agent-hitch 0.2.6 → 0.2.8
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
- package/README.md +30 -3
- package/README.zh-CN.md +14 -2
- package/dist/bin/hitch.js +25 -3
- package/dist/bin/hitch.js.map +1 -1
- package/dist/scripts/canary-benchmark.js +54 -0
- package/dist/scripts/canary-benchmark.js.map +1 -0
- package/dist/scripts/canary-buildkit-secrets.js +166 -0
- package/dist/scripts/canary-buildkit-secrets.js.map +1 -0
- package/dist/scripts/canary-collision-domains.js +77 -0
- package/dist/scripts/canary-collision-domains.js.map +1 -0
- package/dist/scripts/canary-docker-image-observation.js +181 -0
- package/dist/scripts/canary-docker-image-observation.js.map +1 -0
- package/dist/scripts/canary-eval-scheduler-throughput.js +47 -0
- package/dist/scripts/canary-eval-scheduler-throughput.js.map +1 -0
- package/dist/scripts/canary-gpu-hardware.js +95 -0
- package/dist/scripts/canary-gpu-hardware.js.map +1 -0
- package/dist/scripts/canary-harbor-load.js +365 -0
- package/dist/scripts/canary-harbor-load.js.map +1 -0
- package/dist/scripts/canary-packaged-harness.js +75 -0
- package/dist/scripts/canary-packaged-harness.js.map +1 -0
- package/dist/scripts/canary-resource-load.js +188 -0
- package/dist/scripts/canary-resource-load.js.map +1 -0
- package/dist/scripts/check-architecture.js +10 -5
- package/dist/scripts/check-architecture.js.map +1 -1
- package/dist/scripts/docker-reaper-crash-child.js +21 -0
- package/dist/scripts/docker-reaper-crash-child.js.map +1 -0
- package/dist/src/adapters/catalog.js +3 -1
- package/dist/src/adapters/catalog.js.map +1 -1
- package/dist/src/adapters/discovery.js +1 -1
- package/dist/src/adapters/discovery.js.map +1 -1
- package/dist/src/adapters/providers/claude.js +6 -0
- package/dist/src/adapters/providers/claude.js.map +1 -1
- package/dist/src/adapters/providers/codex-auth.js +29 -0
- package/dist/src/adapters/providers/codex-auth.js.map +1 -0
- package/dist/src/adapters/providers/codex.js +29 -6
- package/dist/src/adapters/providers/codex.js.map +1 -1
- package/dist/src/adapters/providers/deepseek.js +6 -0
- package/dist/src/adapters/providers/deepseek.js.map +1 -1
- package/dist/src/adapters/providers/model-call.js +26 -0
- package/dist/src/adapters/providers/model-call.js.map +1 -0
- package/dist/src/adapters/providers/opencode.js +6 -0
- package/dist/src/adapters/providers/opencode.js.map +1 -1
- package/dist/src/adapters/providers/pi.js +6 -0
- package/dist/src/adapters/providers/pi.js.map +1 -1
- package/dist/src/artifacts/index.js +1 -0
- package/dist/src/artifacts/index.js.map +1 -1
- package/dist/src/artifacts/preparer.js +7 -2
- package/dist/src/artifacts/preparer.js.map +1 -1
- package/dist/src/backends/harbor/agent-budget.js +31 -0
- package/dist/src/backends/harbor/agent-budget.js.map +1 -0
- package/dist/src/backends/harbor/backend.js +59 -107
- package/dist/src/backends/harbor/backend.js.map +1 -1
- package/dist/src/backends/harbor/environment-config.js +92 -0
- package/dist/src/backends/harbor/environment-config.js.map +1 -0
- package/dist/src/backends/harbor/index.js +6 -1
- package/dist/src/backends/harbor/index.js.map +1 -1
- package/dist/src/backends/harbor/model-proxy-config.js +23 -0
- package/dist/src/backends/harbor/model-proxy-config.js.map +1 -0
- package/dist/src/backends/harbor/process.js +177 -0
- package/dist/src/backends/harbor/process.js.map +1 -0
- package/dist/src/backends/harbor/regrade.js +77 -0
- package/dist/src/backends/harbor/regrade.js.map +1 -0
- package/dist/src/backends/harbor/runtime-toolchain.js +30 -0
- package/dist/src/backends/harbor/runtime-toolchain.js.map +1 -0
- package/dist/src/backends/harbor/supervisor.js +74 -0
- package/dist/src/backends/harbor/supervisor.js.map +1 -0
- package/dist/src/backends/harbor/task-resources.js +206 -0
- package/dist/src/backends/harbor/task-resources.js.map +1 -0
- package/dist/src/backends/harbor/tools.js +6 -29
- package/dist/src/backends/harbor/tools.js.map +1 -1
- package/dist/src/backends/index.js +5 -1
- package/dist/src/backends/index.js.map +1 -1
- package/dist/src/benchmarks/index.js +3 -0
- package/dist/src/benchmarks/index.js.map +1 -0
- package/dist/src/benchmarks/loader.js +271 -0
- package/dist/src/benchmarks/loader.js.map +1 -0
- package/dist/src/benchmarks/metrics.js +16 -0
- package/dist/src/benchmarks/metrics.js.map +1 -0
- package/dist/src/benchmarks/toml.js +13 -0
- package/dist/src/benchmarks/toml.js.map +1 -0
- package/dist/src/benchmarks/validation.js +242 -0
- package/dist/src/benchmarks/validation.js.map +1 -0
- package/dist/src/cli/arguments.js +6 -4
- package/dist/src/cli/arguments.js.map +1 -1
- package/dist/src/cli/commands/benchmark.js +22 -0
- package/dist/src/cli/commands/benchmark.js.map +1 -0
- package/dist/src/cli/commands/capabilities.js +18 -0
- package/dist/src/cli/commands/capabilities.js.map +1 -0
- package/dist/src/cli/commands/daemon.js +135 -8
- package/dist/src/cli/commands/daemon.js.map +1 -1
- package/dist/src/cli/commands/eval.js +228 -6
- package/dist/src/cli/commands/eval.js.map +1 -1
- package/dist/src/cli/commands/images.js +40 -0
- package/dist/src/cli/commands/images.js.map +1 -0
- package/dist/src/cli/commands/run.js +42 -17
- package/dist/src/cli/commands/run.js.map +1 -1
- package/dist/src/cli/commands/runs.js +25 -3
- package/dist/src/cli/commands/runs.js.map +1 -1
- package/dist/src/cli/commands/trajectory.js +96 -5
- package/dist/src/cli/commands/trajectory.js.map +1 -1
- package/dist/src/cli/commands/verifier.js +36 -0
- package/dist/src/cli/commands/verifier.js.map +1 -0
- package/dist/src/cli/commands/worker.js +79 -0
- package/dist/src/cli/commands/worker.js.map +1 -0
- package/dist/src/cli/main.js +10 -0
- package/dist/src/cli/main.js.map +1 -1
- package/dist/src/cli/output.js +84 -4
- package/dist/src/cli/output.js.map +1 -1
- package/dist/src/control-plane/build-admission.js +69 -0
- package/dist/src/control-plane/build-admission.js.map +1 -0
- package/dist/src/control-plane/build-records.js +6 -0
- package/dist/src/control-plane/build-records.js.map +1 -0
- package/dist/src/control-plane/collisions.js +40 -0
- package/dist/src/control-plane/collisions.js.map +1 -0
- package/dist/src/control-plane/eval-control-state.js +22 -0
- package/dist/src/control-plane/eval-control-state.js.map +1 -0
- package/dist/src/control-plane/eval-control-work.js +44 -0
- package/dist/src/control-plane/eval-control-work.js.map +1 -0
- package/dist/src/control-plane/eval-image-resolution.js +52 -0
- package/dist/src/control-plane/eval-image-resolution.js.map +1 -0
- package/dist/src/control-plane/eval-image-services.js +30 -0
- package/dist/src/control-plane/eval-image-services.js.map +1 -0
- package/dist/src/control-plane/eval-records.js +252 -0
- package/dist/src/control-plane/eval-records.js.map +1 -0
- package/dist/src/control-plane/eval-recovery.js +181 -0
- package/dist/src/control-plane/eval-recovery.js.map +1 -0
- package/dist/src/control-plane/eval-scheduler.js +450 -0
- package/dist/src/control-plane/eval-scheduler.js.map +1 -0
- package/dist/src/control-plane/index.js +20 -0
- package/dist/src/control-plane/index.js.map +1 -0
- package/dist/src/control-plane/local-worker.js +49 -0
- package/dist/src/control-plane/local-worker.js.map +1 -0
- package/dist/src/control-plane/model-capture-planning.js +12 -0
- package/dist/src/control-plane/model-capture-planning.js.map +1 -0
- package/dist/src/control-plane/remote-result-transport.js +175 -0
- package/dist/src/control-plane/remote-result-transport.js.map +1 -0
- package/dist/src/control-plane/remote-work-coordinator.js +333 -0
- package/dist/src/control-plane/remote-work-coordinator.js.map +1 -0
- package/dist/src/control-plane/remote-work-inputs.js +266 -0
- package/dist/src/control-plane/remote-work-inputs.js.map +1 -0
- package/dist/src/control-plane/remote-work-item.js +62 -0
- package/dist/src/control-plane/remote-work-item.js.map +1 -0
- package/dist/src/control-plane/remote-work-recovery.js +220 -0
- package/dist/src/control-plane/remote-work-recovery.js.map +1 -0
- package/dist/src/control-plane/remote-worker-artifacts.js +135 -0
- package/dist/src/control-plane/remote-worker-artifacts.js.map +1 -0
- package/dist/src/control-plane/remote-worker-client.js +202 -0
- package/dist/src/control-plane/remote-worker-client.js.map +1 -0
- package/dist/src/control-plane/remote-worker-configuration.js +34 -0
- package/dist/src/control-plane/remote-worker-configuration.js.map +1 -0
- package/dist/src/control-plane/remote-worker-credentials.js +58 -0
- package/dist/src/control-plane/remote-worker-credentials.js.map +1 -0
- package/dist/src/control-plane/remote-worker-protocol.js +489 -0
- package/dist/src/control-plane/remote-worker-protocol.js.map +1 -0
- package/dist/src/control-plane/remote-worker-runner.js +214 -0
- package/dist/src/control-plane/remote-worker-runner.js.map +1 -0
- package/dist/src/control-plane/remote-workers.js +377 -0
- package/dist/src/control-plane/remote-workers.js.map +1 -0
- package/dist/src/control-plane/rerun-scheduler.js +448 -0
- package/dist/src/control-plane/rerun-scheduler.js.map +1 -0
- package/dist/src/control-plane/rerun-submission.js +62 -0
- package/dist/src/control-plane/rerun-submission.js.map +1 -0
- package/dist/src/control-plane/resources.js +182 -0
- package/dist/src/control-plane/resources.js.map +1 -0
- package/dist/src/control-plane/scheduler-eval-entry.js +32 -0
- package/dist/src/control-plane/scheduler-eval-entry.js.map +1 -0
- package/dist/src/control-plane/synthetic-result.js +21 -0
- package/dist/src/control-plane/synthetic-result.js.map +1 -0
- package/dist/src/control-plane/work-admission.js +23 -0
- package/dist/src/control-plane/work-admission.js.map +1 -0
- package/dist/src/control-plane/work-dispatcher.js +237 -0
- package/dist/src/control-plane/work-dispatcher.js.map +1 -0
- package/dist/src/controller-runtime/hash.js +19 -3
- package/dist/src/controller-runtime/hash.js.map +1 -1
- package/dist/src/controller-runtime/index.js +1 -1
- package/dist/src/controller-runtime/index.js.map +1 -1
- package/dist/src/controller-runtime/store.js +15 -6
- package/dist/src/controller-runtime/store.js.map +1 -1
- package/dist/src/daemon/auth.js +5 -0
- package/dist/src/daemon/auth.js.map +1 -1
- package/dist/src/daemon/client.js.map +1 -1
- package/dist/src/daemon/event-stream.js +46 -0
- package/dist/src/daemon/event-stream.js.map +1 -0
- package/dist/src/daemon/health.js +66 -0
- package/dist/src/daemon/health.js.map +1 -0
- package/dist/src/daemon/launcher.js +28 -1
- package/dist/src/daemon/launcher.js.map +1 -1
- package/dist/src/daemon/logging.js +23 -0
- package/dist/src/daemon/logging.js.map +1 -0
- package/dist/src/daemon/scheduler.js +26 -5
- package/dist/src/daemon/scheduler.js.map +1 -1
- package/dist/src/daemon/server.js +255 -98
- package/dist/src/daemon/server.js.map +1 -1
- package/dist/src/daemon/telemetry.js +163 -0
- package/dist/src/daemon/telemetry.js.map +1 -0
- package/dist/src/daemon/worker-routes.js +219 -0
- package/dist/src/daemon/worker-routes.js.map +1 -0
- package/dist/src/domain/benchmarks.js +2 -0
- package/dist/src/domain/benchmarks.js.map +1 -0
