nniback / evals /__init__.py
Kyrlos11's picture
Upload 842 files
22087d0 verified
Raw History Blame Contribute Delete
752 Bytes
"""
Braintrust Evaluation Framework for Kortix Agent.
This module provides tools for evaluating agent performance using Braintrust.
It integrates with the existing agent system to:
1. Run test cases through the agent
2. Score outputs using various metrics
3. Track experiments in Braintrust dashboard
Usage:
# Run evals from command line:
braintrust eval evals/agent_eval.py
# Or programmatically:
from evals import run_agent_eval
await run_agent_eval()
"""
from evals.runner import AgentEvalRunner
from evals.scorers import (
TaskCompletionScorer,
ToolUsageScorer,
ResponseQualityScorer,
)
__all__ = [
"AgentEvalRunner",
"TaskCompletionScorer",
"ToolUsageScorer",
"ResponseQualityScorer",
]