agent-hitch 0.2.9 → 0.2.11
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 +208 -138
- package/README.zh-CN.md +195 -120
- package/dist/scripts/canary-native-training.js +221 -0
- package/dist/scripts/canary-native-training.js.map +1 -0
- package/dist/scripts/canary-remote-training.js +246 -0
- package/dist/scripts/canary-remote-training.js.map +1 -0
- package/dist/scripts/canary-remote-verifier.js +124 -0
- package/dist/scripts/canary-remote-verifier.js.map +1 -0
- package/dist/scripts/canary-worker-host.js +113 -0
- package/dist/scripts/canary-worker-host.js.map +1 -0
- package/dist/scripts/canary-worker-peer.js +188 -0
- package/dist/scripts/canary-worker-peer.js.map +1 -0
- package/dist/scripts/check-architecture.js +4 -3
- package/dist/scripts/check-architecture.js.map +1 -1
- package/dist/src/adapters/catalog.js +2 -1
- package/dist/src/adapters/catalog.js.map +1 -1
- package/dist/src/adapters/providers/codex.js +65 -2
- package/dist/src/adapters/providers/codex.js.map +1 -1
- package/dist/src/adapters/providers/model-call.js +17 -1
- package/dist/src/adapters/providers/model-call.js.map +1 -1
- package/dist/src/adapters/providers/training-tool.js +29 -0
- package/dist/src/adapters/providers/training-tool.js.map +1 -0
- package/dist/src/artifacts/index.js +1 -1
- package/dist/src/artifacts/index.js.map +1 -1
- package/dist/src/artifacts/store.js +15 -10
- package/dist/src/artifacts/store.js.map +1 -1
- package/dist/src/backends/harbor/backend.js +24 -15
- package/dist/src/backends/harbor/backend.js.map +1 -1
- package/dist/src/backends/harbor/bridge-environment.js +12 -0
- package/dist/src/backends/harbor/bridge-environment.js.map +1 -0
- package/dist/src/backends/harbor/environment-config.js +2 -2
- package/dist/src/backends/harbor/environment-config.js.map +1 -1
- package/dist/src/backends/harbor/host-credential-helper.js +81 -0
- package/dist/src/backends/harbor/host-credential-helper.js.map +1 -0
- package/dist/src/backends/harbor/index.js +3 -0
- package/dist/src/backends/harbor/index.js.map +1 -1
- package/dist/src/backends/harbor/model-proxy-config.js +18 -1
- package/dist/src/backends/harbor/model-proxy-config.js.map +1 -1
- package/dist/src/backends/harbor/observation.js +61 -0
- package/dist/src/backends/harbor/observation.js.map +1 -0
- package/dist/src/backends/harbor/portable-regrade.js +150 -0
- package/dist/src/backends/harbor/portable-regrade.js.map +1 -0
- package/dist/src/backends/harbor/process.js +26 -5
- package/dist/src/backends/harbor/process.js.map +1 -1
- package/dist/src/backends/harbor/regrade.js +4 -1
- package/dist/src/backends/harbor/regrade.js.map +1 -1
- package/dist/src/backends/harbor/supervisor.js +22 -0
- package/dist/src/backends/harbor/supervisor.js.map +1 -1
- package/dist/src/backends/harbor/tools.js +2 -0
- package/dist/src/backends/harbor/tools.js.map +1 -1
- package/dist/src/backends/index.js +3 -0
- package/dist/src/backends/index.js.map +1 -1
- package/dist/src/cli/arguments.js +33 -0
- package/dist/src/cli/arguments.js.map +1 -1
- package/dist/src/cli/commands/capabilities.js +15 -1
- package/dist/src/cli/commands/capabilities.js.map +1 -1
- package/dist/src/cli/commands/daemon.js +19 -0
- package/dist/src/cli/commands/daemon.js.map +1 -1
- package/dist/src/cli/commands/eval.js +38 -10
- package/dist/src/cli/commands/eval.js.map +1 -1
- package/dist/src/cli/commands/local.js +130 -0
- package/dist/src/cli/commands/local.js.map +1 -0
- package/dist/src/cli/commands/model-node.js +26 -0
- package/dist/src/cli/commands/model-node.js.map +1 -0
- package/dist/src/cli/commands/models.js +72 -0
- package/dist/src/cli/commands/models.js.map +1 -0
- package/dist/src/cli/commands/run.js +36 -1
- package/dist/src/cli/commands/run.js.map +1 -1
- package/dist/src/cli/commands/training.js +32 -0
- package/dist/src/cli/commands/training.js.map +1 -0
- package/dist/src/cli/commands/verifier.js +54 -3
- package/dist/src/cli/commands/verifier.js.map +1 -1
- package/dist/src/cli/commands/worker.js +30 -2
- package/dist/src/cli/commands/worker.js.map +1 -1
