Seyyed_arc/docs/ai-agent/evals/phase-8/README.md

Phase 8 evaluation outputs

• tool-adaptive-k/latest.json — Fixed K vs Adaptive
• tool-hybrid/latest.json — weight calibration
• tool-scale/latest.json — 180…10k metadata bench
• latest.json — combined

cd hesabixAPI && .venv/bin/python -m app.services.ai.ai_tool_phase8_eval
cd hesabixAPI && .venv/bin/python -m app.services.ai.ai_tool_scale_bench