Proximal Policy Optimization with Generalized Advantage Estimation
• Deep Q Networks with with Dueling Network, Prioritized Replay and Double Q Network.

✨ Optimizers

• Adam
• AMSGrad
• Adam Optimizer with warmup
• Noam Optimizer
• Rectified Adam Optimizer
• AdaBelief Optimizer

✨ Normalization Layers

• Batch Normalization
• Layer Normalization
• Instance Normalization
• Group Normalization
• Weight Standardization
• Batch-Channel Normalization
• DeepNorm

✨ Distillation

✨ Adaptive Computation

• PonderNet

✨ Uncertainty

• Evidential Deep Learning to Quantify Classification Uncertainty

✨ Activations

• Fuzzy Tiling Activations

✨ Langauge Model Sampling Techniques

• Greedy Sampling
• Temperature Sampling
• Top-k Sampling
• Nucleus Sampling

✨ Scalable Training/Inference

• Zero3 memory optimizations

Highlighted Research Paper PDFs

• FlashAttention: Fast and Memory-Efficient Exact Attention with IO-Awareness
• Autoregressive Search Engines: Generating Substrings as Document Identifiers
• Training Compute-Optimal Large Language Models
• ZeRO: Memory Optimizations Toward Training Trillion Parameter Models
• PaLM: Scaling Language Modeling with Pathways
• Hierarchical Text-Conditional Image Generation with CLIP Latents
• STaR: Self-Taught Reasoner Bootstrapping Reasoning With Reasoning
• Improving language models by retrieving from trillions of tokens
• NeRF: Representing Scenes as Neural Radiance Fields for View Synthesis
• Attention Is All You Need
• Denoising Diffusion Probabilistic Models
• Primer: Searching for Efficient Transformers for Language Modeling
• On First-Order Meta-Learning Algorithms
• Learning Transferable Visual Models From Natural Language Supervision
• The Sensory Neuron as a Transformer: Permutation-Invariant Neural Networks for Reinforcement Learning
• Meta-Gradient Reinforcement Learning
• ETA Prediction with Graph Neural Networks in Google Maps
• PonderNet: Learning to Ponder
• Mastering Atari, Go, Chess and Shogi by Planning with a Learned Model
• GANs N’ Roses: Stable, Controllable, Diverse Image to Image Translation (works for videos too!)
• An Image is Worth 16X16 Word: Transformers for Image Recognition at Scale
• Deep Residual Learning for Image Recognition
• Distilling the Knowledge in a Neural Network

Installation

pip install labml-nn

Citing

If you use this for academic research, please cite it using the following BibTeX entry.

@misc{labml,
 author = {Varuna Jayasiri, Nipun Wijerathne},
 title = {labml.ai Annotated Paper Implementations},
 year = {2020},
 url = {https://nn.labml.ai/},
}

Other Projects

🚀 Trending Research Papers

This shows the most popular research papers on social media. It also aggregates links to useful resources like paper explanations videos and discussions.

🧪 labml.ai/labml

This is a library that let's you monitor deep learning model training and hardware usage from your mobile phone. It also comes with a bunch of other tools to help write deep learning code efficiently.