How cterdam/gpt works
Project Type and Purpose
• This is a Python‐based CLI framework (“General Rapid Prototyping”) for running text‐generation “tasks” against various language‐model backends (OpenAI, Deepseek, etc.).
• Users invoke it through python -m src --task=<task_name>. Tasks live under src/task/... and orchestrate data loading, model calls, and logging.
File Structure Analysis
• Top level:
• src/
– __main__.py: entrypoint, argument parsing, task dispatcher
– core/: argument & config parser, environment setup, logging, utility functions, data_core
– lib/: reusable libraries
· lib/data/: DataBasis, WordBank abstractions
· lib/model/: ModelBasis classes, text‐generation parameters/results, API wrappers for OpenAI/Deepseek
– task/: concrete tasks (dry_run, demo, plus any new ones)
• tests/: unit tests for core and style checkers
• Makefile, pre-commit, requirements: build, test, style enforcement
Key README Insights
• Describes setup (conda, pip), usage (-m src --task), extension points (adding config fields, new tasks in src/task, updating run_task).
• Emphasizes black/isort pre-commit hooks.