- package/dist/src/cli/main.js +8 -0
- package/dist/src/cli/main.js.map +1 -1
- package/dist/src/cli/output.js +26 -2
- package/dist/src/cli/output.js.map +1 -1
- package/dist/src/control-plane/eval-records.js +2 -0
- package/dist/src/control-plane/eval-records.js.map +1 -1
- package/dist/src/control-plane/eval-recovery.js +7 -1
- package/dist/src/control-plane/eval-recovery.js.map +1 -1
- package/dist/src/control-plane/eval-scheduler.js +24 -50
- package/dist/src/control-plane/eval-scheduler.js.map +1 -1
- package/dist/src/control-plane/eval-submission-persist.js +30 -0
- package/dist/src/control-plane/eval-submission-persist.js.map +1 -0
- package/dist/src/control-plane/execution-observation-contract.js +60 -0
- package/dist/src/control-plane/execution-observation-contract.js.map +1 -0
- package/dist/src/control-plane/index.js +10 -0
- package/dist/src/control-plane/index.js.map +1 -1
- package/dist/src/control-plane/inference-manager.js +357 -0
- package/dist/src/control-plane/inference-manager.js.map +1 -0
- package/dist/src/control-plane/managed-gateway-receipt.js +26 -0
- package/dist/src/control-plane/managed-gateway-receipt.js.map +1 -0
- package/dist/src/control-plane/managed-service-recovery.js +158 -0
- package/dist/src/control-plane/managed-service-recovery.js.map +1 -0
- package/dist/src/control-plane/model-capture-planning.js +15 -1
- package/dist/src/control-plane/model-capture-planning.js.map +1 -1
- package/dist/src/control-plane/ordered-eval-control.js +221 -0
- package/dist/src/control-plane/ordered-eval-control.js.map +1 -0
- package/dist/src/control-plane/remote-backend-result.js +18 -0
- package/dist/src/control-plane/remote-backend-result.js.map +1 -0
- package/dist/src/control-plane/remote-execution-lease.js +15 -0
- package/dist/src/control-plane/remote-execution-lease.js.map +1 -0
- package/dist/src/control-plane/remote-execution-observation.js +108 -0
- package/dist/src/control-plane/remote-execution-observation.js.map +1 -0
- package/dist/src/control-plane/remote-model-routes.js +107 -0
- package/dist/src/control-plane/remote-model-routes.js.map +1 -0
- package/dist/src/control-plane/remote-observation-reporter.js +37 -0
- package/dist/src/control-plane/remote-observation-reporter.js.map +1 -0
- package/dist/src/control-plane/remote-offer-contract.js +48 -0
- package/dist/src/control-plane/remote-offer-contract.js.map +1 -0
- package/dist/src/control-plane/remote-rerun-journal.js +61 -0
- package/dist/src/control-plane/remote-rerun-journal.js.map +1 -0
- package/dist/src/control-plane/remote-result-transport.js +55 -9
- package/dist/src/control-plane/remote-result-transport.js.map +1 -1
- package/dist/src/control-plane/remote-verifier-import.js +52 -0
- package/dist/src/control-plane/remote-verifier-import.js.map +1 -0
- package/dist/src/control-plane/remote-verifier-result.js +61 -0
- package/dist/src/control-plane/remote-verifier-result.js.map +1 -0
- package/dist/src/control-plane/remote-verifier-source-transport.js +68 -0
- package/dist/src/control-plane/remote-verifier-source-transport.js.map +1 -0
- package/dist/src/control-plane/remote-work-coordinator.js +103 -47
- package/dist/src/control-plane/remote-work-coordinator.js.map +1 -1
- package/dist/src/control-plane/remote-work-inputs.js +26 -5
- package/dist/src/control-plane/remote-work-inputs.js.map +1 -1
- package/dist/src/control-plane/remote-work-recovery.js +123 -34
- package/dist/src/control-plane/remote-work-recovery.js.map +1 -1
- package/dist/src/control-plane/remote-worker-artifacts.js +10 -7
- package/dist/src/control-plane/remote-worker-artifacts.js.map +1 -1
- package/dist/src/control-plane/remote-worker-cleanup.js +104 -0
- package/dist/src/control-plane/remote-worker-cleanup.js.map +1 -0
- package/dist/src/control-plane/remote-worker-client.js +156 -14
- package/dist/src/control-plane/remote-worker-client.js.map +1 -1
- package/dist/src/control-plane/remote-worker-codecs.js +160 -0
- package/dist/src/control-plane/remote-worker-codecs.js.map +1 -0