- package/dist/src/domain/bundles.js +2 -0
- package/dist/src/domain/bundles.js.map +1 -0
- package/dist/src/domain/control-plane.js +2 -0
- package/dist/src/domain/control-plane.js.map +1 -0
- package/dist/src/domain/execution-evidence.js +2 -0
- package/dist/src/domain/execution-evidence.js.map +1 -0
- package/dist/src/domain/execution-plan.js +2 -0
- package/dist/src/domain/execution-plan.js.map +1 -0
- package/dist/src/domain/images.js +2 -0
- package/dist/src/domain/images.js.map +1 -0
- package/dist/src/domain/index.js +13 -0
- package/dist/src/domain/index.js.map +1 -1
- package/dist/src/domain/interactions.js +2 -0
- package/dist/src/domain/interactions.js.map +1 -0
- package/dist/src/domain/providers.js +2 -0
- package/dist/src/domain/providers.js.map +1 -0
- package/dist/src/domain/resources.js +2 -0
- package/dist/src/domain/resources.js.map +1 -0
- package/dist/src/domain/runs.js.map +1 -1
- package/dist/src/domain/validation.js +11 -4
- package/dist/src/domain/validation.js.map +1 -1
- package/dist/src/domain/verifier-evidence-validation.js +214 -0
- package/dist/src/domain/verifier-evidence-validation.js.map +1 -0
- package/dist/src/domain/verifier-evidence.js +3 -0
- package/dist/src/domain/verifier-evidence.js.map +1 -0
- package/dist/src/domain/verifier-score-contract.js +232 -0
- package/dist/src/domain/verifier-score-contract.js.map +1 -0
- package/dist/src/domain/workers.js +2 -0
- package/dist/src/domain/workers.js.map +1 -0
- package/dist/src/evals/benchmark-adapter-manifest.js +219 -0
- package/dist/src/evals/benchmark-adapter-manifest.js.map +1 -0
- package/dist/src/evals/benchmark-candidate.js +36 -0
- package/dist/src/evals/benchmark-candidate.js.map +1 -0
- package/dist/src/evals/benchmark-run.js +166 -0
- package/dist/src/evals/benchmark-run.js.map +1 -0
- package/dist/src/evals/collect-only-rerun.js +183 -0
- package/dist/src/evals/collect-only-rerun.js.map +1 -0
- package/dist/src/evals/directory.js +30 -0
- package/dist/src/evals/directory.js.map +1 -0
- package/dist/src/evals/docker-engine-stats.js +89 -0
- package/dist/src/evals/docker-engine-stats.js.map +1 -0
- package/dist/src/evals/docker-ownership.js +51 -0
- package/dist/src/evals/docker-ownership.js.map +1 -0
- package/dist/src/evals/docker-reaper.js +144 -0
- package/dist/src/evals/docker-reaper.js.map +1 -0
- package/dist/src/evals/docker-resource-observer.js +207 -0
- package/dist/src/evals/docker-resource-observer.js.map +1 -0
- package/dist/src/evals/duration-estimator.js +126 -0
- package/dist/src/evals/duration-estimator.js.map +1 -0
- package/dist/src/evals/environment-image-planning.js +164 -0
- package/dist/src/evals/environment-image-planning.js.map +1 -0
- package/dist/src/evals/eval-artifact-planning.js +56 -0
- package/dist/src/evals/eval-artifact-planning.js.map +1 -0
- package/dist/src/evals/eval-finalization.js +10 -0
- package/dist/src/evals/eval-finalization.js.map +1 -0
- package/dist/src/evals/eval-lifecycle-events.js +29 -0
- package/dist/src/evals/eval-lifecycle-events.js.map +1 -0
- package/dist/src/evals/eval-logical-plan.js +51 -0
- package/dist/src/evals/eval-logical-plan.js.map +1 -0
- package/dist/src/evals/eval-result-builder.js +81 -0
- package/dist/src/evals/eval-result-builder.js.map +1 -0
- package/dist/src/evals/events.js +53 -5
- package/dist/src/evals/events.js.map +1 -1
- package/dist/src/evals/evolution-baseline.js +191 -0
- package/dist/src/evals/evolution-baseline.js.map +1 -0
- package/dist/src/evals/execution-evidence.js +116 -0
- package/dist/src/evals/execution-evidence.js.map +1 -0
- package/dist/src/evals/execution-leases.js +342 -0
- package/dist/src/evals/execution-leases.js.map +1 -0
- package/dist/src/evals/execution-plan-artifacts.js +72 -0
- package/dist/src/evals/execution-plan-artifacts.js.map +1 -0
- package/dist/src/evals/execution-plan-images.js +69 -0
- package/dist/src/evals/execution-plan-images.js.map +1 -0
- package/dist/src/evals/execution-plan-resources.js +174 -0
- package/dist/src/evals/execution-plan-resources.js.map +1 -0
- package/dist/src/evals/execution-plan.js +451 -0
- package/dist/src/evals/execution-plan.js.map +1 -0
- package/dist/src/evals/failure-classifier.js +61 -0
- package/dist/src/evals/failure-classifier.js.map +1 -0
- package/dist/src/evals/harbor-artifact-builder-image.js +99 -0
- package/dist/src/evals/harbor-artifact-builder-image.js.map +1 -0
- package/dist/src/evals/harbor-artifact-builder.js +358 -0
- package/dist/src/evals/harbor-artifact-builder.js.map +1 -0
- package/dist/src/evals/harbor-bridge-error.js +58 -3
- package/dist/src/evals/harbor-bridge-error.js.map +1 -1
- package/dist/src/evals/harbor-node-runtime.js +146 -0
- package/dist/src/evals/harbor-node-runtime.js.map +1 -0
- package/dist/src/evals/harbor-phase-timings.js +60 -0
- package/dist/src/evals/harbor-phase-timings.js.map +1 -0
- package/dist/src/evals/index.js +40 -2
- package/dist/src/evals/index.js.map +1 -1
- package/dist/src/evals/infrastructure-retry.js +208 -84
- package/dist/src/evals/infrastructure-retry.js.map +1 -1
- package/dist/src/evals/interaction-capture-import.js +49 -0
- package/dist/src/evals/interaction-capture-import.js.map +1 -0
- package/dist/src/evals/local-docker-provider.js +372 -0
- package/dist/src/evals/local-docker-provider.js.map +1 -0
- package/dist/src/evals/local-eval-planning.js +80 -0
- package/dist/src/evals/local-eval-planning.js.map +1 -0
- package/dist/src/evals/model-capture-plan.js +22 -0
- package/dist/src/evals/model-capture-plan.js.map +1 -0
- package/dist/src/evals/model-capture-runtime.js +114 -0
- package/dist/src/evals/model-capture-runtime.js.map +1 -0
- package/dist/src/evals/model-proxy-runtime-state.js +53 -0
- package/dist/src/evals/model-proxy-runtime-state.js.map +1 -0
- package/dist/src/evals/native-phase-evidence.js +440 -0
- package/dist/src/evals/native-phase-evidence.js.map +1 -0
- package/dist/src/evals/physical-retry-work.js +31 -0
- package/dist/src/evals/physical-retry-work.js.map +1 -0
- package/dist/src/evals/planned-execution-support.js +66 -0
- package/dist/src/evals/planned-execution-support.js.map +1 -0
- package/dist/src/evals/planned-execution.js +469 -0
- package/dist/src/evals/planned-execution.js.map +1 -0
- package/dist/src/evals/planned-progress-publisher.js +62 -0
- package/dist/src/evals/planned-progress-publisher.js.map +1 -0
- package/dist/src/evals/planned-remote-execution.js +40 -0
- package/dist/src/evals/planned-remote-execution.js.map +1 -0
- package/dist/src/evals/planned-retry-execution.js +174 -0
- package/dist/src/evals/planned-retry-execution.js.map +1 -0
- package/dist/src/evals/planned-retry-lifecycle.js +188 -0
- package/dist/src/evals/planned-retry-lifecycle.js.map +1 -0
- package/dist/src/evals/preparation-rerun.js +225 -0
- package/dist/src/evals/preparation-rerun.js.map +1 -0
- package/dist/src/evals/prepared-harness.js +76 -0
- package/dist/src/evals/prepared-harness.js.map +1 -0
- package/dist/src/evals/progress.js +52 -12
- package/dist/src/evals/progress.js.map +1 -1
- package/dist/src/evals/records.js +20 -3
- package/dist/src/evals/records.js.map +1 -1
- package/dist/src/evals/recovery.js +242 -0
- package/dist/src/evals/recovery.js.map +1 -0
- package/dist/src/evals/regrade-evidence.js +77 -0
- package/dist/src/evals/regrade-evidence.js.map +1 -0
- package/dist/src/evals/remote-infrastructure-retry.js +165 -0
- package/dist/src/evals/remote-infrastructure-retry.js.map +1 -0
- package/dist/src/evals/request.js +21 -3
- package/dist/src/evals/request.js.map +1 -1
- package/dist/src/evals/rerun-inputs.js +108 -0
- package/dist/src/evals/rerun-inputs.js.map +1 -0
- package/dist/src/evals/rerun-slots.js +4 -3
- package/dist/src/evals/rerun-slots.js.map +1 -1
- package/dist/src/evals/rerun-types.js +45 -0
- package/dist/src/evals/rerun-types.js.map +1 -0
- package/dist/src/evals/rerun.js +133 -88
- package/dist/src/evals/rerun.js.map +1 -1
- package/dist/src/evals/result-helpers.js +78 -0
- package/dist/src/evals/result-helpers.js.map +1 -0
- package/dist/src/evals/resume-state.js +36 -0
- package/dist/src/evals/resume-state.js.map +1 -0
- package/dist/src/evals/retry-backoff.js +20 -0
- package/dist/src/evals/retry-backoff.js.map +1 -0
- package/dist/src/evals/retry-state.js +195 -0
- package/dist/src/evals/retry-state.js.map +1 -0
- package/dist/src/evals/scheduler-metrics.js +95 -0
- package/dist/src/evals/scheduler-metrics.js.map +1 -0
- package/dist/src/evals/scheduler-trace-replay.js +107 -0
- package/dist/src/evals/scheduler-trace-replay.js.map +1 -0
- package/dist/src/evals/service-types.js +2 -0
- package/dist/src/evals/service-types.js.map +1 -0
- package/dist/src/evals/service.js +331 -347
- package/dist/src/evals/service.js.map +1 -1
- package/dist/src/evals/task-resources.js +209 -0
- package/dist/src/evals/task-resources.js.map +1 -0
- package/dist/src/evals/trial-environment-evidence.js +87 -0
- package/dist/src/evals/trial-environment-evidence.js.map +1 -0
- package/dist/src/evals/trial-execution-evidence.js +13 -0
- package/dist/src/evals/trial-execution-evidence.js.map +1 -0
- package/dist/src/evals/trial-import-identity.js +31 -0
- package/dist/src/evals/trial-import-identity.js.map +1 -0
- package/dist/src/evals/trial-import.js +183 -188
- package/dist/src/evals/trial-import.js.map +1 -1
- package/dist/src/evals/trial-publication-recovery.js +144 -0
- package/dist/src/evals/trial-publication-recovery.js.map +1 -0
- package/dist/src/evals/trial-publication.js +25 -0
- package/dist/src/evals/trial-publication.js.map +1 -0
- package/dist/src/evals/trial-reference-validation.js +49 -0
- package/dist/src/evals/trial-reference-validation.js.map +1 -0
- package/dist/src/evals/verifier-artifacts.js +317 -0
- package/dist/src/evals/verifier-artifacts.js.map +1 -0
- package/dist/src/evals/verifier-eligibility.js +25 -0
- package/dist/src/evals/verifier-eligibility.js.map +1 -0
- package/dist/src/evals/verifier-only-rerun.js +210 -0
- package/dist/src/evals/verifier-only-rerun.js.map +1 -0
- package/dist/src/evals/verifier-runtime.js +61 -0
- package/dist/src/evals/verifier-runtime.js.map +1 -0
- package/dist/src/evals/verifier-score-artifacts.js +119 -0
- package/dist/src/evals/verifier-score-artifacts.js.map +1 -0
- package/dist/src/evals/work-item-artifacts.js +23 -0
- package/dist/src/evals/work-item-artifacts.js.map +1 -0
- package/dist/src/foundation/config.js +17 -0
- package/dist/src/foundation/config.js.map +1 -1
- package/dist/src/foundation/contained-file.js +85 -0
- package/dist/src/foundation/contained-file.js.map +1 -0
- package/dist/src/foundation/credential-redaction.js +162 -0
- package/dist/src/foundation/credential-redaction.js.map +1 -0
- package/dist/src/foundation/environment-image-references.js +59 -0
- package/dist/src/foundation/environment-image-references.js.map +1 -0
- package/dist/src/foundation/executable.js +45 -5
- package/dist/src/foundation/executable.js.map +1 -1
- package/dist/src/foundation/fs.js +2 -1
- package/dist/src/foundation/fs.js.map +1 -1
- package/dist/src/foundation/index.js +6 -2
- package/dist/src/foundation/index.js.map +1 -1
