Construyendo un marco de evaluación integral de agentes de IA con métricas, informes y paneles visuales
class AdvancedAIEvaluator: def __init__(self, agent_func: Callable, config: Dict = None): self.agent_func = agent_func self.results = self.evaluation_history = defaultdict(list) self.benchmark_cache = {} self.config = { 'use_llm_judge': True, 'judge_model': 'gpt-4', 'embedding_model': 'sentence-transformers',…