experiments/chunk_100_over_25.yml

data_dir: 'data'
chunk_method: {
    'method': 'token-overlap',
    'args': {
        'chunk_size': 100,
        'chunk_overlap': 25
    }
}

vectorstore_model: {
    'model_source': 'openai',
    'model_name': 'text-embedding-3-large',
    'vector_size': 3072
}

#chunk_method: {
#    'method': 'semantic',
#    'args': {
#        'model_source': 'openai',
#        'model_name': 'text-embedding-3-large',
#        'breakpoint_type': 'percentile'
#    }
#}

# Metrics: faithfulness, answer_relevancy, context_precision, context_recall
metrics: ['faithfulness', 'answer_relevancy', 'context_precision', 'context_recall']
ls_project: 'policy-rag'
ls_dataset_name: "policy-golden-1000-over-100"
ls_experiment_name: 'policy-chunk-100-over-25'