- package/dist/src/foundation/process-identity.js +69 -0
- package/dist/src/foundation/process-identity.js.map +1 -0
- package/dist/src/images/context.js +47 -0
- package/dist/src/images/context.js.map +1 -0
- package/dist/src/images/docker-buildkit.js +137 -0
- package/dist/src/images/docker-buildkit.js.map +1 -0
- package/dist/src/images/dockerfile.js +67 -0
- package/dist/src/images/dockerfile.js.map +1 -0
- package/dist/src/images/gc.js +261 -0
- package/dist/src/images/gc.js.map +1 -0
- package/dist/src/images/index.js +10 -0
- package/dist/src/images/index.js.map +1 -0
- package/dist/src/images/manifest.js +109 -0
- package/dist/src/images/manifest.js.map +1 -0
- package/dist/src/images/ownership.js +5 -0
- package/dist/src/images/ownership.js.map +1 -0
- package/dist/src/images/records.js +32 -0
- package/dist/src/images/records.js.map +1 -0
- package/dist/src/images/registry.js +154 -0
- package/dist/src/images/registry.js.map +1 -0
- package/dist/src/images/service.js +224 -0
- package/dist/src/images/service.js.map +1 -0
- package/dist/src/model-access/capture.js +255 -0
- package/dist/src/model-access/capture.js.map +1 -0
- package/dist/src/model-access/index.js +5 -0
- package/dist/src/model-access/index.js.map +1 -0
- package/dist/src/model-access/policy.js +49 -0
- package/dist/src/model-access/policy.js.map +1 -0
- package/dist/src/model-access/proxy.js +336 -0
- package/dist/src/model-access/proxy.js.map +1 -0
- package/dist/src/model-access/records.js +167 -0
- package/dist/src/model-access/records.js.map +1 -0
- package/dist/src/runs/adapter-process.js +12 -0
- package/dist/src/runs/adapter-process.js.map +1 -0
- package/dist/src/runs/bundle.js +450 -0
- package/dist/src/runs/bundle.js.map +1 -0
- package/dist/src/runs/events.js +12 -1
- package/dist/src/runs/events.js.map +1 -1
- package/dist/src/runs/executor.js +46 -45
- package/dist/src/runs/executor.js.map +1 -1
- package/dist/src/runs/finalizer.js +17 -0
- package/dist/src/runs/finalizer.js.map +1 -1
- package/dist/src/runs/index.js +8 -1
- package/dist/src/runs/index.js.map +1 -1
- package/dist/src/runs/manifest.js +5 -5
- package/dist/src/runs/manifest.js.map +1 -1
- package/dist/src/runs/phase-bundle.js +86 -0
- package/dist/src/runs/phase-bundle.js.map +1 -0
- package/dist/src/runs/phase-cancellation.js +107 -0
- package/dist/src/runs/phase-cancellation.js.map +1 -0
- package/dist/src/runs/phase-group.js +113 -0
- package/dist/src/runs/phase-group.js.map +1 -0
- package/dist/src/runs/query.js +1 -1
- package/dist/src/runs/query.js.map +1 -1
- package/dist/src/runs/records.js +3 -0
- package/dist/src/runs/records.js.map +1 -1
- package/dist/src/runs/request.js +10 -3
- package/dist/src/runs/request.js.map +1 -1
- package/dist/src/runs/training-candidate.js +161 -0
- package/dist/src/runs/training-candidate.js.map +1 -0
- package/dist/src/runs/verifier-evidence-redaction.js +61 -0
- package/dist/src/runs/verifier-evidence-redaction.js.map +1 -0
- package/dist/src/runs/verifier-evidence.js +459 -0
- package/dist/src/runs/verifier-evidence.js.map +1 -0
- package/dist/src/runs/verifier-structured-evidence.js +83 -0
- package/dist/src/runs/verifier-structured-evidence.js.map +1 -0
- package/dist/src/trajectories/analysis.js +288 -0
- package/dist/src/trajectories/analysis.js.map +1 -0
- package/dist/src/trajectories/chunk-projection.js +148 -0
- package/dist/src/trajectories/chunk-projection.js.map +1 -0
- package/dist/src/trajectories/content-projection.js +389 -0
- package/dist/src/trajectories/content-projection.js.map +1 -0
- package/dist/src/trajectories/dsh-chunk-contract.js +176 -0
- package/dist/src/trajectories/dsh-chunk-contract.js.map +1 -0
- package/dist/src/trajectories/dsh-contract.js +284 -0
- package/dist/src/trajectories/dsh-contract.js.map +1 -0
- package/dist/src/trajectories/events-chunk-drill.js +95 -0
- package/dist/src/trajectories/events-chunk-drill.js.map +1 -0
- package/dist/src/trajectories/events-page.js +367 -0
- package/dist/src/trajectories/events-page.js.map +1 -0
- package/dist/src/trajectories/index.js +3 -0
- package/dist/src/trajectories/index.js.map +1 -1
- package/dist/src/trajectories/provider-capture.js +15 -24
- package/dist/src/trajectories/provider-capture.js.map +1 -1
- package/dist/src/trajectories/providers/deepseek.js +3 -3
- package/dist/src/trajectories/providers/deepseek.js.map +1 -1
- package/dist/src/trajectories/request-attempt.js +47 -0
- package/dist/src/trajectories/request-attempt.js.map +1 -0
- package/dist/src/trajectories/stream-reader.js +311 -0
- package/dist/src/trajectories/stream-reader.js.map +1 -0
- package/dist/src/trajectories/surface-fold.js +126 -0
- package/dist/src/trajectories/surface-fold.js.map +1 -0
- package/dist/src/workers/index.js +3 -0
- package/dist/src/workers/index.js.map +1 -0
- package/dist/src/workers/remote-harbor-work-spec.js +97 -0
- package/dist/src/workers/remote-harbor-work-spec.js.map +1 -0
- package/dist/src/workers/remote-harbor-worker.js +223 -0
- package/dist/src/workers/remote-harbor-worker.js.map +1 -0
- package/docs/schemas/benchmark-hook-request-v1.schema.json +65 -0
- package/docs/schemas/benchmark-hook-response-v1.schema.json +75 -0
- package/docs/schemas/benchmark-hook-v1.schema.json +33 -0
- package/docs/schemas/benchmark-lock-v1.schema.json +351 -0
- package/docs/schemas/benchmark-metric-v1.schema.json +37 -0
- package/docs/schemas/benchmark-package-v1.schema.json +167 -0
- package/docs/schemas/benchmark-phase-group.schema.json +41 -0
- package/docs/schemas/benchmark-profile-v1.schema.json +141 -0
- package/docs/schemas/benchmark-task-v1.schema.json +384 -0
- package/docs/schemas/benchmark-tool-result-v1.schema.json +30 -0
- package/docs/schemas/environment-build-index.schema.json +13 -0
- package/docs/schemas/environment-build-record.schema.json +25 -0
- package/docs/schemas/environment-image-manifest.schema.json +62 -0
- package/docs/schemas/error.schema.json +22 -0
- package/docs/schemas/eval-control.schema.json +36 -0
- package/docs/schemas/eval-execution-plan.schema.json +216 -0
- package/docs/schemas/eval-progress.schema.json +1 -15
- package/docs/schemas/eval-rerun-result.schema.json +58 -0
- package/docs/schemas/eval-rerun-state.schema.json +61 -0
- package/docs/schemas/eval-rerun-submission.schema.json +60 -0
- package/docs/schemas/eval-result.schema.json +32 -13
- package/docs/schemas/eval-submission-input.schema.json +14 -0
- package/docs/schemas/eval-submission.schema.json +68 -0
- package/docs/schemas/eval-trial-publication.schema.json +15 -0
- package/docs/schemas/eval-trial-reference.schema.json +129 -0
- package/docs/schemas/execution-evidence.schema.json +96 -0
- package/docs/schemas/execution-lease.schema.json +50 -0
- package/docs/schemas/execution-provider-status.schema.json +65 -0
- package/docs/schemas/execution-worker.schema.json +53 -0
- package/docs/schemas/harbor-process-exit.schema.json +14 -0
- package/docs/schemas/interaction-capture-ref.schema.json +45 -0
- package/docs/schemas/local-provider-execution.schema.json +33 -0
- package/docs/schemas/model-interaction.schema.json +41 -0
- package/docs/schemas/regrade-assessment-reference.schema.json +11 -0
- package/docs/schemas/remote-result-envelope.schema.json +32 -0
- package/docs/schemas/remote-tree-envelope.schema.json +25 -0
- package/docs/schemas/remote-work-offer.schema.json +58 -0
- package/docs/schemas/remote-work-receipt.schema.json +50 -0
- package/docs/schemas/remote-worker-artifact.schema.json +17 -0
- package/docs/schemas/remote-worker-event.schema.json +18 -0
- package/docs/schemas/remote-worker-heartbeat.schema.json +37 -0
- package/docs/schemas/remote-worker-registration.schema.json +52 -0
- package/docs/schemas/result-bundle-index.schema.json +133 -0
- package/docs/schemas/run-context.schema.json +15 -0
- package/docs/schemas/run-environment-images.schema.json +39 -0
- package/docs/schemas/run-manifest.schema.json +4 -0
- package/docs/schemas/run-request.schema.json +4 -0
- package/docs/schemas/training-data-candidate.schema.json +46 -0
- package/docs/schemas/trajectory-analysis.schema.json +258 -0
- package/docs/schemas/trajectory-events-page.schema.json +40 -0
- package/docs/schemas/verifier-assessment.schema.json +77 -0
- package/docs/schemas/verifier-evidence.schema.json +276 -0
- package/integrations/harbor/hitch_benchmark.py +252 -0
- package/integrations/harbor/hitch_candidate_recycle.py +262 -0
- package/integrations/harbor/hitch_harbor_agent.py +698 -568
- package/integrations/harbor/hitch_harbor_environment.py +469 -0
- package/integrations/harbor/hitch_harbor_task_resources.py +275 -0
- package/integrations/harbor/hitch_harbor_verifier.py +83 -2
- package/integrations/harbor/hitch_phase_supervisor.py +452 -0
- package/integrations/harbor/hitch_tool_client.mjs +91 -0
- package/integrations/model-call/cli.js +57 -0
- package/package.json +30 -6
|
@@ -0,0 +1,29 @@
|
|
|
1
|
+
export function emitEvalPlanLifecycle(sink, plan, planningStartedAt) {
|
|
2
|
+
sink.emit({
|
|
3
|
+
type: "eval.plan.created",
|
|
4
|
+
duration_ms: Date.now() - planningStartedAt,
|
|
5
|
+
work_items: plan.work_items.length,
|
|
6
|
+
membership: plan.membership,
|
|
7
|
+
});
|
|
8
|
+
for (const workItem of plan.work_items)
|
|
9
|
+
sink.emit({
|
|
10
|
+
type: "eval.work.queued",
|
|
11
|
+
work_id: workItem.work_id,
|
|
12
|
+
task_ids: workItem.task_ids,
|
|
13
|
+
reservation: workItem.reservation,
|
|
14
|
+
});
|
|
15
|
+
for (const workItem of plan.work_items)
|
|
16
|
+
if (workItem.scheduling)
|
|
17
|
+
sink.emit({
|
|
18
|
+
type: "eval.work.priority-computed",
|
|
19
|
+
work_id: workItem.work_id,
|
|
20
|
+
slot_id: workItem.slots[0],
|
|
21
|
+
task_id: workItem.task_ids[0],
|
|
22
|
+
policy: workItem.scheduling.policy,
|
|
23
|
+
estimated_duration_ms: workItem.scheduling.estimated_duration_ms,
|
|
24
|
+
remaining_path_ms: workItem.scheduling.remaining_path_ms,
|
|
25
|
+
estimate_source: workItem.scheduling.estimate_source,
|
|
26
|
+
estimate_sample_count: workItem.scheduling.estimate_sample_count,
|
|
27
|
+
});
|
|
28
|
+
}
|
|
29
|
+
//# sourceMappingURL=eval-lifecycle-events.js.map
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"eval-lifecycle-events.js","sourceRoot":"","sources":["../../../src/evals/eval-lifecycle-events.ts"],"names":[],"mappings":"AAGA,MAAM,UAAU,qBAAqB,CAAC,IAAmB,EAAE,IAAyB,EAAE,iBAAyB;IAC7G,IAAI,CAAC,IAAI,CAAC;QACR,IAAI,EAAE,mBAAmB;QACzB,WAAW,EAAE,IAAI,CAAC,GAAG,EAAE,GAAG,iBAAiB;QAC3C,UAAU,EAAE,IAAI,CAAC,UAAU,CAAC,MAAM;QAClC,UAAU,EAAE,IAAI,CAAC,UAAU;KAC5B,CAAC,CAAC;IACH,KAAK,MAAM,QAAQ,IAAI,IAAI,CAAC,UAAU;QAAE,IAAI,CAAC,IAAI,CAAC;YAChD,IAAI,EAAE,kBAAkB;YACxB,OAAO,EAAE,QAAQ,CAAC,OAAO;YACzB,QAAQ,EAAE,QAAQ,CAAC,QAAQ;YAC3B,WAAW,EAAE,QAAQ,CAAC,WAAW;SAClC,CAAC,CAAC;IACH,KAAK,MAAM,QAAQ,IAAI,IAAI,CAAC,UAAU;QAAE,IAAI,QAAQ,CAAC,UAAU;YAAE,IAAI,CAAC,IAAI,CAAC;gBACzE,IAAI,EAAE,6BAA6B;gBACnC,OAAO,EAAE,QAAQ,CAAC,OAAO;gBACzB,OAAO,EAAE,QAAQ,CAAC,KAAK,CAAC,CAAC,CAAC;gBAC1B,OAAO,EAAE,QAAQ,CAAC,QAAQ,CAAC,CAAC,CAAC;gBAC7B,MAAM,EAAE,QAAQ,CAAC,UAAU,CAAC,MAAM;gBAClC,qBAAqB,EAAE,QAAQ,CAAC,UAAU,CAAC,qBAAqB;gBAChE,iBAAiB,EAAE,QAAQ,CAAC,UAAU,CAAC,iBAAiB;gBACxD,eAAe,EAAE,QAAQ,CAAC,UAAU,CAAC,eAAe;gBACpD,qBAAqB,EAAE,QAAQ,CAAC,UAAU,CAAC,qBAAqB;aACjE,CAAC,CAAC;AACL,CAAC"}
