How alexengelhardt/startup-name-generator works
SNG is a Python library for learning character patterns from a word corpus and generating name suggestions. The principal workflow is to configure a Generator, supply words or load a bundled wordlist, train its recurrent model, and simulate names. The Generator also saves and reloads trained state. The graph separates corpus loading, model work, and configuration, but the sampled implementation places training, generation, and persistence in one class; unsampled details are not expanded.
Main entry points: Generator.py Generator · __init__.py Package API · builtin_wordlists.py Wordlist Loader · Config.py Generator Config
Components 8
Public API
- Package API
sng/__init__.py - Generator Config
sng/Config.py
Corpus Inputs
- Wordlist Loader
sng/builtin_wordlists.py - Bundled Wordlists
sng/wordlists
Generation Engine
- Generator
sng/Generator.py - Temperature Scaling
sng/helpers.py
Other components
- Library User
- Keras Runtime
Connections 10
- Library User → Package API: uses
README.rst - Package API → Generator: exports
sng/__init__.py - Package API → Generator Config: exports
sng/__init__.py - Package API → Wordlist Loader: exports
sng/__init__.py - Library User → Wordlist Loader: chooses corpus
sng/builtin_wordlists.py - Wordlist Loader → Bundled Wordlists: reads
sng/builtin_wordlists.py - Library User → Generator: supplies words
sng/Generator.py - Generator → Generator Config: uses settings
sng/Generator.py - Generator → Temperature Scaling: scales probabilities
sng/Generator.py - Generator → Keras Runtime: trains and predicts
sng/Generator.py