- package/dist/src/control-plane/remote-worker-generation-recovery.js +67 -0
- package/dist/src/control-plane/remote-worker-generation-recovery.js.map +1 -0
- package/dist/src/control-plane/remote-worker-ownership.js +51 -0
- package/dist/src/control-plane/remote-worker-ownership.js.map +1 -0
- package/dist/src/control-plane/remote-worker-protocol.js +85 -119
- package/dist/src/control-plane/remote-worker-protocol.js.map +1 -1
- package/dist/src/control-plane/remote-worker-runner.js +95 -18
- package/dist/src/control-plane/remote-worker-runner.js.map +1 -1
- package/dist/src/control-plane/remote-workers.js +33 -5
- package/dist/src/control-plane/remote-workers.js.map +1 -1
- package/dist/src/control-plane/rerun-recovery.js +89 -0
- package/dist/src/control-plane/rerun-recovery.js.map +1 -0
- package/dist/src/control-plane/rerun-scheduler.js +70 -96
- package/dist/src/control-plane/rerun-scheduler.js.map +1 -1
- package/dist/src/control-plane/rerun-source.js +57 -0
- package/dist/src/control-plane/rerun-source.js.map +1 -0
- package/dist/src/control-plane/worker-model-relay.js +76 -0
- package/dist/src/control-plane/worker-model-relay.js.map +1 -0
- package/dist/src/controller-runtime/hash.js +3 -0
- package/dist/src/controller-runtime/hash.js.map +1 -1
- package/dist/src/controller-runtime/index.js +2 -0
- package/dist/src/controller-runtime/index.js.map +1 -1
- package/dist/src/controller-runtime/observation.js +29 -0
- package/dist/src/controller-runtime/observation.js.map +1 -0
- package/dist/src/controller-runtime/resident.js +22 -0
- package/dist/src/controller-runtime/resident.js.map +1 -0
- package/dist/src/daemon/client.js +34 -1
- package/dist/src/daemon/client.js.map +1 -1
- package/dist/src/daemon/execution-observation.js +38 -0
- package/dist/src/daemon/execution-observation.js.map +1 -0
- package/dist/src/daemon/http.js +57 -0
- package/dist/src/daemon/http.js.map +1 -0
- package/dist/src/daemon/inference-prepare.js +47 -0
- package/dist/src/daemon/inference-prepare.js.map +1 -0
- package/dist/src/daemon/scheduler.js +4 -1
- package/dist/src/daemon/scheduler.js.map +1 -1
- package/dist/src/daemon/server.js +80 -82
- package/dist/src/daemon/server.js.map +1 -1
- package/dist/src/daemon/worker-observation.js +34 -0
- package/dist/src/daemon/worker-observation.js.map +1 -0
- package/dist/src/daemon/worker-ownership.js +71 -0
- package/dist/src/daemon/worker-ownership.js.map +1 -0
- package/dist/src/daemon/worker-routes.js +83 -15
- package/dist/src/daemon/worker-routes.js.map +1 -1
- package/dist/src/domain/execution-observation.js +2 -0
- package/dist/src/domain/execution-observation.js.map +1 -0
- package/dist/src/domain/index.js +1 -0
- package/dist/src/domain/index.js.map +1 -1
- package/dist/src/domain/inference.js +2 -0
- package/dist/src/domain/inference.js.map +1 -0
- package/dist/src/domain/runs.js.map +1 -1
- package/dist/src/domain/training.js +2 -0
- package/dist/src/domain/training.js.map +1 -0
- package/dist/src/domain/validation.js +11 -0
- package/dist/src/domain/validation.js.map +1 -1
- package/dist/src/domain/verifier-source.js +2 -0
- package/dist/src/domain/verifier-source.js.map +1 -0
- package/dist/src/domain/worker-recovery.js +2 -0
- package/dist/src/domain/worker-recovery.js.map +1 -0
- package/dist/src/evals/eval-logical-plan.js +8 -1
- package/dist/src/evals/eval-logical-plan.js.map +1 -1
- package/dist/src/evals/execution-leases.js +15 -2
- package/dist/src/evals/execution-leases.js.map +1 -1
- package/dist/src/evals/execution-plan.js +11 -1
- package/dist/src/evals/execution-plan.js.map +1 -1
- package/dist/src/evals/harbor-bridge-error.js +8 -0
- package/dist/src/evals/harbor-bridge-error.js.map +1 -1
- package/dist/src/evals/index.js +20 -0
- package/dist/src/evals/index.js.map +1 -1
- package/dist/src/evals/model-capture-plan.js +12 -0
- package/dist/src/evals/model-capture-plan.js.map +1 -1
- package/dist/src/evals/model-capture-runtime.js +25 -0
- package/dist/src/evals/model-capture-runtime.js.map +1 -1
- package/dist/src/evals/physical-work-plan.js +57 -0