|
|
@@ -0,0 +1,51 @@
|
|
|
1
|
+
import path from "node:path";
|
|
2
|
+
import { SCHEMA_VERSION, atomicWriteJSON, readJSON } from "../foundation/index.js";
|
|
3
|
+
import { writeEvalProgress } from "./progress.js";
|
|
4
|
+
/** Persist the reproducible, artifact-independent part of an eval before setup work begins. */
|
|
5
|
+
export async function writeEvalPlanningCheckpoint(evalDirectory, plan, progress) {
|
|
6
|
+
await Promise.all([
|
|
7
|
+
atomicWriteJSON(path.join(evalDirectory, "logical-plan.json"), plan),
|
|
8
|
+
writeEvalProgress(evalDirectory, progress),
|
|
9
|
+
]);
|
|
10
|
+
}
|
|
11
|
+
export function materializeEvalPlan(logical, modelCapture, artifactFields) {
|
|
12
|
+
const { kind: _kind, model_capture: _modelCapture, ...base } = logical;
|
|
13
|
+
return {
|
|
14
|
+
...base,
|
|
15
|
+
...(modelCapture ? { model_capture: modelCapture } : {}),
|
|
16
|
+
...artifactFields,
|
|
17
|
+
};
|
|
18
|
+
}
|
|
19
|
+
/** Read only the restart inputs that cannot be reconstructed from request.json. */
|
|
20
|
+
export async function readEvalLogicalPlan(evalDirectory, evalId, request) {
|
|
21
|
+
const value = await readJSON(path.join(evalDirectory, "logical-plan.json"), null);
|
|
22
|
+
if (value === null)
|
|
23
|
+
return null;
|
|
24
|
+
if (!value || typeof value !== "object" || Array.isArray(value))
|
|
25
|
+
throw new TypeError("eval logical plan must be an object");
|
|
26
|
+
const plan = value;
|
|
27
|
+
if (plan.schema_version !== SCHEMA_VERSION || plan.kind !== "eval-logical-plan" || plan.eval_id !== evalId
|
|
28
|
+
|| plan.dataset !== request.dataset || plan.benchmark_id !== request.benchmark_id
|
|
29
|
+
|| plan.benchmark_revision !== request.benchmark_revision) {
|
|
30
|
+
throw new TypeError("eval logical plan identity is invalid");
|
|
31
|
+
}
|
|
32
|
+
if (plan.model_capture !== undefined)
|
|
33
|
+
parseModelCapturePlan(plan.model_capture);
|
|
34
|
+
return plan;
|
|
35
|
+
}
|
|
36
|
+
export function logicalPlanModelCapture(plan) {
|
|
37
|
+
return plan?.model_capture === undefined ? undefined : parseModelCapturePlan(plan.model_capture);
|
|
38
|
+
}
|
|
39
|
+
function parseModelCapturePlan(value) {
|
|
40
|
+
if (!value || typeof value !== "object" || Array.isArray(value))
|
|
41
|
+
throw new TypeError("eval logical model capture plan is invalid");
|
|
42
|
+
const plan = value;
|
|
43
|
+
const modes = new Set(["off", "native", "proxy", "hybrid"]);
|
|
44
|
+
if (!modes.has(String(plan.requested_mode)) || !modes.has(String(plan.effective_mode)) || typeof plan.required !== "boolean"
|
|
45
|
+
|| plan.topology !== undefined && plan.topology !== "host-side" && plan.topology !== "in-sandbox"
|
|
46
|
+
|| plan.degraded_reason !== undefined && typeof plan.degraded_reason !== "string") {
|
|
47
|
+
throw new TypeError("eval logical model capture plan is invalid");
|
|
48
|
+
}
|
|
49
|
+
return plan;
|
|
50
|
+
}
|
|
51
|
+
//# sourceMappingURL=eval-logical-plan.js.map
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"eval-logical-plan.js","sourceRoot":"","sources":["../../../src/evals/eval-logical-plan.ts"],"names":[],"mappings":"AAAA,OAAO,IAAI,MAAM,WAAW,CAAC;AAE7B,OAAO,EAAE,cAAc,EAAE,eAAe,EAAE,QAAQ,EAAE,MAAM,wBAAwB,CAAC;AACnF,OAAO,EAAE,iBAAiB,EAAE,MAAM,eAAe,CAAC;AAoBlD,+FAA+F;AAC/F,MAAM,CAAC,KAAK,UAAU,2BAA2B,CAC/C,aAAqB,EACrB,IAAuB,EACvB,QAAwB;IAExB,MAAM,OAAO,CAAC,GAAG,CAAC;QAChB,eAAe,CAAC,IAAI,CAAC,IAAI,CAAC,aAAa,EAAE,mBAAmB,CAAC,EAAE,IAAI,CAAC;QACpE,iBAAiB,CAAC,aAAa,EAAE,QAAQ,CAAC;KAC3C,CAAC,CAAC;AACL,CAAC;AAED,MAAM,UAAU,mBAAmB,CACjC,OAA0B,EAC1B,YAA4C,EAC5C,cAAuC;IAEvC,MAAM,EAAE,IAAI,EAAE,KAAK,EAAE,aAAa,EAAE,aAAa,EAAE,GAAG,IAAI,EAAE,GAAG,OAAO,CAAC;IACvE,OAAO;QACL,GAAG,IAAI;QACP,GAAG,CAAC,YAAY,CAAC,CAAC,CAAC,EAAE,aAAa,EAAE,YAAY,EAAE,CAAC,CAAC,CAAC,EAAE,CAAC;QACxD,GAAG,cAAc;KAClB,CAAC;AACJ,CAAC;AAED,mFAAmF;AACnF,MAAM,CAAC,KAAK,UAAU,mBAAmB,CACvC,aAAqB,EACrB,MAAc,EACd,OAAoB;IAEpB,MAAM,KAAK,GAAG,MAAM,QAAQ,CAAiB,IAAI,CAAC,IAAI,CAAC,aAAa,EAAE,mBAAmB,CAAC,EAAE,IAAI,CAAC,CAAC;IAClG,IAAI,KAAK,KAAK,IAAI;QAAE,OAAO,IAAI,CAAC;IAChC,IAAI,CAAC,KAAK,IAAI,OAAO,KAAK,KAAK,QAAQ,IAAI,KAAK,CAAC,OAAO,CAAC,KAAK,CAAC;QAAE,MAAM,IAAI,SAAS,CAAC,qCAAqC,CAAC,CAAC;IAC5H,MAAM,IAAI,GAAG,KAAgC,CAAC;IAC9C,IAAI,IAAI,CAAC,cAAc,KAAK,cAAc,IAAI,IAAI,CAAC,IAAI,KAAK,mBAAmB,IAAI,IAAI,CAAC,OAAO,KAAK,MAAM;WACrG,IAAI,CAAC,OAAO,KAAK,OAAO,CAAC,OAAO,IAAI,IAAI,CAAC,YAAY,KAAK,OAAO,CAAC,YAAY;WAC9E,IAAI,CAAC,kBAAkB,KAAK,OAAO,CAAC,kBAAkB,EAAE,CAAC;QAC5D,MAAM,IAAI,SAAS,CAAC,uCAAuC,CAAC,CAAC;IAC/D,CAAC;IACD,IAAI,IAAI,CAAC,aAAa,KAAK,SAAS;QAAE,qBAAqB,CAAC,IAAI,CAAC,aAAa,CAAC,CAAC;IAChF,OAAO,IAAoC,CAAC;AAC9C,CAAC;AAED,MAAM,UAAU,uBAAuB,CAAC,IAA8B;IACpE,OAAO,IAAI,EAAE,aAAa,KAAK,SAAS,CAAC,CAAC,CAAC,SAAS,CAAC,CAAC,CAAC,qBAAqB,CAAC,IAAI,CAAC,aAAa,CAAC,CAAC;AACnG,CAAC;AAED,SAAS,qBAAqB,CAAC,KAAc;IAC3C,IAAI,CAAC,KAAK,IAAI,OAAO,KAAK,KAAK,QAAQ,IAAI,KAAK,CAAC,OAAO,CAAC,KAAK,CAAC;QAAE,MAAM,IAAI,SAAS,CAAC,4CAA4C,CAAC,CAAC;IACnI,MAAM,IAAI,GAAG,KAAgC,CAAC;IAC9C,MAAM,KAAK,GAAG,IAAI,GAAG,CAAC,CAAC,KAAK,EAAE,QAAQ,EAAE,OAAO,EAAE,QAAQ,CAAC,CAAC,CAAC;IAC5D,IAAI,CAAC,KAAK,CAAC,GAAG,CAAC,MAAM,CAAC,IAAI,CAAC,cAAc,CAAC,CAAC,IAAI,CAAC,KAAK,CAAC,GAAG,CAAC,MAAM,CAAC,IAAI,CAAC,cAAc,CAAC,CAAC,IAAI,OAAO,IAAI,CAAC,QAAQ,KAAK,SAAS;WACvH,IAAI,CAAC,QAAQ,KAAK,SAAS,IAAI,IAAI,CAAC,QAAQ,KAAK,WAAW,IAAI,IAAI,CAAC,QAAQ,KAAK,YAAY;WAC9F,IAAI,CAAC,eAAe,KAAK,SAAS,IAAI,OAAO,IAAI,CAAC,eAAe,KAAK,QAAQ,EAAE,CAAC;QACpF,MAAM,IAAI,SAAS,CAAC,4CAA4C,CAAC,CAAC;IACpE,CAAC;IACD,OAAO,IAAqC,CAAC;AAC/C,CAAC"}
|
|
@@ -0,0 +1,81 @@
|
|
|
1
|
+
import { SCHEMA_VERSION } from "../foundation/index.js";
|
|
2
|
+
import { preparedArtifactPlanFields } from "./eval-artifact-planning.js";
|
|
3
|
+
import { infrastructureFailureTrials } from "./infrastructure-retry.js";
|
|
4
|
+
import { invalidTrialSlots } from "./rerun-slots.js";
|
|
5
|
+
import { localSourceBackendFailure, summarizeTrialRefs, transportSummary } from "./result-helpers.js";
|
|
6
|
+
export function buildCompletedEvalResult(input) {
|
|
7
|
+
const trialRefs = input.progress.trials;
|
|
8
|
+
const localSourceFailure = input.localTransport
|
|
9
|
+
? input.backendRuns.some(({ run }) => localSourceBackendFailure(run.rawResult))
|
|
10
|
+
|| input.infrastructureRetryRuns.some(({ run }) => localSourceBackendFailure(run.rawResult))
|
|
11
|
+
: false;
|
|
12
|
+
const expectedBackendRuns = input.plannedTaskExecution ? input.executionWorkItems : input.request.attempts;
|
|
13
|
+
const backendsSucceeded = input.backendRuns.every(({ run }) => run.backend.process_exit_code === 0 && run.rawResult !== null)
|
|
14
|
+
&& (input.plannedTaskExecution ? trialRefs.length === input.plannedTrials : input.backendRuns.length === expectedBackendRuns);
|
|
15
|
+
const invalidTrials = input.localTaskIds === null
|
|
16
|
+
? trialRefs.filter((trial) => trial.observation_status !== "valid").map((trial) => ({ task_id: trial.task_id, attempt: trial.attempt }))
|
|
17
|
+
: invalidTrialSlots(input.localTaskIds, input.request.attempts, input.progress);
|
|
18
|
+
const infrastructureFailures = infrastructureFailureTrials(trialRefs);
|
|
19
|
+
const infrastructureFailureSlots = infrastructureFailures.map((trial) => ({ task_id: trial.task_id, attempt: trial.attempt }));
|
|
20
|
+
const verifierRetriesExhausted = input.request.infrastructure_retries > 0
|
|
21
|
+
&& infrastructureFailures.some((trial) => trial.invalid_reason === "verifier_infrastructure_failure");
|
|
22
|
+
const infrastructureErrorCode = verifierRetriesExhausted
|
|
23
|
+
|| (input.request.infrastructure_retries > 0 && input.infrastructureRetryRuns.length > 0)
|
|
24
|
+
? "eval_infrastructure_retries_exhausted"
|
|
25
|
+
: "eval_has_infrastructure_failures";
|
|
26
|
+
const succeeded = !input.cancelled && !localSourceFailure && backendsSucceeded && invalidTrials.length === 0;
|
|
27
|
+
const singleBackend = input.backendRuns.length === 1 ? input.backendRuns[0].run : undefined;
|
|
28
|
+
return {
|
|
29
|
+
schema_version: SCHEMA_VERSION,
|
|
30
|
+
eval_id: input.evalId,
|
|
31
|
+
status: input.cancelled ? "cancelled" : succeeded ? "succeeded" : "failed",
|
|
32
|
+
exit_code: input.cancelled ? 9 : succeeded ? 0 : 13,
|
|
33
|
+
...(singleBackend ? { backend: singleBackend.backend, backend_summary: singleBackend.summary } : {}),
|
|
34
|
+
...(input.plannedTaskExecution ? {
|
|
35
|
+
backend_work_items: input.backendRuns.map(({ attempt, workId, tasks, leaseId, run }) => ({
|
|
36
|
+
work_id: workId, lease_id: leaseId, attempt, tasks, backend: run.backend, backend_summary: run.summary,
|
|
37
|
+
})),
|
|
38
|
+
} : input.request.attempts > 1 ? {
|
|
39
|
+
backend_runs: input.backendRuns.map(({ attempt, run }) => ({ attempt, backend: run.backend, backend_summary: run.summary })),
|
|
40
|
+
} : {}),
|
|
41
|
+
infrastructure_retry_policy: {
|
|
42
|
+
max_retries: input.request.infrastructure_retries,
|
|
43
|
+
backoff_ms: input.request.infrastructure_retry_backoff_ms,
|
|
44
|
+
verifier_execution: "same_trial_verifier_only",
|
|
45
|
+
candidate_rerun_on_verifier_failure: false,
|
|
46
|
+
},
|
|
47
|
+
...(input.infrastructureRetryRuns.length > 0 ? {
|
|
48
|
+
infrastructure_retry_runs: input.infrastructureRetryRuns.map(({ attempt, retry, tasks, triggers, refs, run, leaseId, workId }) => ({
|
|
49
|
+
execution_kind: "physical-infrastructure-retry", ...(leaseId ? { lease_id: leaseId } : {}),
|
|
50
|
+
...(workId ? { work_id: workId } : {}), attempt, retry, tasks, trigger_trials: triggers,
|
|
51
|
+
trials: refs, backend: run.backend, backend_summary: run.summary,
|
|
52
|
+
})),
|
|
53
|
+
} : {}),
|
|
54
|
+
candidate: input.candidate,
|
|
55
|
+
dataset: input.request.dataset,
|
|
56
|
+
benchmark_id: input.request.benchmark_id,
|
|
57
|
+
benchmark_revision: input.request.benchmark_revision,
|
|
58
|
+
generation: input.progress.generation,
|
|
59
|
+
trials: trialRefs,
|
|
60
|
+
summary: summarizeTrialRefs(trialRefs),
|
|
61
|
+
...(input.schedulerSummary ? { scheduler_summary: input.schedulerSummary } : {}),
|
|
62
|
+
...preparedArtifactPlanFields(input.preparedAssignments),
|
|
63
|
+
...(input.localTransport ? { local_source_transport: transportSummary(input.localTransport) } : {}),
|
|
64
|
+
...(succeeded ? {} : {
|
|
65
|
+
error: {
|
|
66
|
+
code: input.cancelled ? "cancelled" : localSourceFailure ? "local_source_materialize_failed"
|
|
67
|
+
: !backendsSucceeded ? "harbor_failed" : infrastructureFailures.length > 0
|
|
68
|
+
? infrastructureErrorCode : "eval_has_invalid_tasks",
|
|
69
|
+
message: input.cancelled ? "eval was cancelled"
|
|
70
|
+
: localSourceFailure ? "Harbor rejected the transported local Git source before candidate execution"
|
|
71
|
+
: !backendsSucceeded ? `Harbor work items completed ${input.backendRuns.length}/${expectedBackendRuns}`
|
|
72
|
+
: infrastructureFailures.length > 0
|
|
73
|
+
? `${infrastructureErrorCode === "eval_infrastructure_retries_exhausted" ? "infrastructure retries exhausted" : "verifier infrastructure failure"}: ${infrastructureFailureSlots.map(trial => `${trial.task_id}#${trial.attempt}`).join(", ")}`
