pytorch architecture assistant-PyTorch Code Generation

Example

Given a dataset of satellite images for land cover classification, the assistant suggests a CNN architecture with specifics on layer types, activation functions, and optimization strategies.

Scenario

A user is working on a project for classifying types of land cover in satellite images but is unsure about the optimal neural network architecture to use. The assistant analyzes the task and dataset characteristics, then recommends an architecture tailored to spatial feature extraction and classification.

Example

For a regression task predicting house prices from a set of features, the assistant generates a multi-layer perceptron (MLP) PyTorch code template, including data loading, model definition, training, and validation loops.

Scenario

A real estate analyst wants to develop a model to predict house prices based on features like square footage, location, and number of bedrooms. The assistant provides a ready-to-use PyTorch code template that the analyst can customize and run with their dataset.

Example

Advises on implementing early stopping and model checkpointing to improve training efficiency and model performance for a text classification task.

Scenario

An NLP practitioner is experiencing overfitting and long training times with their text classification model. The assistant suggests implementing early stopping to halt training when validation loss stops improving, and model checkpointing to save the model at its best performance state, thereby optimizing the training process and results.