Parameter                   Description                                          Default
------------------------    ------------------------------------------------     ----------------------
-p --problem_description    Description of your classification problem           (Required)

--model-type                ML library to use for the classifier model           tensorflow
                            tensorflow | torch | sklearn

--data-gen-model            OpenRouter model name for bulk data generation       openai/gtp-4o-mini

--config-llm-model          OpenRouter model for config, refinement,             x-ai/grok-3-beta
                            edge cases, analysis (should be powerful,
                            e.g., GPT-4, Claude Opus).",

--refinement-cycles         Number of prompt refinement cycles (0 to disable).   1

--generate-edge-cases       Generate challenging edge case data (true/false)     true

--batch-size                Number of parallel LLM API requests for bulk data
                            generation batches.                                  10

--lang                      Primary language for the generated dataset           english
                            (e.g., english, german, spanish, etc.).

Parameter                   Description
------------------------    ----------------------------------------------------------------------------
--config-path               Continue from a previous configuration file. Can be used for additional data
                            generation or retraining with user data or another library.

--skip-data-gen             Skip data generation and only train the model from existing data. Works only
                            if training_data.csv is present in the model directory and --config-path is
                            specified.

--skip-model-training       Skip model training and only generate data

                    
.[model_prefix]/
    ├── api_requests/                  # API requests used for training data generation
    ├── generation_config.json         # Configuration and analysis
    ├── edge_case_data.csv             # Sophisticated data used for model evaluation by LLM
    ├── model.onnx                     # ONNX model file
    ├── scaler.json                    # StandardScaler parameters
    ├── tokenizer.json                 # Tokenizer configuration (for softmax models)
    ├── training_data.csv              # Training data used for model training
    └── vocab.json                     # TF-IDF vocabulary (for sigmoid models)
    + ... native model files .h5, .keras