|
|
74
|
+
: `eval has invalid or missing trials: ${invalidTrials.map(trial => `${trial.task_id}#${trial.attempt}`).join(", ")}`,
|
|
75
|
+
},
|
|
76
|
+
}),
|
|
77
|
+
started_at: input.startedAt.toISOString(),
|
|
78
|
+
completed_at: new Date().toISOString(),
|
|
79
|
+
};
|
|
80
|
+
}
|
|
81
|
+
//# sourceMappingURL=eval-result-builder.js.map
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"eval-result-builder.js","sourceRoot":"","sources":["../../../src/evals/eval-result-builder.ts"],"names":[],"mappings":"AAEA,OAAO,EAAE,cAAc,EAAE,MAAM,wBAAwB,CAAC;AACxD,OAAO,EAAE,0BAA0B,EAAuC,MAAM,6BAA6B,CAAC;AAC9G,OAAO,EAAE,2BAA2B,EAA+B,MAAM,2BAA2B,CAAC;AACrG,OAAO,EAAE,iBAAiB,EAAE,MAAM,kBAAkB,CAAC;AACrD,OAAO,EAAE,yBAAyB,EAAE,kBAAkB,EAAE,gBAAgB,EAAE,MAAM,qBAAqB,CAAC;AAWtG,MAAM,UAAU,wBAAwB,CAAC,KAgBxC;IACC,MAAM,SAAS,GAAG,KAAK,CAAC,QAAQ,CAAC,MAAM,CAAC;IACxC,MAAM,kBAAkB,GAAG,KAAK,CAAC,cAAc;QAC7C,CAAC,CAAC,KAAK,CAAC,WAAW,CAAC,IAAI,CAAC,CAAC,EAAE,GAAG,EAAE,EAAE,EAAE,CAAC,yBAAyB,CAAC,GAAG,CAAC,SAAS,CAAC,CAAC;eAC1E,KAAK,CAAC,uBAAuB,CAAC,IAAI,CAAC,CAAC,EAAE,GAAG,EAAE,EAAE,EAAE,CAAC,yBAAyB,CAAC,GAAG,CAAC,SAAS,CAAC,CAAC;QAC9F,CAAC,CAAC,KAAK,CAAC;IACV,MAAM,mBAAmB,GAAG,KAAK,CAAC,oBAAoB,CAAC,CAAC,CAAC,KAAK,CAAC,kBAAkB,CAAC,CAAC,CAAC,KAAK,CAAC,OAAO,CAAC,QAAQ,CAAC;IAC3G,MAAM,iBAAiB,GAAG,KAAK,CAAC,WAAW,CAAC,KAAK,CAAC,CAAC,EAAE,GAAG,EAAE,EAAE,EAAE,CAAC,GAAG,CAAC,OAAO,CAAC,iBAAiB,KAAK,CAAC,IAAI,GAAG,CAAC,SAAS,KAAK,IAAI,CAAC;WACxH,CAAC,KAAK,CAAC,oBAAoB,CAAC,CAAC,CAAC,SAAS,CAAC,MAAM,KAAK,KAAK,CAAC,aAAa,CAAC,CAAC,CAAC,KAAK,CAAC,WAAW,CAAC,MAAM,KAAK,mBAAmB,CAAC,CAAC;IAChI,MAAM,aAAa,GAAG,KAAK,CAAC,YAAY,KAAK,IAAI;QAC/C,CAAC,CAAC,SAAS,CAAC,MAAM,CAAC,CAAC,KAAK,EAAE,EAAE,CAAC,KAAK,CAAC,kBAAkB,KAAK,OAAO,CAAC,CAAC,GAAG,CAAC,CAAC,KAAK,EAAE,EAAE,CAAC,CAAC,EAAE,OAAO,EAAE,KAAK,CAAC,OAAO,EAAE,OAAO,EAAE,KAAK,CAAC,OAAO,EAAE,CAAC,CAAC;QACxI,CAAC,CAAC,iBAAiB,CAAC,KAAK,CAAC,YAAY,EAAE,KAAK,CAAC,OAAO,CAAC,QAAQ,EAAE,KAAK,CAAC,QAAQ,CAAC,CAAC;IAClF,MAAM,sBAAsB,GAAG,2BAA2B,CAAC,SAAS,CAAC,CAAC;IACtE,MAAM,0BAA0B,GAAG,sBAAsB,CAAC,GAAG,CAAC,CAAC,KAAK,EAAE,EAAE,CAAC,CAAC,EAAE,OAAO,EAAE,KAAK,CAAC,OAAO,EAAE,OAAO,EAAE,KAAK,CAAC,OAAO,EAAE,CAAC,CAAC,CAAC;IAC/H,MAAM,wBAAwB,GAAG,KAAK,CAAC,OAAO,CAAC,sBAAsB,GAAG,CAAC;WACpE,sBAAsB,CAAC,IAAI,CAAC,CAAC,KAAK,EAAE,EAAE,CAAC,KAAK,CAAC,cAAc,KAAK,iCAAiC,CAAC,CAAC;IACxG,MAAM,uBAAuB,GAAG,wBAAwB;WACnD,CAAC,KAAK,CAAC,OAAO,CAAC,sBAAsB,GAAG,CAAC,IAAI,KAAK,CAAC,uBAAuB,CAAC,MAAM,GAAG,CAAC,CAAC;QACzF,CAAC,CAAC,uCAAuC;QACzC,CAAC,CAAC,kCAAkC,CAAC;IACvC,MAAM,SAAS,GAAG,CAAC,KAAK,CAAC,SAAS,IAAI,CAAC,kBAAkB,IAAI,iBAAiB,IAAI,aAAa,CAAC,MAAM,KAAK,CAAC,CAAC;IAC7G,MAAM,aAAa,GAAG,KAAK,CAAC,WAAW,CAAC,MAAM,KAAK,CAAC,CAAC,CAAC,CAAC,KAAK,CAAC,WAAW,CAAC,CAAC,CAAE,CAAC,GAAG,CAAC,CAAC,CAAC,SAAS,CAAC;IAC7F,OAAO;QACL,cAAc,EAAE,cAAc;QAC9B,OAAO,EAAE,KAAK,CAAC,MAAM;QACrB,MAAM,EAAE,KAAK,CAAC,SAAS,CAAC,CAAC,CAAC,WAAW,CAAC,CAAC,CAAC,SAAS,CAAC,CAAC,CAAC,WAAW,CAAC,CAAC,CAAC,QAAQ;QAC1E,SAAS,EAAE,KAAK,CAAC,SAAS,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,SAAS,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,EAAE;QACnD,GAAG,CAAC,aAAa,CAAC,CAAC,CAAC,EAAE,OAAO,EAAE,aAAa,CAAC,OAAO,EAAE,eAAe,EAAE,aAAa,CAAC,OAAO,EAAE,CAAC,CAAC,CAAC,EAAE,CAAC;QACpG,GAAG,CAAC,KAAK,CAAC,oBAAoB,CAAC,CAAC,CAAC;YAC/B,kBAAkB,EAAE,KAAK,CAAC,WAAW,CAAC,GAAG,CAAC,CAAC,EAAE,OAAO,EAAE,MAAM,EAAE,KAAK,EAAE,OAAO,EAAE,GAAG,EAAE,EAAE,EAAE,CAAC,CAAC;gBACvF,OAAO,EAAE,MAAM,EAAE,QAAQ,EAAE,OAAO,EAAE,OAAO,EAAE,KAAK,EAAE,OAAO,EAAE,GAAG,CAAC,OAAO,EAAE,eAAe,EAAE,GAAG,CAAC,OAAO;aACvG,CAAC,CAAC;SACJ,CAAC,CAAC,CAAC,KAAK,CAAC,OAAO,CAAC,QAAQ,GAAG,CAAC,CAAC,CAAC,CAAC;YAC/B,YAAY,EAAE,KAAK,CAAC,WAAW,CAAC,GAAG,CAAC,CAAC,EAAE,OAAO,EAAE,GAAG,EAAE,EAAE,EAAE,CAAC,CAAC,EAAE,OAAO,EAAE,OAAO,EAAE,GAAG,CAAC,OAAO,EAAE,eAAe,EAAE,GAAG,CAAC,OAAO,EAAE,CAAC,CAAC;SAC7H,CAAC,CAAC,CAAC,EAAE,CAAC;QACP,2BAA2B,EAAE;YAC3B,WAAW,EAAE,KAAK,CAAC,OAAO,CAAC,sBAAsB;YACjD,UAAU,EAAE,KAAK,CAAC,OAAO,CAAC,+BAA+B;YACzD,kBAAkB,EAAE,0BAA0B;YAC9C,mCAAmC,EAAE,KAAK;SAC3C;QACD,GAAG,CAAC,KAAK,CAAC,uBAAuB,CAAC,MAAM,GAAG,CAAC,CAAC,CAAC,CAAC;YAC7C,yBAAyB,EAAE,KAAK,CAAC,uBAAuB,CAAC,GAAG,CAAC,CAAC,EAAE,OAAO,EAAE,KAAK,EAAE,KAAK,EAAE,QAAQ,EAAE,IAAI,EAAE,GAAG,EAAE,OAAO,EAAE,MAAM,EAAE,EAAE,EAAE,CAAC,CAAC;gBACjI,cAAc,EAAE,+BAA+B,EAAE,GAAG,CAAC,OAAO,CAAC,CAAC,CAAC,EAAE,QAAQ,EAAE,OAAO,EAAE,CAAC,CAAC,CAAC,EAAE,CAAC;gBAC1F,GAAG,CAAC,MAAM,CAAC,CAAC,CAAC,EAAE,OAAO,EAAE,MAAM,EAAE,CAAC,CAAC,CAAC,EAAE,CAAC,EAAE,OAAO,EAAE,KAAK,EAAE,KAAK,EAAE,cAAc,EAAE,QAAQ;gBACvF,MAAM,EAAE,IAAI,EAAE,OAAO,EAAE,GAAG,CAAC,OAAO,EAAE,eAAe,EAAE,GAAG,CAAC,OAAO;aACjE,CAAC,CAAC;SACJ,CAAC,CAAC,CAAC,EAAE,CAAC;QACP,SAAS,EAAE,KAAK,CAAC,SAAS;QAC1B,OAAO,EAAE,KAAK,CAAC,OAAO,CAAC,OAAO;QAC9B,YAAY,EAAE,KAAK,CAAC,OAAO,CAAC,YAAY;QACxC,kBAAkB,EAAE,KAAK,CAAC,OAAO,CAAC,kBAAkB;QACpD,UAAU,EAAE,KAAK,CAAC,QAAQ,CAAC,UAAU;QACrC,MAAM,EAAE,SAAS;QACjB,OAAO,EAAE,kBAAkB,CAAC,SAAS,CAAC;QACtC,GAAG,CAAC,KAAK,CAAC,gBAAgB,CAAC,CAAC,CAAC,EAAE,iBAAiB,EAAE,KAAK,CAAC,gBAAgB,EAAE,CAAC,CAAC,CAAC,EAAE,CAAC;QAChF,GAAG,0BAA0B,CAAC,KAAK,CAAC,mBAAmB,CAAC;QACxD,GAAG,CAAC,KAAK,CAAC,cAAc,CAAC,CAAC,CAAC,EAAE,sBAAsB,EAAE,gBAAgB,CAAC,KAAK,CAAC,cAAc,CAAC,EAAE,CAAC,CAAC,CAAC,EAAE,CAAC;QACnG,GAAG,CAAC,SAAS,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC;YACnB,KAAK,EAAE;gBACL,IAAI,EAAE,KAAK,CAAC,SAAS,CAAC,CAAC,CAAC,WAAW,CAAC,CAAC,CAAC,kBAAkB,CAAC,CAAC,CAAC,iCAAiC;oBAC1F,CAAC,CAAC,CAAC,iBAAiB,CAAC,CAAC,CAAC,eAAe,CAAC,CAAC,CAAC,sBAAsB,CAAC,MAAM,GAAG,CAAC;wBACxE,CAAC,CAAC,uBAAuB,CAAC,CAAC,CAAC,wBAAwB;gBACxD,OAAO,EAAE,KAAK,CAAC,SAAS,CAAC,CAAC,CAAC,oBAAoB;oBAC7C,CAAC,CAAC,kBAAkB,CAAC,CAAC,CAAC,6EAA6E;wBAClG,CAAC,CAAC,CAAC,iBAAiB,CAAC,CAAC,CAAC,+BAA+B,KAAK,CAAC,WAAW,CAAC,MAAM,IAAI,mBAAmB,EAAE;4BACrG,CAAC,CAAC,sBAAsB,CAAC,MAAM,GAAG,CAAC;gCACjC,CAAC,CAAC,GAAG,uBAAuB,KAAK,uCAAuC,CAAC,CAAC,CAAC,kCAAkC,CAAC,CAAC,CAAC,iCAAiC,KAAK,0BAA0B,CAAC,GAAG,CAAC,KAAK,CAAC,EAAE,CAAC,GAAG,KAAK,CAAC,OAAO,IAAI,KAAK,CAAC,OAAO,EAAE,CAAC,CAAC,IAAI,CAAC,IAAI,CAAC,EAAE;gCAC/O,CAAC,CAAC,uCAAuC,aAAa,CAAC,GAAG,CAAC,KAAK,CAAC,EAAE,CAAC,GAAG,KAAK,CAAC,OAAO,IAAI,KAAK,CAAC,OAAO,EAAE,CAAC,CAAC,IAAI,CAAC,IAAI,CAAC,EAAE;aAC9H;SACF,CAAC;QACF,UAAU,EAAE,KAAK,CAAC,SAAS,CAAC,WAAW,EAAE;QACzC,YAAY,EAAE,IAAI,IAAI,EAAE,CAAC,WAAW,EAAE;KACvC,CAAC;AACJ,CAAC"}
|
package/dist/src/evals/events.js
CHANGED
|
@@ -1,6 +1,8 @@
|
|
|
1
1
|
import { createWriteStream } from "node:fs";
|
|
2
|
+
import { open, stat } from "node:fs/promises";
|
|
2
3
|
import path from "node:path";
|
|
3
4
|
import { SCHEMA_VERSION, ensureDir } from "../foundation/index.js";
|
|
5
|
+
export const MAX_EVAL_EVENT_BYTES = 64 * 1024;
|
|
4
6
|
export class EvalEventSink {
|
|
5
7
|
path;
|
|
6
8
|
evalId;
|
|
@@ -16,23 +18,33 @@ export class EvalEventSink {
|
|
|
16
18
|
}
|
|
17
19
|
async open() {
|
|
18
20
|
await ensureDir(path.dirname(this.path));
|
|
21
|
+
this.sequence = await lastCommittedSequence(this.path);
|
|
19
22
|
this.stream = createWriteStream(this.path, { flags: "a", mode: 0o600 });
|
|
20
23
|
this.stream.on("error", (error) => { this.streamError ||= error; });
|
|
21
24
|
}
|
|
22
25
|
emit(event) {
|
|
23
|
-
const
|
|
26
|
+
const identity = {
|
|
24
27
|
schema_version: SCHEMA_VERSION,
|
|
25
28
|
sequence: ++this.sequence,
|
|
26
29
|
timestamp: new Date().toISOString(),
|
|
27
30
|
eval_id: this.evalId,
|
|
28
|
-
...event,
|
|
29
31
|
};
|
|
30
|
-
|
|
32
|
+
const framed = { ...identity, ...event };
|
|
33
|
+
const encoded = JSON.stringify(framed);
|
|
34
|
+
const persisted = Buffer.byteLength(encoded) <= MAX_EVAL_EVENT_BYTES
|
|
35
|
+
? framed
|
|
36
|
+
: {
|
|
37
|
+
...identity,
|
|
38
|
+
type: boundedType(event.type),
|
|
39
|
+
truncated: true,
|
|
40
|
+
original_bytes: Buffer.byteLength(encoded),
|
|
41
|
+
};
|
|
42
|
+
this.pending = this.pending.then(() => writeChunk(this.stream, `${JSON.stringify(persisted)}\n`));
|
|
31
43
|
try {
|
|
32
|
-
this.onEvent(
|
|
44
|
+
this.onEvent(persisted);
|
|
33
45
|
}
|
|
34
46
|
catch { /* Observers cannot break eval persistence. */ }
|
|
35
|
-
return
|
|
47
|
+
return persisted;
|
|
36
48
|
}
|
|
37
49
|
async close() {
|
|
38
50
|
let failure;
|
|
@@ -53,6 +65,42 @@ export class EvalEventSink {
|
|
|
53
65
|
throw failure;
|
|
54
66
|
}
|
|
55
67
|
}
|
|
68
|
+
function boundedType(value) {
|
|
69
|
+
const type = typeof value === "string" && value ? value : "eval.event.truncated";
|
|
70
|
+
return type.slice(0, 256);
|
|
71
|
+
}
|
|
72
|
+
async function lastCommittedSequence(file) {
|
|
73
|
+
let size;
|
|
74
|
+
try {
|
|
75
|
+
size = (await stat(file)).size;
|
|
76
|
+
}
|
|
77
|
+
catch (error) {
|
|
78
|
+
if (error.code === "ENOENT")
|
|
79
|
+
return 0;
|
|
80
|
+
throw error;
|
|
81
|
+
}
|
|
82
|
+
if (size === 0)
|
|
83
|
+
return 0;
|
|
84
|
+
const handle = await open(file, "r");
|
|
85
|
+
try {
|
|
86
|
+
const length = Math.min(size, 64 * 1024);
|
|
87
|
+
const buffer = Buffer.allocUnsafe(length);
|
|
88
|
+
const { bytesRead } = await handle.read(buffer, 0, length, size - length);
|
|
89
|
+
const committed = buffer.subarray(0, bytesRead).toString("utf8").replace(/[^\n]*$/, "");
|
|
90
|
+
const lines = committed.trimEnd().split("\n");
|
|
91
|
+
const last = lines.at(-1);
|
|
92
|
+
if (!last)
|
|
93
|
+
return 0;
|
|
94
|
+
const parsed = JSON.parse(last);
|
|
95
|
+
if (!Number.isSafeInteger(parsed.sequence) || parsed.sequence < 0) {
|
|
96
|
+
throw new TypeError("eval event sequence is invalid");
|
|
97
|
+
}
|
|
98
|
+
return parsed.sequence;
|
|
99
|
+
}
|
|
100
|
+
finally {
|
|
101
|
+
await handle.close();
|
|
102
|
+
}
|
|
103
|
+
}
|
|
56
104
|
function writeChunk(stream, chunk) {
|
|
57
105
|
return new Promise((resolve, reject) => stream.write(chunk, (error) => error ? reject(error) : resolve()));
|
|
58
106
|
}
|
|
@@ -1 +1 @@
|
|
|
1
|
-
{"version":3,"file":"events.js","sourceRoot":"","sources":["../../../src/evals/events.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,iBAAiB,EAAE,MAAM,SAAS,CAAC;AAE5C,OAAO,IAAI,MAAM,WAAW,CAAC;AAE7B,OAAO,EAAE,cAAc,EAAE,SAAS,EAAE,MAAM,wBAAwB,CAAC;AAEnE,MAAM,OAAO,aAAa;IACf,IAAI,CAAS;IACL,MAAM,CAAS;IACf,OAAO,CAA2C;IAC3D,QAAQ,GAAG,CAAC,CAAC;IACb,OAAO,GAAkB,OAAO,CAAC,OAAO,EAAE,CAAC;IAC3C,MAAM,CAA0B;IAChC,WAAW,CAAoB;IAEvC,YAAY,aAAqB,EAAE,MAAc,EAAE,UAAoD,GAAG,EAAE,GAAE,CAAC;QAC7G,IAAI,CAAC,IAAI,GAAG,IAAI,CAAC,IAAI,CAAC,aAAa,EAAE,cAAc,CAAC,CAAC;QACrD,IAAI,CAAC,MAAM,GAAG,MAAM,CAAC;QACrB,IAAI,CAAC,OAAO,GAAG,OAAO,IAAI,CAAC,GAAG,EAAE,GAAE,CAAC,CAAC,CAAC;IACvC,CAAC;IAED,KAAK,CAAC,IAAI;QACR,MAAM,SAAS,CAAC,IAAI,CAAC,OAAO,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC,CAAC;QACzC,IAAI,CAAC,MAAM,GAAG,iBAAiB,CAAC,IAAI,CAAC,IAAI,EAAE,EAAE,KAAK,EAAE,GAAG,EAAE,IAAI,EAAE,KAAK,EAAE,CAAC,CAAC;QACxE,IAAI,CAAC,MAAM,CAAC,EAAE,CAAC,OAAO,EAAE,CAAC,KAAY,EAAE,EAAE,GAAG,IAAI,CAAC,WAAW,KAAK,KAAK,CAAC,CAAC,CAAC,CAAC,CAAC;IAC7E,CAAC;IAED,IAAI,CAAC,KAA8B;QACjC,MAAM,
|
|
1
|
+
