MAT-HPO Library

from MAT_HPO_LIB import MAT_HPO_Optimizer, BaseEnvironment, HyperparameterSpace
from MAT_HPO_LIB.utils import DefaultConfigs

#  1. Define your optimization environment
class MyEnvironment(BaseEnvironment):
    def step(self, hyperparams):
        """ Core optimization logic"""
        f1, auc, gmean = train_and_evaluate(hyperparams)
        done = check_stopping_condition()
        return f1, auc, gmean, done

#  2. Define hyperparameter space with agent assignment
space = HyperparameterSpace()
space.add_continuous('class_weight_0', 0.5, 3.0, agent=0)  # Problem-specific
space.add_discrete('hidden_size', [64, 128, 256, 512], agent=1)    # Architecture
space.add_continuous('learning_rate', 1e-5, 1e-2, agent=2)             # Training
space.add_discrete('batch_size', [16, 32, 64, 128], agent=2)    # Training

#  3. Run optimization
config = DefaultConfigs.standard()  # 100 optimization steps
optimizer = MAT_HPO_Optimizer(MyEnvironment(), space, config)
results = optimizer.optimize()

#  View results
print(" Best hyperparameters:", results['best_hyperparameters'])

from MAT_HPO_LIB import FullControlHPO

#  Production-grade LLM enhancement (exact usage from SPNV2)
optimizer = FullControlHPO(
    task_type="time_series_classification",
    max_trials=30,

    # LLM Configuration (from SPNV2)
    llm_enabled=True,
    llm_model="llama3.2:3b",
    llm_base_url="http://localhost:11434",
    mixing_strategy="adaptive",              # or "fixed_alpha"
    alpha=0.3,                              # LLM/RL mixing ratio
    slope_threshold=0.01,                   # For adaptive strategy
    llm_cooldown_episodes=6,              # ICBEB: 6, PTBXL: 4
    min_episodes_before_llm=6,          # ICBEB: 6, PTBXL: 4

    # Advanced RL Parameters (dataset-specific)
    replay_buffer_size=150,              # ICBEB: 150, PTBXL: 200
    batch_size=12,                        # ICBEB: 12, PTBXL: 16
    actor_lr=0.0008,                       # ICBEB: 0.0008, PTBXL: 0.001
    critic_lr=0.0015,                      # ICBEB: 0.0015, PTBXL: 0.002
    
    # System Configuration
    seed=42,
    device="cuda",                        # Auto-detect CUDA
    verbose=True,
    output_dir="./mat_hpo_logs",
    dataset_name="ICBEB_Fold1",
    task_description="ECG classification on ICBEB dataset with 9 classes"
)

#  Optimize with custom environment (SPNV2 pattern)
results = optimizer.optimize(
    X_train, y_train, X_val, y_val,
    X_test, y_test,
    custom_environment=your_environment,
    custom_space=your_hyperparameter_space
)

print(f" Best F1: {results['best_performance']['f1']:.4f}")

#  Clone the repository
git clone https://github.com/VM230705/MAT-HPO-Library.git
cd MAT-HPO-Library

#  Install dependencies
pip install torch numpy scikit-learn

# ✅ Test installation
python test_working_examples.py

from MAT_HPO_LIB.utils import DefaultConfigs

#  Quick test (10 steps) - Perfect for debugging
config = DefaultConfigs.quick_test()

#  Standard optimization (100 steps) - Recommended for most use cases
config = DefaultConfigs.standard()

#  CPU-only mode - For environments without GPU
config = DefaultConfigs.cpu_only()