- package/dist/src/evals/physical-work-plan.js.map +1 -0
- package/dist/src/evals/preparation-rerun.js +6 -0
- package/dist/src/evals/preparation-rerun.js.map +1 -1
- package/dist/src/evals/recovery.js +8 -1
- package/dist/src/evals/recovery.js.map +1 -1
- package/dist/src/evals/regrade-evidence.js +4 -4
- package/dist/src/evals/regrade-evidence.js.map +1 -1
- package/dist/src/evals/remote-execution-ownership.js +50 -0
- package/dist/src/evals/remote-execution-ownership.js.map +1 -0
- package/dist/src/evals/remote-infrastructure-retry.js +2 -0
- package/dist/src/evals/remote-infrastructure-retry.js.map +1 -1
- package/dist/src/evals/remote-lease-release.js +58 -0
- package/dist/src/evals/remote-lease-release.js.map +1 -0
- package/dist/src/evals/remote-rerun-execution.js +85 -0
- package/dist/src/evals/remote-rerun-execution.js.map +1 -0
- package/dist/src/evals/remote-trial.js +6 -0
- package/dist/src/evals/remote-trial.js.map +1 -0
- package/dist/src/evals/remote-verifier-assessment.js +85 -0
- package/dist/src/evals/remote-verifier-assessment.js.map +1 -0
- package/dist/src/evals/remote-verifier-contract.js +43 -0
- package/dist/src/evals/remote-verifier-contract.js.map +1 -0
- package/dist/src/evals/remote-verifier-execution.js +105 -0
- package/dist/src/evals/remote-verifier-execution.js.map +1 -0
- package/dist/src/evals/remote-verifier-journal.js +21 -0
- package/dist/src/evals/remote-verifier-journal.js.map +1 -0
- package/dist/src/evals/remote-verifier-outcome.js +51 -0
- package/dist/src/evals/remote-verifier-outcome.js.map +1 -0
- package/dist/src/evals/remote-verifier-rerun.js +134 -0
- package/dist/src/evals/remote-verifier-rerun.js.map +1 -0
- package/dist/src/evals/remote-verifier-selection.js +61 -0
- package/dist/src/evals/remote-verifier-selection.js.map +1 -0
- package/dist/src/evals/remote-verifier-source.js +33 -0
- package/dist/src/evals/remote-verifier-source.js.map +1 -0
- package/dist/src/evals/remote-worker-cleanup.js +46 -0
- package/dist/src/evals/remote-worker-cleanup.js.map +1 -0
- package/dist/src/evals/request.js +58 -1
- package/dist/src/evals/request.js.map +1 -1
- package/dist/src/evals/rerun-completion.js +39 -0
- package/dist/src/evals/rerun-completion.js.map +1 -0
- package/dist/src/evals/rerun-resume.js +70 -0
- package/dist/src/evals/rerun-resume.js.map +1 -0
- package/dist/src/evals/rerun-state.js +25 -0
- package/dist/src/evals/rerun-state.js.map +1 -0
- package/dist/src/evals/rerun-types.js.map +1 -1
- package/dist/src/evals/rerun.js +90 -46
- package/dist/src/evals/rerun.js.map +1 -1
- package/dist/src/evals/service.js +59 -5
- package/dist/src/evals/service.js.map +1 -1
- package/dist/src/evals/verifier-artifacts.js +98 -153
- package/dist/src/evals/verifier-artifacts.js.map +1 -1
- package/dist/src/evals/verifier-diagnostic-repair.js +276 -0
- package/dist/src/evals/verifier-diagnostic-repair.js.map +1 -0
- package/dist/src/evals/verifier-only-rerun.js +2 -16
- package/dist/src/evals/verifier-only-rerun.js.map +1 -1
- package/dist/src/evals/verifier-rerun-result.js +24 -0
- package/dist/src/evals/verifier-rerun-result.js.map +1 -0
- package/dist/src/evals/verifier-source-import.js +48 -0
- package/dist/src/evals/verifier-source-import.js.map +1 -0
- package/dist/src/evals/verifier-source.js +139 -0
- package/dist/src/evals/verifier-source.js.map +1 -0
- package/dist/src/foundation/base64.js +19 -0
- package/dist/src/foundation/base64.js.map +1 -0
- package/dist/src/foundation/config.js +10 -0
- package/dist/src/foundation/config.js.map +1 -1
- package/dist/src/foundation/credential-redaction.js +22 -9
- package/dist/src/foundation/credential-redaction.js.map +1 -1
- package/dist/src/foundation/index.js +1 -0
- package/dist/src/foundation/index.js.map +1 -1
- package/dist/src/foundation/process-identity.js +13 -9
- package/dist/src/foundation/process-identity.js.map +1 -1
- package/dist/src/inference/daemon-capacity.js +41 -0
- package/dist/src/inference/daemon-capacity.js.map +1 -0
- package/dist/src/inference/device-reservation.js +58 -0
