Priyam Mazumdar

3.59K subscribers

76 videos

I am a Neuroscientist and Statistician that has become obsessed with AI and I hope to explore different architectures in a way that is accessible to anyone! ...

Latest Videos

The Annotated Flash Attention

Triton Layernorm Kernel A MyTorch Sidequest

Triton Embedding Kernel and Atomic Sum A MyTorch Sidequest

Triton Softmax Kernel A MyTorch Sidequest

REINFORCE with Baseline Variance Reduction via Advantage Estimation

Masked AutoEncoders (MAE) Implementation From Scratch Self-Supervised Image Pretraining!

Masked AutoEncoders (MAE) Paper Overview Self-Supervised Image Pretraining!

Triton Blocked Matrix Multiplication A MyTorch Sidequest!

Triton Grouped Matrix Multiplication (Almost CUDA Performance!) A MyTorch Sidequest

Triton Naive Matrix Multiplication A MyTorch Sidequest!

EnCodec From Scratch Lets Build a Neural Audio Codec!!

Triton Vector Addition Kernel A MyTorch Sidequest

Intro to Triton A MyTorch Sidequest!

Why do LLMs have to Think Exploring the importance of Reasoning

Can Speech Be Tokenized Like Text A Brief Exploration of Neural Codec Models

HifiGAN From Scratch Building a Neural Vocoder for Speech Synthesis

MyTorch Part 4 Lets Build a Baby GPT Model

Stable Diffusion Part 3 Training the Latent Diffusion Model

MyTorch Part 3 Writing the Base nn.Module

Stable Diffusion Part 2 Training the Variational AutoEncoder

Deriving the Policy Gradient Theorem and REINFORCE

MyTorch Part 2 Lets Build AutoGrad From Scratch

MyTorch Part 1 Running GPU Ops on Numpy

Dueling Deep-Q-Learning What's My Advantage

Stable Diffusion From Scratch Part 1 Perceptual Loss Functions (LPIPS + PatchGAN)

The Autogradless Transformer Training a GPT2 Model With Nothing but Numpy!

Double Deep-Q-Learning Solving the Overestimation Bias

Build your own Voice Generator w Tacotron 2 Text-To-Speech (TTS) From Scratch

Language Diffusion Models From Scratch Maybe Diffusion is All We Need

Let's Reproduce Wav2Vec2 Quantized Speech Pre-Training!