{"version":3,"file":"events.js","sourceRoot":"","sources":["../../../src/evals/events.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,iBAAiB,EAAE,MAAM,SAAS,CAAC;AAE5C,OAAO,EAAE,IAAI,EAAE,IAAI,EAAE,MAAM,kBAAkB,CAAC;AAC9C,OAAO,IAAI,MAAM,WAAW,CAAC;AAE7B,OAAO,EAAE,cAAc,EAAE,SAAS,EAAE,MAAM,wBAAwB,CAAC;AAEnE,MAAM,CAAC,MAAM,oBAAoB,GAAG,EAAE,GAAG,IAAI,CAAC;AAE9C,MAAM,OAAO,aAAa;IACf,IAAI,CAAS;IACL,MAAM,CAAS;IACf,OAAO,CAA2C;IAC3D,QAAQ,GAAG,CAAC,CAAC;IACb,OAAO,GAAkB,OAAO,CAAC,OAAO,EAAE,CAAC;IAC3C,MAAM,CAA0B;IAChC,WAAW,CAAoB;IAEvC,YAAY,aAAqB,EAAE,MAAc,EAAE,UAAoD,GAAG,EAAE,GAAE,CAAC;QAC7G,IAAI,CAAC,IAAI,GAAG,IAAI,CAAC,IAAI,CAAC,aAAa,EAAE,cAAc,CAAC,CAAC;QACrD,IAAI,CAAC,MAAM,GAAG,MAAM,CAAC;QACrB,IAAI,CAAC,OAAO,GAAG,OAAO,IAAI,CAAC,GAAG,EAAE,GAAE,CAAC,CAAC,CAAC;IACvC,CAAC;IAED,KAAK,CAAC,IAAI;QACR,MAAM,SAAS,CAAC,IAAI,CAAC,OAAO,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC,CAAC;QACzC,IAAI,CAAC,QAAQ,GAAG,MAAM,qBAAqB,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC;QACvD,IAAI,CAAC,MAAM,GAAG,iBAAiB,CAAC,IAAI,CAAC,IAAI,EAAE,EAAE,KAAK,EAAE,GAAG,EAAE,IAAI,EAAE,KAAK,EAAE,CAAC,CAAC;QACxE,IAAI,CAAC,MAAM,CAAC,EAAE,CAAC,OAAO,EAAE,CAAC,KAAY,EAAE,EAAE,GAAG,IAAI,CAAC,WAAW,KAAK,KAAK,CAAC,CAAC,CAAC,CAAC,CAAC;IAC7E,CAAC;IAED,IAAI,CAAC,KAA8B;QACjC,MAAM,QAAQ,GAAG;YACf,cAAc,EAAE,cAAc;YAC9B,QAAQ,EAAE,EAAE,IAAI,CAAC,QAAQ;YACzB,SAAS,EAAE,IAAI,IAAI,EAAE,CAAC,WAAW,EAAE;YACnC,OAAO,EAAE,IAAI,CAAC,MAAM;SACrB,CAAC;QACF,MAAM,MAAM,GAAG,EAAE,GAAG,QAAQ,EAAE,GAAG,KAAK,EAAE,CAAC;QACzC,MAAM,OAAO,GAAG,IAAI,CAAC,SAAS,CAAC,MAAM,CAAC,CAAC;QACvC,MAAM,SAAS,GAAG,MAAM,CAAC,UAAU,CAAC,OAAO,CAAC,IAAI,oBAAoB;YAClE,CAAC,CAAC,MAAM;YACR,CAAC,CAAC;gBACE,GAAG,QAAQ;gBACX,IAAI,EAAE,WAAW,CAAC,KAAK,CAAC,IAAI,CAAC;gBAC7B,SAAS,EAAE,IAAI;gBACf,cAAc,EAAE,MAAM,CAAC,UAAU,CAAC,OAAO,CAAC;aAC3C,CAAC;QACN,IAAI,CAAC,OAAO,GAAG,IAAI,CAAC,OAAO,CAAC,IAAI,CAAC,GAAG,EAAE,CAAC,UAAU,CAAC,IAAI,CAAC,MAAqB,EAAE,GAAG,IAAI,CAAC,SAAS,CAAC,SAAS,CAAC,IAAI,CAAC,CAAC,CAAC;QACjH,IAAI,CAAC;YAAC,IAAI,CAAC,OAAO,CAAC,SAAS,CAAC,CAAC;QAAC,CAAC;QAAC,MAAM,CAAC,CAAC,8CAA8C,CAAC,CAAC;QACzF,OAAO,SAAS,CAAC;IACnB,CAAC;IAED,KAAK,CAAC,KAAK;QACT,IAAI,OAA0B,CAAC;QAC/B,IAAI,CAAC;YAAC,MAAM,IAAI,CAAC,OAAO,CAAC;QAAC,CAAC;QAAC,OAAO,KAAK,EAAE,CAAC;YAAC,OAAO,GAAG,KAAc,CAAC;QAAC,CAAC;QACvE,IAAI,CAAC;YAAC,MAAM,WAAW,CAAC,IAAI,CAAC,MAAM,CAAC,CAAC;QAAC,CAAC;QAAC,OAAO,KAAK,EAAE,CAAC;YAAC,OAAO,KAAK,KAAc,CAAC;QAAC,CAAC;QACrF,OAAO,KAAK,IAAI,CAAC,WAAW,CAAC;QAC7B,IAAI,OAAO;YAAE,MAAM,OAAO,CAAC;IAC7B,CAAC;CACF;AAED,SAAS,WAAW,CAAC,KAAc;IACjC,MAAM,IAAI,GAAG,OAAO,KAAK,KAAK,QAAQ,IAAI,KAAK,CAAC,CAAC,CAAC,KAAK,CAAC,CAAC,CAAC,sBAAsB,CAAC;IACjF,OAAO,IAAI,CAAC,KAAK,CAAC,CAAC,EAAE,GAAG,CAAC,CAAC;AAC5B,CAAC;AAED,KAAK,UAAU,qBAAqB,CAAC,IAAY;IAC/C,IAAI,IAAY,CAAC;IACjB,IAAI,CAAC;QACH,IAAI,GAAG,CAAC,MAAM,IAAI,CAAC,IAAI,CAAC,CAAC,CAAC,IAAI,CAAC;IACjC,CAAC;IAAC,OAAO,KAAK,EAAE,CAAC;QACf,IAAK,KAA+B,CAAC,IAAI,KAAK,QAAQ;YAAE,OAAO,CAAC,CAAC;QACjE,MAAM,KAAK,CAAC;IACd,CAAC;IACD,IAAI,IAAI,KAAK,CAAC;QAAE,OAAO,CAAC,CAAC;IACzB,MAAM,MAAM,GAAG,MAAM,IAAI,CAAC,IAAI,EAAE,GAAG,CAAC,CAAC;IACrC,IAAI,CAAC;QACH,MAAM,MAAM,GAAG,IAAI,CAAC,GAAG,CAAC,IAAI,EAAE,EAAE,GAAG,IAAI,CAAC,CAAC;QACzC,MAAM,MAAM,GAAG,MAAM,CAAC,WAAW,CAAC,MAAM,CAAC,CAAC;QAC1C,MAAM,EAAE,SAAS,EAAE,GAAG,MAAM,MAAM,CAAC,IAAI,CAAC,MAAM,EAAE,CAAC,EAAE,MAAM,EAAE,IAAI,GAAG,MAAM,CAAC,CAAC;QAC1E,MAAM,SAAS,GAAG,MAAM,CAAC,QAAQ,CAAC,CAAC,EAAE,SAAS,CAAC,CAAC,QAAQ,CAAC,MAAM,CAAC,CAAC,OAAO,CAAC,SAAS,EAAE,EAAE,CAAC,CAAC;QACxF,MAAM,KAAK,GAAG,SAAS,CAAC,OAAO,EAAE,CAAC,KAAK,CAAC,IAAI,CAAC,CAAC;QAC9C,MAAM,IAAI,GAAG,KAAK,CAAC,EAAE,CAAC,CAAC,CAAC,CAAC,CAAC;QAC1B,IAAI,CAAC,IAAI;YAAE,OAAO,CAAC,CAAC;QACpB,MAAM,MAAM,GAAG,IAAI,CAAC,KAAK,CAAC,IAAI,CAA2B,CAAC;QAC1D,IAAI,CAAC,MAAM,CAAC,aAAa,CAAC,MAAM,CAAC,QAAQ,CAAC,IAAK,MAAM,CAAC,QAAmB,GAAG,CAAC,EAAE,CAAC;YAC9E,MAAM,IAAI,SAAS,CAAC,gCAAgC,CAAC,CAAC;QACxD,CAAC;QACD,OAAO,MAAM,CAAC,QAAkB,CAAC;IACnC,CAAC;YAAS,CAAC;QACT,MAAM,MAAM,CAAC,KAAK,EAAE,CAAC;IACvB,CAAC;AACH,CAAC;AAED,SAAS,UAAU,CAAC,MAAmB,EAAE,KAAa;IACpD,OAAO,IAAI,OAAO,CAAC,CAAC,OAAO,EAAE,MAAM,EAAE,EAAE,CAAC,MAAM,CAAC,KAAK,CAAC,KAAK,EAAE,CAAC,KAAK,EAAE,EAAE,CAAC,KAAK,CAAC,CAAC,CAAC,MAAM,CAAC,KAAK,CAAC,CAAC,CAAC,CAAC,OAAO,EAAE,CAAC,CAAC,CAAC;AAC7G,CAAC;AAED,SAAS,WAAW,CAAC,MAA+B;IAClD,IAAI,CAAC,MAAM;QAAE,OAAO,OAAO,CAAC,OAAO,EAAE,CAAC;IACtC,OAAO,IAAI,OAAO,CAAC,CAAC,OAAO,EAAE,MAAM,EAAE,EAAE;QACrC,MAAM,CAAC,IAAI,CAAC,OAAO,EAAE,MAAM,CAAC,CAAC;QAC7B,MAAM,CAAC,GAAG,CAAC,OAAO,CAAC,CAAC;IACtB,CAAC,CAAC,CAAC;AACL,CAAC"}
|
|
@@ -0,0 +1,191 @@
|
|
|
1
|
+
import path from "node:path";
|
|
2
|
+
import { setTimeout as delay } from "node:timers/promises";
|
|
3
|
+
import { atomicWriteJSON, ensureDir, readJSON, sha256JSON, withFileLock } from "../foundation/index.js";
|
|
4
|
+
import { HitchError } from "../foundation/index.js";
|
|
5
|
+
const localFlights = new Map();
|
|
6
|
+
/** Persisted lineage singleflight. Only the caller that creates `submitting`
|
|
7
|
+
* may invoke the materializer; later rounds validate and return `ready`. */
|
|
8
|
+
export async function ensureEvolutionBaseline(options) {
|
|
9
|
+
validateInputs(options);
|
|
10
|
+
const file = baselinePath(options.root, options.evolutionId);
|
|
11
|
+
const local = localFlights.get(file);
|
|
12
|
+
if (local)
|
|
13
|
+
return local;
|
|
14
|
+
const operation = ensureInternal(options, file);
|
|
15
|
+
localFlights.set(file, operation);
|
|
16
|
+
try {
|
|
17
|
+
return await operation;
|
|
18
|
+
}
|
|
19
|
+
finally {
|
|
20
|
+
if (localFlights.get(file) === operation)
|
|
21
|
+
localFlights.delete(file);
|
|
22
|
+
}
|
|
23
|
+
}
|
|
24
|
+
export function evolutionBaselineFingerprint(value) {
|
|
25
|
+
if (!value || typeof value !== "object" || Array.isArray(value) || Object.keys(value).length === 0) {
|
|
26
|
+
throw new TypeError("evolution baseline fingerprint input is invalid");
|
|
27
|
+
}
|
|
28
|
+
return sha256JSON(value);
|
|
29
|
+
}
|
|
30
|
+
export async function readEvolutionBaseline(root, evolutionId) {
|
|
31
|
+
const value = await readJSON(baselinePath(root, evolutionId), null);
|
|
32
|
+
return value === null ? null : parseEvolutionBaseline(value, evolutionId);
|
|
33
|
+
}
|
|
34
|
+
export function parseEvolutionBaseline(value, expectedEvolutionId) {
|
|
35
|
+
if (!value || typeof value !== "object" || Array.isArray(value))
|
|
36
|
+
throw new TypeError("evolution baseline must be an object");
|
|
37
|
+
const record = value;
|
|
38
|
+
const fields = ["schema_version", "evolution_id", "baseline_fingerprint", "state", "seed_eval_id", "heldout_eval_id", "seed_result_digest", "heldout_result_digest", "created_at", "updated_at"];
|
|
39
|
+
if (Object.keys(record).some((key) => !fields.includes(key)) || record.schema_version !== "1"
|
|
40
|
+
|| typeof record.evolution_id !== "string" || !record.evolution_id
|
|
41
|
+
|| expectedEvolutionId !== undefined && record.evolution_id !== expectedEvolutionId
|
|
42
|
+
|| typeof record.baseline_fingerprint !== "string" || !/^sha256:[a-f0-9]{64}$/.test(record.baseline_fingerprint)
|
|
43
|
+
|| !["submitting", "running", "ready", "failed"].includes(record.state)
|
|
44
|
+
|| typeof record.created_at !== "string" || !Number.isFinite(Date.parse(record.created_at))
|
|
45
|
+
|| typeof record.updated_at !== "string" || !Number.isFinite(Date.parse(record.updated_at)))
|
|
46
|
+
throw new TypeError("evolution baseline identity is invalid");
|
|
47
|
+
const optional = ["seed_eval_id", "heldout_eval_id"];
|
|
48
|
+
for (const name of optional)
|
|
49
|
+
if (record[name] !== undefined && (typeof record[name] !== "string" || !/^eval_[a-f0-9]{32}$/.test(record[name]))) {
|
|
50
|
+
throw new TypeError(`evolution baseline ${name} is invalid`);
|
|
51
|
+
}
|
|
52
|
+
for (const name of ["seed_result_digest", "heldout_result_digest"])
|
|
53
|
+
if (record[name] !== undefined
|
|
54
|
+
&& (typeof record[name] !== "string" || !/^sha256:[a-f0-9]{64}$/.test(record[name])))
|
|
55
|
+
throw new TypeError(`evolution baseline ${name} is invalid`);
|
|
56
|
+
if (record.state === "running" && (!record.seed_eval_id || !record.heldout_eval_id))
|
|
57
|
+
throw new TypeError("running evolution baseline eval identities are missing");
|
|
58
|
+
if (record.state === "ready" && (!record.seed_eval_id || !record.heldout_eval_id || !record.seed_result_digest || !record.heldout_result_digest)) {
|
|
59
|
+
throw new TypeError("ready evolution baseline references are incomplete");
|
|
60
|
+
}
|
|
61
|
+
return record;
|
|
62
|
+
}
|
|
63
|
+
async function ensureInternal(options, file) {
|
|
64
|
+
const ownership = await claim(options, file);
|
|
65
|
+
if (!ownership.owner)
|
|
66
|
+
return waitForOwner(options, file, ownership.baseline);
|
|
67
|
+
try {
|
|
68
|
+
const materialized = await options.materialize({
|
|
69
|
+
markRunning: (seedEvalId, heldoutEvalId) => updateOwned(options, file, (current) => ({
|
|
70
|
+
...assertRunningIdentity(current, seedEvalId, heldoutEvalId), state: "running",
|
|
71
|
+
})).then(() => undefined),
|
|
72
|
+
});
|
|
73
|
+
const ready = await updateOwned(options, file, (current) => {
|
|
74
|
+
if (current.seed_eval_id && current.seed_eval_id !== materialized.seed_eval_id
|
|
75
|
+
|| current.heldout_eval_id && current.heldout_eval_id !== materialized.heldout_eval_id) {
|
|
76
|
+
throw new TypeError("evolution baseline eval identity changed during materialization");
|
|
77
|
+
}
|
|
78
|
+
return {
|
|
79
|
+
...current, state: "ready", seed_eval_id: checkedEvalId(materialized.seed_eval_id), heldout_eval_id: checkedEvalId(materialized.heldout_eval_id),
|
|
80
|
+
seed_result_digest: checkedDigest(materialized.seed_result_digest), heldout_result_digest: checkedDigest(materialized.heldout_result_digest),
|
|
81
|
+
};
|
|
82
|
+
});
|
|
83
|
+
await options.validateReady?.(ready);
|
|
84
|
+
return ready;
|
|
85
|
+
}
|
|
86
|
+
catch (error) {
|
|
87
|
+
await updateOwned(options, file, (current) => ({ ...current, state: "failed" })).catch(() => undefined);
|
|
88
|
+
throw error;
|
|
89
|
+
}
|
|
90
|
+
}
|
|
91
|
+
async function claim(options, file) {
|
|
92
|
+
await ensureDir(path.dirname(file));
|
|
93
|
+
return withBaselineLock(options, async () => {
|
|
94
|
+
const existing = await readEvolutionBaseline(options.root, options.evolutionId);
|
|
95
|
+
if (existing) {
|
|
96
|
+
assertFingerprint(existing, options.fingerprint);
|
|
97
|
+
if (existing.state === "failed")
|
|
98
|
+
throw baselineError("evolution baseline materialization previously failed", "evolution_baseline_failed");
|
|
99
|
+
if (existing.state === "ready")
|
|
100
|
+
await options.validateReady?.(existing);
|
|
101
|
+
return { owner: false, baseline: existing };
|
|
102
|
+
}
|
|
103
|
+
const now = new Date().toISOString();
|
|
104
|
+
const baseline = parseEvolutionBaseline({
|
|
105
|
+
schema_version: "1", evolution_id: options.evolutionId, baseline_fingerprint: options.fingerprint,
|
|
106
|
+
state: "submitting", created_at: now, updated_at: now,
|
|
107
|
+
}, options.evolutionId);
|
|
108
|
+
await atomicWriteJSON(file, baseline);
|
|
109
|
+
return { owner: true, baseline };
|
|
110
|
+
});
|
|
111
|
+
}
|
|
112
|
+
async function waitForOwner(options, file, initial) {
|
|
113
|
+
if (initial.state === "ready")
|
|
114
|
+
return initial;
|
|
115
|
+
const deadline = Date.now() + (options.maxWaitMs ?? 24 * 60 * 60 * 1_000);
|
|
116
|
+
const poll = options.pollIntervalMs ?? 250;
|
|
117
|
+
for (;;) {
|
|
118
|
+
if (options.signal?.aborted)
|
|
119
|
+
throw options.signal.reason;
|
|
120
|
+
if (Date.now() >= deadline)
|
|
121
|
+
throw baselineError("evolution baseline materialization is still in progress", "evolution_baseline_in_progress");