- package/dist/src/inference/device-reservation.js.map +1 -0
- package/dist/src/inference/doctor.js +95 -0
- package/dist/src/inference/doctor.js.map +1 -0
- package/dist/src/inference/generation-recovery.js +83 -0
- package/dist/src/inference/generation-recovery.js.map +1 -0
- package/dist/src/inference/index.js +24 -0
- package/dist/src/inference/index.js.map +1 -0
- package/dist/src/inference/lock.js +331 -0
- package/dist/src/inference/lock.js.map +1 -0
- package/dist/src/inference/managed-launcher.js +35 -0
- package/dist/src/inference/managed-launcher.js.map +1 -0
- package/dist/src/inference/manifest.js +262 -0
- package/dist/src/inference/manifest.js.map +1 -0
- package/dist/src/inference/materialize.js +66 -0
- package/dist/src/inference/materialize.js.map +1 -0
- package/dist/src/inference/model-location.js +34 -0
- package/dist/src/inference/model-location.js.map +1 -0
- package/dist/src/inference/model-store.js +439 -0
- package/dist/src/inference/model-store.js.map +1 -0
- package/dist/src/inference/node-client.js +167 -0
- package/dist/src/inference/node-client.js.map +1 -0
- package/dist/src/inference/node-registry.js +79 -0
- package/dist/src/inference/node-registry.js.map +1 -0
- package/dist/src/inference/observation.js +79 -0
- package/dist/src/inference/observation.js.map +1 -0
- package/dist/src/inference/preflight.js +132 -0
- package/dist/src/inference/preflight.js.map +1 -0
- package/dist/src/inference/process-sglang.js +169 -0
- package/dist/src/inference/process-sglang.js.map +1 -0
- package/dist/src/inference/protocol-probe.js +29 -0
- package/dist/src/inference/protocol-probe.js.map +1 -0
- package/dist/src/inference/remote-model-store.js +24 -0
- package/dist/src/inference/remote-model-store.js.map +1 -0
- package/dist/src/inference/runtime-catalog.js +41 -0
- package/dist/src/inference/runtime-catalog.js.map +1 -0
- package/dist/src/inference/runtime-store.js +93 -0
- package/dist/src/inference/runtime-store.js.map +1 -0
- package/dist/src/inference/service-attachment.js +46 -0
- package/dist/src/inference/service-attachment.js.map +1 -0
- package/dist/src/inference/service-inspection.js +38 -0
- package/dist/src/inference/service-inspection.js.map +1 -0
- package/dist/src/inference/sglang.js +279 -0
- package/dist/src/inference/sglang.js.map +1 -0
- package/dist/src/inference/supervisor.js +460 -0
- package/dist/src/inference/supervisor.js.map +1 -0
- package/dist/src/model-access/environment.js +28 -0
- package/dist/src/model-access/environment.js.map +1 -0
- package/dist/src/model-access/index.js +4 -0
- package/dist/src/model-access/index.js.map +1 -1
- package/dist/src/model-access/local-gateway-request.js +139 -0
- package/dist/src/model-access/local-gateway-request.js.map +1 -0
- package/dist/src/model-access/local-gateway.js +323 -0
- package/dist/src/model-access/local-gateway.js.map +1 -0
- package/dist/src/model-access/proxy.js +90 -5
- package/dist/src/model-access/proxy.js.map +1 -1
- package/dist/src/model-access/remote-model.js +58 -0
- package/dist/src/model-access/remote-model.js.map +1 -0
- package/dist/src/model-access/training.js +115 -0
- package/dist/src/model-access/training.js.map +1 -0
- package/dist/src/runs/adapter-process.js +2 -2
- package/dist/src/runs/adapter-process.js.map +1 -1
- package/dist/src/runs/executor-types.js +2 -0
- package/dist/src/runs/executor-types.js.map +1 -0
- package/dist/src/runs/executor.js +54 -34
- package/dist/src/runs/executor.js.map +1 -1
- package/dist/src/runs/index.js +4 -2
- package/dist/src/runs/index.js.map +1 -1
- package/dist/src/runs/local-inference-environment.js +59 -0
- package/dist/src/runs/local-inference-environment.js.map +1 -0
- package/dist/src/runs/local-inference-run.js +53 -0
- package/dist/src/runs/local-inference-run.js.map +1 -0
- package/dist/src/runs/outcome.js +22 -0
- package/dist/src/runs/outcome.js.map +1 -1
- package/dist/src/runs/records.js +9 -0
- package/dist/src/runs/records.js.map +1 -1
- package/dist/src/runs/request.js +58 -3
