Machine Learning for Alignment Bootcamp (MLAB)

| Day | Title | Folder | | ---- | ------------------------------------------------------------------------------------------- | ----------------------- | | Prep | Tensor operations | tensor_operations | | W1D1 | Triangle rendering | | | W1D2 | Implementing ResNet34 | | | W1D3 | MiniTorch | | | W1D4 | Training (= learning = optimization) | basic_training | | W1D5 | ConvNet interpretability | | | W1D6 | CUDA programming | | | W2D1 | Implementing BERT | bert_model | | W2D2 | BERT training and inference | bert_training | | W2D3 | Implementing GPT-2 | | | W2D4 | Remote hyperparameter search and tokenizers | hyperparameter_search | | W2D5 | Data parallel training | data_parallel_training | | W3D1 | Model parallel training | model_parallel_training | | W3D2 | Transformer interpretability | | | W3D3 | Deep Q-Learning (DQN) for Atari | dqn_atari | | W3D4 | RL for Human Feedback | rl_human_feedback | | W3D5 | Transformer interpretability | |