Wavenet Inspired Language Model
GitHubNLPDeep LearningPyTorchCausal ConvolutionsLanguage Modeling
TLDR
Character level language model using causal convolutions. Generates text by learning patterns from training data.
Detailed
Tech Stack:
PyTorch, Marimo notebook
Goal:
Build a character level autoregressive language model to generate company names.
What I did:
- •Implemented causal convolutions inspired by Google DeepMind's WaveNet paper
- •Trained on large dataset of company names
- •Deployed interactive model in Marimo notebook for hyperparameter tuning
What was achieved:
Model captures long range text dependencies. Architecture scales to word and sentence level.