- package/dist/src/runs/request.js.map +1 -1
- package/dist/src/runs/training-evidence.js +22 -0
- package/dist/src/runs/training-evidence.js.map +1 -0
- package/dist/src/runs/training-runtime.js +20 -0
- package/dist/src/runs/training-runtime.js.map +1 -0
- package/dist/src/runs/verifier-diagnostic-storage.js +237 -0
- package/dist/src/runs/verifier-diagnostic-storage.js.map +1 -0
- package/dist/src/runs/verifier-diagnostic-supplement.js +228 -0
- package/dist/src/runs/verifier-diagnostic-supplement.js.map +1 -0
- package/dist/src/runs/verifier-evidence.js +119 -118
- package/dist/src/runs/verifier-evidence.js.map +1 -1
- package/dist/src/trajectories/providers/deepseek.js +52 -38
- package/dist/src/trajectories/providers/deepseek.js.map +1 -1
- package/dist/src/trajectories/store.js +17 -0
- package/dist/src/trajectories/store.js.map +1 -1
- package/dist/src/workers/execution-observation.js +32 -0
- package/dist/src/workers/execution-observation.js.map +1 -0
- package/dist/src/workers/index.js +2 -0
- package/dist/src/workers/index.js.map +1 -1
- package/dist/src/workers/remote-harbor-ownership.js +266 -0
- package/dist/src/workers/remote-harbor-ownership.js.map +1 -0
- package/dist/src/workers/remote-harbor-verifier.js +76 -0
- package/dist/src/workers/remote-harbor-verifier.js.map +1 -0
- package/dist/src/workers/remote-harbor-work-spec.js +66 -10
- package/dist/src/workers/remote-harbor-work-spec.js.map +1 -1
- package/dist/src/workers/remote-harbor-worker.js +56 -49
- package/dist/src/workers/remote-harbor-worker.js.map +1 -1
- package/dist/src/workers/remote-worker-host.js +33 -0
- package/dist/src/workers/remote-worker-host.js.map +1 -0
- package/docs/local-model-inference-spec.zh-CN.md +666 -0
- package/docs/schemas/controller-runtime-observation.schema.json +21 -0
- package/docs/schemas/eval-execution-plan.schema.json +1 -0
- package/docs/schemas/eval-request.schema.json +3 -1
- package/docs/schemas/execution-lease.schema.json +95 -1
- package/docs/schemas/execution-provider-status.schema.json +162 -28
- package/docs/schemas/inference-execution-evidence.schema.json +487 -0
- package/docs/schemas/inference-lock.schema.json +487 -0
- package/docs/schemas/inference-runtime-manifest.schema.json +156 -0
- package/docs/schemas/inference-service-record.schema.json +227 -0
- package/docs/schemas/local-inference-selection.schema.json +98 -0
- package/docs/schemas/local-model-manifest.schema.json +43 -0
- package/docs/schemas/model-node-binding.schema.json +34 -0
- package/docs/schemas/model-node-generation-release.schema.json +46 -0
- package/docs/schemas/ordered-eval-control.schema.json +39 -0
- package/docs/schemas/remote-execution-lease-grant.schema.json +45 -0
- package/docs/schemas/remote-harbor-work-spec.schema.json +325 -0
- package/docs/schemas/remote-model-binding.schema.json +21 -0
- package/docs/schemas/remote-physical-execution.schema.json +108 -0
- package/docs/schemas/remote-result-envelope.schema.json +9 -3
- package/docs/schemas/remote-tree-envelope.schema.json +17 -3
- package/docs/schemas/remote-verifier-result.schema.json +203 -0
- package/docs/schemas/remote-verifier-source.schema.json +52 -0
- package/docs/schemas/remote-verifier-work.schema.json +125 -0
- package/docs/schemas/remote-work-offer.schema.json +170 -34
- package/docs/schemas/remote-worker-execution-admission.schema.json +214 -0
- package/docs/schemas/remote-worker-generation-cleanup.schema.json +199 -0
- package/docs/schemas/remote-worker-registration.schema.json +226 -25
- package/docs/schemas/run-manifest.schema.json +211 -44
- package/docs/schemas/run-request.schema.json +2 -1
- package/docs/schemas/training-external-binding.schema.json +26 -0
- package/docs/schemas/verifier-diagnostic-page.schema.json +97 -0
- package/integrations/harbor/hitch_harbor_agent.py +241 -13
- package/integrations/harbor/hitch_harbor_environment.py +22 -1
- package/integrations/harbor/hitch_host_credentials.py +354 -0
- package/integrations/harbor/hitch_private_exec.py +150 -0
- package/integrations/model-call/cli.js +13 -5
