add fine-tuning an embedding model for domain knowledge

add confusion matrix and decision boundary

Kaggle Notebook | Implement GPT2-124M and Load HF GPT2 | Version 20

Add speech recognition via Whisper

Add visualisation for a simple multiple perceptron

add back propagation and training functions

Init commit for multilayer perceptron with numpy

Kaggle Notebook | Tokenize the Data for Pre-training Neural Network |…

add DistributedDataParallel training

Add the gradient accumulation minimal example

hidden code comment

add acknowledgement

add temporal fusion transformer with Pytorch Forecasting

show the code cell in the notebook

reduce the micro batch number to fit lower memory GPU

Kaggle Notebook | Implement GPT2-124M and Load HF GPT2 | Version 15

add weight decay, only for 2D params, and add fused AdamW

Add learning rate scheduler

add torch compile and bfloat16

Support gpt2 initialisation with std=0.2 parameter

fix the issue weight tie the embedding and unembedding matrix

Add data loader

Add 50 training loop with a little batch

calculate the loss function: cross entropy loss

Do sampling from the customize model

add forward() function of GPT2 nn.Module

Kaggle Notebook | Implement GPT2-124M and Load HF GPT2 | Version 4

Kaggle Notebook | Implement GPT2-124M | Version 2

Kaggle Notebook | Build a Tokenizer | Version 3

Kaggle Notebook | Anomaly Detection on Bank Card Fraud | Version 2