Inspect the search settings
Locate search and evaluation parameters. Treat the evaluator and synthetic workload as part of the experiment, not as universal measures of code quality.
demo/AlphaEvolve-v0/config/alphaevolve.json ↗Learning & research
AlphaEvolve
Explore evolutionary search, candidate evaluation and an inspectable results viewer.
A CLOSER LOOK
Can a small code change improve an allocation heuristic?
For developers: follow candidate generation, program changes, evaluation and selection on synthetic agents and jobs.
Follow candidate evaluation
The CLI builds a population, evaluates candidates and records utility. Read the baseline construction and guardrails before trusting a winning edit.
demo/AlphaEvolve-v0/alphaevolve_runner.py ↗Read a concrete improvement
Expand best.diff and history. The recorded example changes the scoring weights and cost penalty; compare baseline utility and cost with the candidate on this workload.
demo/AlphaEvolve-v0/alphaevolve_summary.json ↗REAL REPOSITORY MATERIAL
Inspect. Understand. Reproduce.
Reading the exact source at revision 5b4cebb3.
This browser inspection does not execute the demo.
demo/AlphaEvolve-v0/config/alphaevolve.json
Select a walkthrough step to explore its source.
Full source text
{
"evolvable_functions": [
"alphaevolve.heuristics.score_match",
"alphaevolve.heuristics.price_job",
"alphaevolve.heuristics.rank_candidates",
"alphaevolve.heuristics.schedule_agents"
],
"prompt": {
"explicit_context": "Economic optimization of AGIJobs marketplace heuristics.",
"include_metrics": ["Utility", "GMV", "Cost", "Fairness"],
"stochastic_templates": {
"task_intro": [
"Improve the allocation logic for higher ROI.",
"Discover a breakthrough increase in Utility.",
"Engineer a resilient uplift in marketplace economics."
],
"use_probability": 0.35
}
},
"models": {
"fast_model": "gpt-fast",
"strong_model": "gpt-strong",
"strong_invoke_ratio": 0.15
},
"controller": {
"max_parallel_evaluations": 4,
"max_generations_per_run": 200,
"wallclock_time_limit_min": 60
},
"thermostat": {
"success_window": 12,
"low_success_threshold": 0.15,
"high_success_threshold": 0.6,
"min_temperature": 0.2,
"max_temperature": 0.9
},
"guardrails": {
"max_cost_pct_baseline": 1.1,
"min_utility_pct_baseline": 0.98,
"min_fairness": 0.3,
"rollback_on_latency_ms": 450
},
"baseline_metrics": {
"GMV": 2100.0,
"Cost": 880.0,
"Utility": 1220.0,
"Latency": 0.4,
"Fairness": 0.35,
"Acceptance": 0.65
}
}
SHA-256 51e1afb1da68d888492f8b3ebbc065cf1d7d6ec5af76c7e760a678da0232824f
FROM READING TO A REPRODUCIBLE RUN
Try the selected path.
Python, no demo package installRun from the repository root with Python 3.12. This selected entry point uses the Python standard library. Keep output in a separate directory so that you can compare runs.
Complete environment setup ↗python demo/AlphaEvolve-v0/run_demo.py run --seed 7 --output /tmp/alphaevolve-report.jsonWhat you should observe
A local evolutionary run writes /tmp/alphaevolve-report.json. Inspect baseline, selected candidate and evaluation assumptions; the historical summary is a separate recorded example.
The source inspector above reads bundled repository material. Local commands run separately on your computer. Recorded examples may contain historical timestamps, placeholders and simulated metrics.
MAKE IT YOUR OWN
One useful experiment.
Examine the winning diff, then ask which held-out workloads would expose overfitting. A better synthetic utility score does not establish safer general code.
THE SYSTEM, MADE VISIBLE
Architecture & relationships
View original Mermaid source
flowchart LR
Operators((Mission Owners)) --> demo_AlphaEvolve_v0[[Demo → AlphaEvolve v0]]
demo_AlphaEvolve_v0 --> Core[[AGI Jobs v0 (v2) Core Intelligence]]
Core --> Observability[[Unified CI / CD & Observability]]
Core --> Governance[[Owner Control Plane]]WHEN SOMETHING DOESN’T MATCH
Troubleshooting
Import or dependency error
Confirm the active virtual environment and the selected demo’s requirements. Run python -m pip check; do not install unrelated demo requirements over a working environment.
Unexpected result or missing file
Check the selected entry point, configuration and output argument. Keep the seed and implementation fixed before comparing outcomes.
TRACE THE CHECKS
Verification & next steps
6 tracked test source files are available in this directory. Inspect the tests and their environment before choosing a suite; file counts do not establish test results.
Browse the test sources
For live commissioning, consult the production readiness record.
EVERY VARIANT, PRESERVED
Complete document library
REPRODUCE & INSPECT
Registered commands
Run commands from the repository root after following this demo's guide. Network and owner actions require their documented setup.
No root-level launch command is associated with this source path. Follow the guide or source directory for its own entry point.