- package/integrations/training-tool/cli.js +72 -0
- package/package.json +8 -2
|
@@ -1,3 +1,4 @@
|
|
|
1
|
+
import { parseModelNodeBinding } from "../domain/index.js";
|
|
1
2
|
export const DEFAULT_EVAL_TIMEOUT_MS = 15 * 60 * 1_000;
|
|
2
3
|
export const DEFAULT_EVAL_SETUP_TIMEOUT_MS = 30 * 60 * 1_000;
|
|
3
4
|
export const DEFAULT_INFRASTRUCTURE_RETRIES = 1;
|
|
@@ -20,7 +21,7 @@ export async function validateEvalRequest(input) {
|
|
|
20
21
|
const allowed = new Set([
|
|
21
22
|
"schema_version", "backend", "dataset", "harness_ref", "model", "attempts",
|
|
22
23
|
"max_concurrent", "infrastructure_retries", "infrastructure_retry_backoff_ms",
|
|
23
|
-
"timeout_ms", "setup_timeout_ms", "agent_args", "pass_env",
|
|
24
|
+
"timeout_ms", "setup_timeout_ms", "agent_args", "pass_env", "local_inference", "training_binding",
|
|
24
25
|
]);
|
|
25
26
|
const unexpected = Object.keys(input).find((field) => !allowed.has(field));
|
|
26
27
|
if (unexpected)
|
|
@@ -53,6 +54,14 @@ export async function validateEvalRequest(input) {
|
|
|
53
54
|
if (input.pass_env !== undefined && (!Array.isArray(input.pass_env) || input.pass_env.some((value) => typeof value !== "string"))) {
|
|
54
55
|
throw invalidInput("pass_env must be an array of strings");
|
|
55
56
|
}
|
|
57
|
+
const localInference = validateEvalLocalInference(input.local_inference, typeof input.model === "string" ? input.model : "");
|
|
58
|
+
const training = input.training_binding === undefined ? undefined : parseTrainingBinding(input.training_binding);
|
|
59
|
+
if (training && (localInference || input.model !== `training/${training.bindingId}` || attempts !== 1 || infrastructureRetries !== 0
|
|
60
|
+
|| reference.harness_id !== "training-tool" || input.agent_args?.length || input.pass_env?.length)) {
|
|
61
|
+
throw invalidInput("training-external requires training-tool, one logical attempt, zero automatic retries, no argument/env overrides, and its exact training/<bindingId> model");
|
|
62
|
+
}
|
|
63
|
+
if (!training && String(input.model).startsWith("training/"))
|
|
64
|
+
throw invalidInput("training models require an explicit training binding");
|
|
56
65
|
const agentArgs = Array.isArray(input.agent_args) ? [...input.agent_args] : [];
|
|
57
66
|
if (benchmark.manifest) {
|
|
58
67
|
await assertStandardBenchmarkCandidate(input.dataset.trim(), benchmark.manifest, reference.harness_id, agentArgs);
|
|
@@ -76,6 +85,53 @@ export async function validateEvalRequest(input) {
|
|
|
76
85
|
pass_env: [...new Set(Array.isArray(input.pass_env) ? input.pass_env : [])],
|
|
77
86
|
benchmark_id: benchmark.benchmark_id,
|
|
78
87
|
benchmark_revision: benchmark.benchmark_revision,
|
|
88
|
+
...(localInference ? { local_inference: localInference } : {}),
|
|
89
|
+
...(training ? { training_binding: training } : {}),
|
|
90
|
+
};
|
|
91
|
+
}
|
|
92
|
+
function validateEvalLocalInference(value, model) {
|
|
93
|
+
const localModel = model.startsWith("local/");
|
|
94
|
+
if (!localModel) {
|
|
95
|
+
if (value !== undefined)
|
|
96
|
+
throw invalidInput("--device, --local-profile, --inference and --offline require a local/<name> model");
|
|
97
|
+
return undefined;
|
|
98
|
+
}
|
|
99
|
+
if (value === undefined)
|
|
100
|
+
return { model, device: "auto", profile: "baseline", offline: false };
|
|
101
|
+
if (!value || typeof value !== "object" || Array.isArray(value))
|
|
102
|
+
throw invalidInput("local_inference must be an object");
|
|
103
|
+
const record = value;
|
|
104
|
+
const allowed = new Set(["model", "device", "profile", "inference_id", "offline", "model_node"]);
|
|
105
|
+
const unexpected = Object.keys(record).find((field) => !allowed.has(field));
|
|
106
|
+
if (unexpected)
|
|
107
|
+
throw invalidInput(`local_inference has unknown field: ${unexpected}`);
|
|
108
|
+
if (record.model !== undefined && record.model !== model)
|
|
109
|
+
throw invalidInput("local_inference.model must match model");
|
|
110
|
+
const modelNode = record.model_node === undefined ? undefined : parseModelNodeBinding(record.model_node);