|
|
122
|
+
await delay(Math.min(poll, Math.max(1, deadline - Date.now())), undefined, options.signal ? { signal: options.signal } : undefined);
|
|
123
|
+
const current = await readEvolutionBaseline(options.root, options.evolutionId);
|
|
124
|
+
if (!current)
|
|
125
|
+
throw baselineError("evolution baseline record disappeared", "evolution_baseline_state_conflict");
|
|
126
|
+
assertFingerprint(current, options.fingerprint);
|
|
127
|
+
if (current.state === "failed")
|
|
128
|
+
throw baselineError("evolution baseline materialization failed", "evolution_baseline_failed");
|
|
129
|
+
if (current.state === "ready") {
|
|
130
|
+
await options.validateReady?.(current);
|
|
131
|
+
return current;
|
|
132
|
+
}
|
|
133
|
+
}
|
|
134
|
+
}
|
|
135
|
+
async function updateOwned(options, file, update) {
|
|
136
|
+
return withBaselineLock(options, async () => {
|
|
137
|
+
const current = await readEvolutionBaseline(options.root, options.evolutionId);
|
|
138
|
+
if (!current)
|
|
139
|
+
throw baselineError("evolution baseline ownership was lost", "evolution_baseline_state_conflict");
|
|
140
|
+
assertFingerprint(current, options.fingerprint);
|
|
141
|
+
if (current.state === "ready" || current.state === "failed")
|
|
142
|
+
throw baselineError("evolution baseline is already terminal", "evolution_baseline_state_conflict");
|
|
143
|
+
const next = parseEvolutionBaseline({ ...update(current), updated_at: new Date().toISOString() }, options.evolutionId);
|
|
144
|
+
await atomicWriteJSON(file, next);
|
|
145
|
+
return next;
|
|
146
|
+
});
|
|
147
|
+
}
|
|
148
|
+
function withBaselineLock(options, operation) {
|
|
149
|
+
const key = sha256JSON({ evolution_id: options.evolutionId }).slice("sha256:".length);
|
|
150
|
+
return withFileLock(path.join(options.root, "locks", "evolution-baseline"), key, operation, {
|
|
151
|
+
timeoutCode: "evolution_baseline_locked", timeoutExitCode: 12,
|
|
152
|
+
});
|
|
153
|
+
}
|
|
154
|
+
function validateInputs(options) {
|
|
155
|
+
if (!options.root || !options.evolutionId || !/^sha256:[a-f0-9]{64}$/.test(options.fingerprint))
|
|
156
|
+
throw new TypeError("evolution baseline request is invalid");
|
|
157
|
+
if (options.pollIntervalMs !== undefined && (!Number.isSafeInteger(options.pollIntervalMs) || options.pollIntervalMs < 1))
|
|
158
|
+
throw new TypeError("evolution baseline poll interval is invalid");
|
|
159
|
+
if (options.maxWaitMs !== undefined && (!Number.isSafeInteger(options.maxWaitMs) || options.maxWaitMs < 1))
|
|
160
|
+
throw new TypeError("evolution baseline wait is invalid");
|
|
161
|
+
}
|
|
162
|
+
function baselinePath(root, evolutionId) {
|
|
163
|
+
const id = sha256JSON({ evolution_id: evolutionId }).slice("sha256:".length);
|
|
164
|
+
return path.join(root, "evolutions", id, "baseline.json");
|
|
165
|
+
}
|
|
166
|
+
function assertFingerprint(record, fingerprint) {
|
|
167
|
+
if (record.baseline_fingerprint !== fingerprint)
|
|
168
|
+
throw baselineError("evolution baseline fingerprint changed", "evolution_baseline_fingerprint_mismatch");
|
|
169
|
+
}
|
|
170
|
+
function checkedEvalId(value) {
|
|
171
|
+
if (!/^eval_[a-f0-9]{32}$/.test(value))
|
|
172
|
+
throw new TypeError("evolution baseline eval id is invalid");
|
|
173
|
+
return value;
|
|
174
|
+
}
|
|
175
|
+
function checkedDigest(value) {
|
|
176
|
+
if (!/^sha256:[a-f0-9]{64}$/.test(value))
|
|
177
|
+
throw new TypeError("evolution baseline result digest is invalid");
|
|
178
|
+
return value;
|
|
179
|
+
}
|
|
180
|
+
function assertRunningIdentity(current, seedEvalId, heldoutEvalId) {
|
|
181
|
+
const seed = checkedEvalId(seedEvalId);
|
|
182
|
+
const heldout = checkedEvalId(heldoutEvalId);
|
|
183
|
+
if (current.seed_eval_id && current.seed_eval_id !== seed || current.heldout_eval_id && current.heldout_eval_id !== heldout) {
|
|
184
|
+
throw new TypeError("evolution baseline eval identity changed while running");
|
|
185
|
+
}
|
|
186
|
+
return { ...current, seed_eval_id: seed, heldout_eval_id: heldout };
|
|
187
|
+
}
|
|
188
|
+
function baselineError(message, code) {
|
|
189
|
+
return new HitchError(message, { code, exitCode: 12 });
|
|
190
|
+
}
|
|
191
|
+
//# sourceMappingURL=evolution-baseline.js.map
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"evolution-baseline.js","sourceRoot":"","sources":["../../../src/evals/evolution-baseline.ts"],"names":[],"mappings":"AAAA,OAAO,IAAI,MAAM,WAAW,CAAC;AAC7B,OAAO,EAAE,UAAU,IAAI,KAAK,EAAE,MAAM,sBAAsB,CAAC;AAC3D,OAAO,EAAE,eAAe,EAAE,SAAS,EAAE,QAAQ,EAAE,UAAU,EAAE,YAAY,EAAE,MAAM,wBAAwB,CAAC;AACxG,OAAO,EAAE,UAAU,EAAE,MAAM,wBAAwB,CAAC;AAqCpD,MAAM,YAAY,GAAG,IAAI,GAAG,EAAwC,CAAC;AAErE;4EAC4E;AAC5E,MAAM,CAAC,KAAK,UAAU,uBAAuB,CAAC,OAAuC;IACnF,cAAc,CAAC,OAAO,CAAC,CAAC;IACxB,MAAM,IAAI,GAAG,YAAY,CAAC,OAAO,CAAC,IAAI,EAAE,OAAO,CAAC,WAAW,CAAC,CAAC;IAC7D,MAAM,KAAK,GAAG,YAAY,CAAC,GAAG,CAAC,IAAI,CAAC,CAAC;IACrC,IAAI,KAAK;QAAE,OAAO,KAAK,CAAC;IACxB,MAAM,SAAS,GAAG,cAAc,CAAC,OAAO,EAAE,IAAI,CAAC,CAAC;IAChD,YAAY,CAAC,GAAG,CAAC,IAAI,EAAE,SAAS,CAAC,CAAC;IAClC,IAAI,CAAC;QAAC,OAAO,MAAM,SAAS,CAAC;IAAC,CAAC;YAAS,CAAC;QAAC,IAAI,YAAY,CAAC,GAAG,CAAC,IAAI,CAAC,KAAK,SAAS;YAAE,YAAY,CAAC,MAAM,CAAC,IAAI,CAAC,CAAC;IAAC,CAAC;AAClH,CAAC;AAED,MAAM,UAAU,4BAA4B,CAAC,KAA8B;IACzE,IAAI,CAAC,KAAK,IAAI,OAAO,KAAK,KAAK,QAAQ,IAAI,KAAK,CAAC,OAAO,CAAC,KAAK,CAAC,IAAI,MAAM,CAAC,IAAI,CAAC,KAAK,CAAC,CAAC,MAAM,KAAK,CAAC,EAAE,CAAC;QACnG,MAAM,IAAI,SAAS,CAAC,iDAAiD,CAAC,CAAC;IACzE,CAAC;IACD,OAAO,UAAU,CAAC,KAAK,CAAC,CAAC;AAC3B,CAAC;AAED,MAAM,CAAC,KAAK,UAAU,qBAAqB,CAAC,IAAY,EAAE,WAAmB;IAC3E,MAAM,KAAK,GAAG,MAAM,QAAQ,CAAiB,YAAY,CAAC,IAAI,EAAE,WAAW,CAAC,EAAE,IAAI,CAAC,CAAC;IACpF,OAAO,KAAK,KAAK,IAAI,CAAC,CAAC,CAAC,IAAI,CAAC,CAAC,CAAC,sBAAsB,CAAC,KAAK,EAAE,WAAW,CAAC,CAAC;AAC5E,CAAC;AAED,MAAM,UAAU,sBAAsB,CAAC,KAAc,EAAE,mBAA4B;IACjF,IAAI,CAAC,KAAK,IAAI,OAAO,KAAK,KAAK,QAAQ,IAAI,KAAK,CAAC,OAAO,CAAC,KAAK,CAAC;QAAE,MAAM,IAAI,SAAS,CAAC,sCAAsC,CAAC,CAAC;IAC7H,MAAM,MAAM,GAAG,KAAgC,CAAC;IAChD,MAAM,MAAM,GAAG,CAAC,gBAAgB,EAAE,cAAc,EAAE,sBAAsB,EAAE,OAAO,EAAE,cAAc,EAAE,iBAAiB,EAAE,oBAAoB,EAAE,uBAAuB,EAAE,YAAY,EAAE,YAAY,CAAC,CAAC;IACjM,IAAI,MAAM,CAAC,IAAI,CAAC,MAAM,CAAC,CAAC,IAAI,CAAC,CAAC,GAAG,EAAE,EAAE,CAAC,CAAC,MAAM,CAAC,QAAQ,CAAC,GAAG,CAAC,CAAC,IAAI,MAAM,CAAC,cAAc,KAAK,GAAG;WACxF,OAAO,MAAM,CAAC,YAAY,KAAK,QAAQ,IAAI,CAAC,MAAM,CAAC,YAAY;WAC/D,mBAAmB,KAAK,SAAS,IAAI,MAAM,CAAC,YAAY,KAAK,mBAAmB;WAChF,OAAO,MAAM,CAAC,oBAAoB,KAAK,QAAQ,IAAI,CAAC,uBAAuB,CAAC,IAAI,CAAC,MAAM,CAAC,oBAAoB,CAAC;WAC7G,CAAC,CAAC,YAAY,EAAE,SAAS,EAAE,OAAO,EAAE,QAAQ,CAAC,CAAC,QAAQ,CAAC,MAAM,CAAC,KAAe,CAAC;WAC9E,OAAO,MAAM,CAAC,UAAU,KAAK,QAAQ,IAAI,CAAC,MAAM,CAAC,QAAQ,CAAC,IAAI,CAAC,KAAK,CAAC,MAAM,CAAC,UAAU,CAAC,CAAC;WACxF,OAAO,MAAM,CAAC,UAAU,KAAK,QAAQ,IAAI,CAAC,MAAM,CAAC,QAAQ,CAAC,IAAI,CAAC,KAAK,CAAC,MAAM,CAAC,UAAU,CAAC,CAAC;QAAE,MAAM,IAAI,SAAS,CAAC,wCAAwC,CAAC,CAAC;IAC7J,MAAM,QAAQ,GAAG,CAAC,cAAc,EAAE,iBAAiB,CAAU,CAAC;IAC9D,KAAK,MAAM,IAAI,IAAI,QAAQ;QAAE,IAAI,MAAM,CAAC,IAAI,CAAC,KAAK,SAAS,IAAI,CAAC,OAAO,MAAM,CAAC,IAAI,CAAC,KAAK,QAAQ,IAAI,CAAC,qBAAqB,CAAC,IAAI,CAAC,MAAM,CAAC,IAAI,CAAW,CAAC,CAAC,EAAE,CAAC;YACzJ,MAAM,IAAI,SAAS,CAAC,sBAAsB,IAAI,aAAa,CAAC,CAAC;QAC/D,CAAC;IACD,KAAK,MAAM,IAAI,IAAI,CAAC,oBAAoB,EAAE,uBAAuB,CAAU;QAAE,IAAI,MAAM,CAAC,IAAI,CAAC,KAAK,SAAS;eACtG,CAAC,OAAO,MAAM,CAAC,IAAI,CAAC,KAAK,QAAQ,IAAI,CAAC,uBAAuB,CAAC,IAAI,CAAC,MAAM,CAAC,IAAI,CAAW,CAAC,CAAC;YAAE,MAAM,IAAI,SAAS,CAAC,sBAAsB,IAAI,aAAa,CAAC,CAAC;IAC/J,IAAI,MAAM,CAAC,KAAK,KAAK,SAAS,IAAI,CAAC,CAAC,MAAM,CAAC,YAAY,IAAI,CAAC,MAAM,CAAC,eAAe,CAAC;QAAE,MAAM,IAAI,SAAS,CAAC,wDAAwD,CAAC,CAAC;IACnK,IAAI,MAAM,CAAC,KAAK,KAAK,OAAO,IAAI,CAAC,CAAC,MAAM,CAAC,YAAY,IAAI,CAAC,MAAM,CAAC,eAAe,IAAI,CAAC,MAAM,CAAC,kBAAkB,IAAI,CAAC,MAAM,CAAC,qBAAqB,CAAC,EAAE,CAAC;QACjJ,MAAM,IAAI,SAAS,CAAC,oDAAoD,CAAC,CAAC;IAC5E,CAAC;IACD,OAAO,MAAwC,CAAC;AAClD,CAAC;AAED,KAAK,UAAU,cAAc,CAAC,OAAuC,EAAE,IAAY;IACjF,MAAM,SAAS,GAAG,MAAM,KAAK,CAAC,OAAO,EAAE,IAAI,CAAC,CAAC;IAC7C,IAAI,CAAC,SAAS,CAAC,KAAK;QAAE,OAAO,YAAY,CAAC,OAAO,EAAE,IAAI,EAAE,SAAS,CAAC,QAAQ,CAAC,CAAC;IAC7E,IAAI,CAAC;QACH,MAAM,YAAY,GAAG,MAAM,OAAO,CAAC,WAAW,CAAC;YAC7C,WAAW,EAAE,CAAC,UAAU,EAAE,aAAa,EAAE,EAAE,CAAC,WAAW,CAAC,OAAO,EAAE,IAAI,EAAE,CAAC,OAAO,EAAE,EAAE,CAAC,CAAC;gBACnF,GAAG,qBAAqB,CAAC,OAAO,EAAE,UAAU,EAAE,aAAa,CAAC,EAAE,KAAK,EAAE,SAAS;aAC/E,CAAC,CAAC,CAAC,IAAI,CAAC,GAAG,EAAE,CAAC,SAAS,CAAC;SAC1B,CAAC,CAAC;QACH,MAAM,KAAK,GAAG,MAAM,WAAW,CAAC,OAAO,EAAE,IAAI,EAAE,CAAC,OAAO,EAAE,EAAE;YACzD,IAAI,OAAO,CAAC,YAAY,IAAI,OAAO,CAAC,YAAY,KAAK,YAAY,CAAC,YAAY;mBACzE,OAAO,CAAC,eAAe,IAAI,OAAO,CAAC,eAAe,KAAK,YAAY,CAAC,eAAe,EAAE,CAAC;gBACzF,MAAM,IAAI,SAAS,CAAC,iEAAiE,CAAC,CAAC;YACzF,CAAC;YACD,OAAO;gBACL,GAAG,OAAO,EAAE,KAAK,EAAE,OAAO,EAAE,YAAY,EAAE,aAAa,CAAC,YAAY,CAAC,YAAY,CAAC,EAAE,eAAe,EAAE,aAAa,CAAC,YAAY,CAAC,eAAe,CAAC;gBAChJ,kBAAkB,EAAE,aAAa,CAAC,YAAY,CAAC,kBAAkB,CAAC,EAAE,qBAAqB,EAAE,aAAa,CAAC,YAAY,CAAC,qBAAqB,CAAC;aAC7I,CAAC;QACJ,CAAC,CAAC,CAAC;QACH,MAAM,OAAO,CAAC,aAAa,EAAE,CAAC,KAAK,CAAC,CAAC;QACrC,OAAO,KAAK,CAAC;IACf,CAAC;IAAC,OAAO,KAAK,EAAE,CAAC;QACf,MAAM,WAAW,CAAC,OAAO,EAAE,IAAI,EAAE,CAAC,OAAO,EAAE,EAAE,CAAC,CAAC,EAAE,GAAG,OAAO,EAAE,KAAK,EAAE,QAAQ,EAAE,CAAC,CAAC,CAAC,KAAK,CAAC,GAAG,EAAE,CAAC,SAAS,CAAC,CAAC;QACxG,MAAM,KAAK,CAAC;IACd,CAAC;AACH,CAAC;AAED,KAAK,UAAU,KAAK,CAAC,OAAuC,EAAE,IAAY;IACxE,MAAM,SAAS,CAAC,IAAI,CAAC,OAAO,CAAC,IAAI,CAAC,CAAC,CAAC;IACpC,OAAO,gBAAgB,CAAC,OAAO,EAAE,KAAK,IAAI,EAAE;QAC1C,MAAM,QAAQ,GAAG,MAAM,qBAAqB,CAAC,OAAO,CAAC,IAAI,EAAE,OAAO,CAAC,WAAW,CAAC,CAAC;QAChF,IAAI,QAAQ,EAAE,CAAC;YACb,iBAAiB,CAAC,QAAQ,EAAE,OAAO,CAAC,WAAW,CAAC,CAAC;YACjD,IAAI,QAAQ,CAAC,KAAK,KAAK,QAAQ;gBAAE,MAAM,aAAa,CAAC,sDAAsD,EAAE,2BAA2B,CAAC,CAAC;YAC1I,IAAI,QAAQ,CAAC,KAAK,KAAK,OAAO;gBAAE,MAAM,OAAO,CAAC,aAAa,EAAE,CAAC,QAAQ,CAAC,CAAC;YACxE,OAAO,EAAE,KAAK,EAAE,KAAK,EAAE,QAAQ,EAAE,QAAQ,EAAE,CAAC;QAC9C,CAAC;QACD,MAAM,GAAG,GAAG,IAAI,IAAI,EAAE,CAAC,WAAW,EAAE,CAAC;QACrC,MAAM,QAAQ,GAAG,sBAAsB,CAAC;YACtC,cAAc,EAAE,GAAG,EAAE,YAAY,EAAE,OAAO,CAAC,WAAW,EAAE,oBAAoB,EAAE,OAAO,CAAC,WAAW;YACjG,KAAK,EAAE,YAAY,EAAE,UAAU,EAAE,GAAG,EAAE,UAAU,EAAE,GAAG;SACtD,EAAE,OAAO,CAAC,WAAW,CAAC,CAAC;QACxB,MAAM,eAAe,CAAC,IAAI,EAAE,QAAQ,CAAC,CAAC;QACtC,OAAO,EAAE,KAAK,EAAE,IAAI,EAAE,QAAQ,EAAE,CAAC;IACnC,CAAC,CAAC,CAAC;AACL,CAAC;AAED,KAAK,UAAU,YAAY,CAAC,OAAuC,EAAE,IAAY,EAAE,OAA4B;IAC7G,IAAI,OAAO,CAAC,KAAK,KAAK,OAAO;QAAE,OAAO,OAAO,CAAC;IAC9C,MAAM,QAAQ,GAAG,IAAI,CAAC,GAAG,EAAE,GAAG,CAAC,OAAO,CAAC,SAAS,IAAI,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,KAAK,CAAC,CAAC;IAC1E,MAAM,IAAI,GAAG,OAAO,CAAC,cAAc,IAAI,GAAG,CAAC;IAC3C,SAAS,CAAC;QACR,IAAI,OAAO,CAAC,MAAM,EAAE,OAAO;YAAE,MAAM,OAAO,CAAC,MAAM,CAAC,MAAM,CAAC;QACzD,IAAI,IAAI,CAAC,GAAG,EAAE,IAAI,QAAQ;YAAE,MAAM,aAAa,CAAC,yDAAyD,EAAE,gCAAgC,CAAC,CAAC;QAC7I,MAAM,KAAK,CAAC,IAAI,CAAC,GAAG,CAAC,IAAI,EAAE,IAAI,CAAC,GAAG,CAAC,CAAC,EAAE,QAAQ,GAAG,IAAI,CAAC,GAAG,EAAE,CAAC,CAAC,EAAE,SAAS,EAAE,OAAO,CAAC,MAAM,CAAC,CAAC,CAAC,EAAE,MAAM,EAAE,OAAO,CAAC,MAAM,EAAE,CAAC,CAAC,CAAC,SAAS,CAAC,CAAC;QACpI,MAAM,OAAO,GAAG,MAAM,qBAAqB,CAAC,OAAO,CAAC,IAAI,EAAE,OAAO,CAAC,WAAW,CAAC,CAAC;QAC/E,IAAI,CAAC,OAAO;YAAE,MAAM,aAAa,CAAC,uCAAuC,EAAE,mCAAmC,CAAC,CAAC;QAChH,iBAAiB,CAAC,OAAO,EAAE,OAAO,CAAC,WAAW,CAAC,CAAC;QAChD,IAAI,OAAO,CAAC,KAAK,KAAK,QAAQ;YAAE,MAAM,aAAa,CAAC,2CAA2C,EAAE,2BAA2B,CAAC,CAAC;QAC9H,IAAI,OAAO,CAAC,KAAK,KAAK,OAAO,EAAE,CAAC;YAC9B,MAAM,OAAO,CAAC,aAAa,EAAE,CAAC,OAAO,CAAC,CAAC;YACvC,OAAO,OAAO,CAAC;QACjB,CAAC;IACH,CAAC;AACH,CAAC;AAED,KAAK,UAAU,WAAW,CACxB,OAAuC,EACvC,IAAY,EACZ,MAA6D;IAE7D,OAAO,gBAAgB,CAAC,OAAO,EAAE,KAAK,IAAI,EAAE;QAC1C,MAAM,OAAO,GAAG,MAAM,qBAAqB,CAAC,OAAO,CAAC,IAAI,EAAE,OAAO,CAAC,WAAW,CAAC,CAAC;QAC/E,IAAI,CAAC,OAAO;YAAE,MAAM,aAAa,CAAC,uCAAuC,EAAE,mCAAmC,CAAC,CAAC;QAChH,iBAAiB,CAAC,OAAO,EAAE,OAAO,CAAC,WAAW,CAAC,CAAC;QAChD,IAAI,OAAO,CAAC,KAAK,KAAK,OAAO,IAAI,OAAO,CAAC,KAAK,KAAK,QAAQ;YAAE,MAAM,aAAa,CAAC,wCAAwC,EAAE,mCAAmC,CAAC,CAAC;QAChK,MAAM,IAAI,GAAG,sBAAsB,CAAC,EAAE,GAAG,MAAM,CAAC,OAAO,CAAC,EAAE,UAAU,EAAE,IAAI,IAAI,EAAE,CAAC,WAAW,EAAE,EAAE,EAAE,OAAO,CAAC,WAAW,CAAC,CAAC;QACvH,MAAM,eAAe,CAAC,IAAI,EAAE,IAAI,CAAC,CAAC;QAClC,OAAO,IAAI,CAAC;IACd,CAAC,CAAC,CAAC;AACL,CAAC;AAED,SAAS,gBAAgB,CAAI,OAAuC,EAAE,SAA2B;IAC/F,MAAM,GAAG,GAAG,UAAU,CAAC,EAAE,YAAY,EAAE,OAAO,CAAC,WAAW,EAAE,CAAC,CAAC,KAAK,CAAC,SAAS,CAAC,MAAM,CAAC,CAAC;IACtF,OAAO,YAAY,CAAC,IAAI,CAAC,IAAI,CAAC,OAAO,CAAC,IAAI,EAAE,OAAO,EAAE,oBAAoB,CAAC,EAAE,GAAG,EAAE,SAAS,EAAE;QAC1F,WAAW,EAAE,2BAA2B,EAAE,eAAe,EAAE,EAAE;KAC9D,CAAC,CAAC;AACL,CAAC;AAED,SAAS,cAAc,CAAC,OAAuC;IAC7D,IAAI,CAAC,OAAO,CAAC,IAAI,IAAI,CAAC,OAAO,CAAC,WAAW,IAAI,CAAC,uBAAuB,CAAC,IAAI,CAAC,OAAO,CAAC,WAAW,CAAC;QAAE,MAAM,IAAI,SAAS,CAAC,uCAAuC,CAAC,CAAC;IAC9J,IAAI,OAAO,CAAC,cAAc,KAAK,SAAS,IAAI,CAAC,CAAC,MAAM,CAAC,aAAa,CAAC,OAAO,CAAC,cAAc,CAAC,IAAI,OAAO,CAAC,cAAc,GAAG,CAAC,CAAC;QAAE,MAAM,IAAI,SAAS,CAAC,6CAA6C,CAAC,CAAC;IAC9L,IAAI,OAAO,CAAC,SAAS,KAAK,SAAS,IAAI,CAAC,CAAC,MAAM,CAAC,aAAa,CAAC,OAAO,CAAC,SAAS,CAAC,IAAI,OAAO,CAAC,SAAS,GAAG,CAAC,CAAC;QAAE,MAAM,IAAI,SAAS,CAAC,oCAAoC,CAAC,CAAC;AACxK,CAAC;AAED,SAAS,YAAY,CAAC,IAAY,EAAE,WAAmB;IACrD,MAAM,EAAE,GAAG,UAAU,CAAC,EAAE,YAAY,EAAE,WAAW,EAAE,CAAC,CAAC,KAAK,CAAC,SAAS,CAAC,MAAM,CAAC,CAAC;IAC7E,OAAO,IAAI,CAAC,IAAI,CAAC,IAAI,EAAE,YAAY,EAAE,EAAE,EAAE,eAAe,CAAC,CAAC;AAC5D,CAAC;AAED,SAAS,iBAAiB,CAAC,MAA2B,EAAE,WAAmB;IACzE,IAAI,MAAM,CAAC,oBAAoB,KAAK,WAAW;QAAE,MAAM,aAAa,CAAC,wCAAwC,EAAE,yCAAyC,CAAC,CAAC;AAC5J,CAAC;AAED,SAAS,aAAa,CAAC,KAAa;IAClC,IAAI,CAAC,qBAAqB,CAAC,IAAI,CAAC,KAAK,CAAC;QAAE,MAAM,IAAI,SAAS,CAAC,uCAAuC,CAAC,CAAC;IACrG,OAAO,KAAK,CAAC;AACf,CAAC;AAED,SAAS,aAAa,CAAC,KAAa;IAClC,IAAI,CAAC,uBAAuB,CAAC,IAAI,CAAC,KAAK,CAAC;QAAE,MAAM,IAAI,SAAS,CAAC,6CAA6C,CAAC,CAAC;IAC7G,OAAO,KAAK,CAAC;AACf,CAAC;AAED,SAAS,qBAAqB,CAAC,OAA4B,EAAE,UAAkB,EAAE,aAAqB;IACpG,MAAM,IAAI,GAAG,aAAa,CAAC,UAAU,CAAC,CAAC;IACvC,MAAM,OAAO,GAAG,aAAa,CAAC,aAAa,CAAC,CAAC;IAC7C,IAAI,OAAO,CAAC,YAAY,IAAI,OAAO,CAAC,YAAY,KAAK,IAAI,IAAI,OAAO,CAAC,eAAe,IAAI,OAAO,CAAC,eAAe,KAAK,OAAO,EAAE,CAAC;QAC5H,MAAM,IAAI,SAAS,CAAC,wDAAwD,CAAC,CAAC;IAChF,CAAC;IACD,OAAO,EAAE,GAAG,OAAO,EAAE,YAAY,EAAE,IAAI,EAAE,eAAe,EAAE,OAAO,EAAE,CAAC;AACtE,CAAC;AAED,SAAS,aAAa,CAAC,OAAe,EAAE,IAAY;IAClD,OAAO,IAAI,UAAU,CAAC,OAAO,EAAE,EAAE,IAAI,EAAE,QAAQ,EAAE,EAAE,EAAE,CAAC,CAAC;AACzD,CAAC"}
|
|
@@ -0,0 +1,116 @@
|
|
|
1
|
+
const RESOURCE_FIELDS = ["cpu_millis", "memory_bytes", "container_slots", "build_slots"];
|
|
2
|
+
export function parseExecutionEvidence(value) {
|
|
3
|
+
const record = exact(value, [
|
|
4
|
+
"schema_version", "provider", "worker_id", "collision_domain_id", "eval_id", "work_id", "lease_id", "lease_epoch",
|
|
5
|
+
"task_id", "reservation", "enforced", "observed",
|
|
6
|
+
], "execution evidence");
|
|
7
|
+
if (record.schema_version !== "1" || !validText(record.provider)
|
|
8
|
+
|| !validText(record.worker_id) || !validText(record.collision_domain_id)
|
|
9
|
+
|| typeof record.eval_id !== "string" || !/^eval_[a-f0-9]{32}$/.test(record.eval_id)
|
|
10
|
+
|| typeof record.work_id !== "string" || !/^work_[a-f0-9]{32}$/.test(record.work_id)
|
|
11
|
+
|| typeof record.lease_id !== "string" || !/^lease_[a-f0-9]{32}$/.test(record.lease_id)
|
|
12
|
+
|| !Number.isSafeInteger(record.lease_epoch) || record.lease_epoch < 1 || !validText(record.task_id)) {
|
|
13
|
+
throw new TypeError("execution evidence identity is invalid");
|
|
14
|
+
}
|
|
15
|
+
const reservation = resources(record.reservation, "execution evidence reservation");
|
|
16
|
+
const enforcedRecord = exact(record.enforced, ["main_limits", "sidecar_limits"], "execution evidence enforced limits");
|
|
17
|
+
const mainLimits = resources(enforcedRecord.main_limits, "execution evidence main limits");
|
|
18
|
+
const sidecarLimits = sidecars(enforcedRecord.sidecar_limits);
|
|
19
|
+
const observedRecord = exact(record.observed, [
|
|
20
|
+
"status", "started_at", "collected_at", "sample_count", "containers", "unavailable_fields", "issues",
|
|
21
|
+
], "execution evidence observation");
|
|
22
|
+
if ((observedRecord.status !== "partial" && observedRecord.status !== "unavailable")
|
|
23
|
+
|| !timestamp(observedRecord.started_at) || !timestamp(observedRecord.collected_at)
|
|
24
|
+
|| Date.parse(observedRecord.collected_at) < Date.parse(observedRecord.started_at)
|
|
25
|
+
|| !Number.isSafeInteger(observedRecord.sample_count) || observedRecord.sample_count < 0
|
|
26
|
+
|| !Array.isArray(observedRecord.containers) || observedRecord.containers.length > 256
|
|
27
|
+
|| !Array.isArray(observedRecord.unavailable_fields)
|
|
28
|
+
|| observedRecord.unavailable_fields.some((entry) => !["cpu_time_ns", "peak_memory_bytes", "exit_status", "image_identity"].includes(entry))
|
|
29
|
+
|| new Set(observedRecord.unavailable_fields).size !== observedRecord.unavailable_fields.length
|
|
30
|
+
|| !Array.isArray(observedRecord.issues) || observedRecord.issues.length > 32
|
|
31
|
+
|| observedRecord.issues.some((entry) => !validText(entry, 512)))
|
|
32
|
+
throw new TypeError("execution evidence observation is invalid");
|
|
33
|
+
const containers = observedRecord.containers.map((entry, index) => container(entry, index));
|
|
34
|
+
if (new Set(containers.map((entry) => entry.container_id)).size !== containers.length
|
|
35
|
+
|| (observedRecord.status === "unavailable" && containers.length !== 0))
|
|
36
|
+
throw new TypeError("execution evidence containers are invalid");
|
|
37
|
+
return {
|
|
38
|
+
schema_version: "1",
|
|
39
|
+
provider: record.provider,
|
|
40
|
+
worker_id: record.worker_id,
|
|
41
|
+
collision_domain_id: record.collision_domain_id,
|
|
42
|
+
eval_id: record.eval_id,
|
|
43
|
+
work_id: record.work_id,
|
|
44
|
+
lease_id: record.lease_id,
|
|
45
|
+
lease_epoch: record.lease_epoch,
|
|
46
|
+
task_id: record.task_id,
|
|
47
|
+
reservation,
|
|
48
|
+
enforced: { main_limits: mainLimits, sidecar_limits: sidecarLimits },
|
|
49
|
+
observed: {
|
|
50
|
+
status: observedRecord.status,
|
|
51
|
+
started_at: observedRecord.started_at,
|
|
52
|
+
collected_at: observedRecord.collected_at,
|
|
53
|
+
sample_count: observedRecord.sample_count,
|
|
54
|
+
containers,
|
|
55
|
+
unavailable_fields: [...observedRecord.unavailable_fields],
|
|
56
|
+
issues: [...observedRecord.issues],
|
|
57
|
+
},
|
|
58
|
+
};
|
|
59
|
+
}
|
|
60
|
+
function container(value, index) {
|
|
61
|
+
const record = exact(value, ["container_id", "first_observed_at", "last_observed_at"], `execution evidence container ${index}`, [
|
|
62
|
+
"name", "image_reference", "image_config_digest", "peak_memory_bytes", "cpu_time_ns", "oom_killed", "exit_code", "exit_reason",
|
|
63
|
+
]);
|
|
64
|
+
if (typeof record.container_id !== "string" || !/^[a-f0-9]{12,64}$/.test(record.container_id)
|
|
65
|
+
|| (record.name !== undefined && !validText(record.name, 256)) || !timestamp(record.first_observed_at) || !timestamp(record.last_observed_at)
|
|
66
|
+
|| (record.image_reference !== undefined && !validText(record.image_reference, 1_024))
|
|
67
|
+
|| (record.image_config_digest !== undefined && (typeof record.image_config_digest !== "string" || !/^sha256:[a-f0-9]{64}$/.test(record.image_config_digest)))
|
|
68
|
+
|| Date.parse(record.last_observed_at) < Date.parse(record.first_observed_at)
|
|
69
|
+
|| (record.peak_memory_bytes !== undefined && (!Number.isSafeInteger(record.peak_memory_bytes) || record.peak_memory_bytes < 0))
|
|
70
|
+
|| (record.cpu_time_ns !== undefined && (!Number.isSafeInteger(record.cpu_time_ns) || record.cpu_time_ns < 0))
|
|
71
|
+
|| (record.oom_killed !== undefined && typeof record.oom_killed !== "boolean")
|
|
72
|
+
|| (record.exit_code !== undefined && (!Number.isSafeInteger(record.exit_code) || record.exit_code < 0 || record.exit_code > 255))
|
|
73
|
+
|| (record.exit_reason !== undefined && !validText(record.exit_reason, 512)))
|
|
74
|
+
throw new TypeError(`execution evidence container ${index} is invalid`);
|
|
75
|
+
return record;
|
|
76
|
+
}
|
|
77
|
+
function sidecars(value) {
|
|
78
|
+
if (!value || typeof value !== "object" || Array.isArray(value))
|
|
79
|
+
throw new TypeError("execution evidence sidecar limits are invalid");
|
|
80
|
+
const result = {};
|
|
81
|
+
for (const [name, entry] of Object.entries(value).sort(([left], [right]) => left.localeCompare(right))) {
|
|
82
|
+
const limit = exact(entry, ["cpu_millis", "memory_bytes"], `execution evidence sidecar ${name}`, ["gpu_count"]);
|
|
83
|
+
if (!validText(name, 255) || !Number.isSafeInteger(limit.cpu_millis) || limit.cpu_millis < 1
|
|
84
|
+
|| !Number.isSafeInteger(limit.memory_bytes) || limit.memory_bytes < 1
|
|
85
|
+
|| (limit.gpu_count !== undefined && (!Number.isSafeInteger(limit.gpu_count) || limit.gpu_count < 1)))
|
|
86
|
+
throw new TypeError("execution evidence sidecar limits are invalid");
|
|
87
|
+
result[name] = { cpu_millis: limit.cpu_millis, memory_bytes: limit.memory_bytes, ...(limit.gpu_count === undefined ? {} : { gpu_count: limit.gpu_count }) };
|
|
88
|
+
}
|
|
89
|
+
return result;
|
|
90
|
+
}
|
|
91
|
+
function resources(value, label) {
|
|
92
|
+
const record = exact(value, RESOURCE_FIELDS, label, ["gpu_count", "ephemeral_disk_bytes"]);
|
|
93
|
+
if (RESOURCE_FIELDS.some((name) => !Number.isSafeInteger(record[name]) || record[name] < 0))
|
|
94
|
+
throw new TypeError(`${label} is invalid`);
|
|
95
|
+
if (record.gpu_count !== undefined && (!Number.isSafeInteger(record.gpu_count) || record.gpu_count < 0))
|
|
96
|
+
throw new TypeError(`${label} is invalid`);
|
|
97
|
+
if (record.ephemeral_disk_bytes !== undefined && (!Number.isSafeInteger(record.ephemeral_disk_bytes) || record.ephemeral_disk_bytes < 0))
|
|
98
|
+
throw new TypeError(`${label} is invalid`);
|
|
99
|
+
return record;
|
|
100
|
+
}
|
|
101
|
+
function exact(value, required, label, optional = []) {
|
|
102
|
+
if (!value || typeof value !== "object" || Array.isArray(value))
|
|
103
|
+
throw new TypeError(`${label} must be an object`);
|
|
104
|
+
const record = value;
|
|
105
|
+
const allowed = new Set([...required, ...optional]);
|
|
106
|
+
if (required.some((key) => !(key in record)) || Object.keys(record).some((key) => !allowed.has(key)))
|
|
107
|
+
throw new TypeError(`${label} fields are invalid`);
|
|
108
|
+
return record;
|
|
109
|
+
}
|
|
110
|
+
function timestamp(value) {
|
|
111
|
+
return typeof value === "string" && Number.isFinite(Date.parse(value));
|
|
112
|
+
}
|
|
113
|
+
function validText(value, maximum = 4_096) {
|
|
114
|
+
return typeof value === "string" && value.length > 0 && value.length <= maximum && !/[\0\r\n]/.test(value);
|
|
115
|
+
}
|
|
116
|
+
//# sourceMappingURL=execution-evidence.js.map
|