|
|
111
|
+
if (modelNode && record.inference_id === undefined)
|
|
112
|
+
throw new TypeError("model-node selection requires a planned inference lock");
|
|
113
|
+
const normalizedNodeSelection = modelNode && (record.device === undefined || record.device === "auto") && (record.profile === undefined || record.profile === "baseline");
|
|
114
|
+
if (record.inference_id !== undefined && !normalizedNodeSelection && (record.device !== undefined || record.profile !== undefined)) {
|
|
115
|
+
throw invalidInput("--inference cannot be combined with --device or --local-profile");
|
|
116
|
+
}
|
|
117
|
+
const device = record.device ?? "auto";
|
|
118
|
+
const profile = record.profile ?? "baseline";
|
|
119
|
+
if (!new Set(["auto", "cpu", "cuda", "metal"]).has(String(device)))
|
|
120
|
+
throw invalidInput("local_inference.device must be auto, cpu, cuda, or metal");
|
|
121
|
+
if (profile !== "baseline" && profile !== "throughput")
|
|
122
|
+
throw invalidInput("local_inference.profile must be baseline or throughput");
|
|
123
|
+
if (record.offline !== undefined && typeof record.offline !== "boolean")
|
|
124
|
+
throw invalidInput("local_inference.offline must be a boolean");
|
|
125
|
+
if (record.inference_id !== undefined && (typeof record.inference_id !== "string" || !/^sha256:[a-f0-9]{64}$/.test(record.inference_id))) {
|
|
126
|
+
throw invalidInput("local_inference.inference_id must be a SHA-256 digest");
|
|
127
|
+
}
|
|
128
|
+
return {
|
|
129
|
+
model,
|
|
130
|
+
device: device,
|
|
131
|
+
profile,
|
|
132
|
+
offline: record.offline === true,
|
|
133
|
+
...(modelNode ? { model_node: modelNode } : {}),
|
|
134
|
+
...(record.inference_id === undefined ? {} : { inference_id: record.inference_id }),
|
|
79
135
|
};
|
|
80
136
|
}
|
|
81
137
|
export async function resolveBenchmarkReference(dataset) {
|
|
@@ -173,4 +229,5 @@ import { assertExactLocalGitEvalReference, parseHarnessReference } from "../revi
|
|
|
173
229
|
import { workspaceDigest } from "../workspaces/index.js";
|
|
174
230
|
import { loadBenchmarkAdapterManifest } from "./benchmark-adapter-manifest.js";
|
|
175
231
|
import { assertStandardBenchmarkCandidate } from "./benchmark-candidate.js";
|
|
232
|
+
import { parseTrainingBinding } from "../model-access/index.js";
|
|
176
233
|
//# sourceMappingURL=request.js.map
|
|
@@ -1 +1 @@
|
|
|
1
|
-
{"version":3,"file":"request.js","sourceRoot":"","sources":["../../../src/evals/request.ts"],"names":[],"mappings":"AAAA,MAAM,CAAC,MAAM,uBAAuB,GAAG,EAAE,GAAG,EAAE,GAAG,KAAK,CAAC;AACvD,MAAM,CAAC,MAAM,6BAA6B,GAAG,EAAE,GAAG,EAAE,GAAG,KAAK,CAAC;AAC7D,MAAM,CAAC,MAAM,8BAA8B,GAAG,CAAC,CAAC;AAChD,MAAM,CAAC,MAAM,uCAAuC,GAAG,KAAK,CAAC;AAE7D,MAAM,sBAAsB,GAAqC;IAC/D,KAAK,EAAE,4CAA4C;IACnD,QAAQ,EAAE,gCAAgC;CAC3C,CAAC;AAEF,MAAM,UAAU,SAAS;IACvB,OAAO,QAAQ,UAAU,EAAE,CAAC,UAAU,CAAC,GAAG,EAAE,EAAE,CAAC,EAAY,CAAC;AAC9D,CAAC;AAED,MAAM,UAAU,cAAc,CAAC,KAAa;IAC1C,IAAI,CAAC,qBAAqB,CAAC,IAAI,CAAC,KAAK,CAAC;QAAE,MAAM,YAAY,CAAC,oBAAoB,KAAK,EAAE,CAAC,CAAC;IACxF,OAAO,KAAe,CAAC;AACzB,CAAC;AAkBD,MAAM,CAAC,KAAK,UAAU,mBAAmB,CAAC,KAAuB;IAC/D,IAAI,CAAC,KAAK,IAAI,OAAO,KAAK,KAAK,QAAQ,IAAI,KAAK,CAAC,OAAO,CAAC,KAAK,CAAC;QAAE,MAAM,YAAY,CAAC,oCAAoC,CAAC,CAAC;IAC1H,MAAM,OAAO,GAAG,IAAI,GAAG,CAAC;QACtB,gBAAgB,EAAE,SAAS,EAAE,SAAS,EAAE,aAAa,EAAE,OAAO,EAAE,UAAU;QAC1E,gBAAgB,EAAE,wBAAwB,EAAE,iCAAiC;QAC7E,YAAY,EAAE,kBAAkB,EAAE,YAAY,EAAE,UAAU;KAC3D,CAAC,CAAC;IACH,MAAM,UAAU,GAAG,MAAM,CAAC,IAAI,CAAC,KAAK,CAAC,CAAC,IAAI,CAAC,CAAC,KAAK,EAAE,EAAE,CAAC,CAAC,OAAO,CAAC,GAAG,CAAC,KAAK,CAAC,CAAC,CAAC;IAC3E,IAAI,UAAU;QAAE,MAAM,YAAY,CAAC,+BAA+B,UAAU,EAAE,CAAC,CAAC;IAChF,IAAI,KAAK,CAAC,cAAc,KAAK,SAAS,IAAI,KAAK,CAAC,cAAc,KAAK,cAAc;QAAE,MAAM,YAAY,CAAC,+BAA+B,KAAK,CAAC,cAAc,EAAE,CAAC,CAAC;IAC7J,IAAI,CAAC,KAAK,CAAC,OAAO,IAAI,QAAQ,CAAC,KAAK,QAAQ;QAAE,MAAM,YAAY,CAAC,qDAAqD,CAAC,CAAC;IACxH,IAAI,OAAO,KAAK,CAAC,OAAO,KAAK,QAAQ,IAAI,CAAC,KAAK,CAAC,OAAO,CAAC,IAAI,EAAE;QAAE,MAAM,YAAY,CAAC,oCAAoC,CAAC,CAAC;IACzH,IAAI,OAAO,KAAK,CAAC,WAAW,KAAK,QAAQ,IAAI,CAAC,KAAK,CAAC,WAAW,CAAC,IAAI,EAAE;QAAE,MAAM,YAAY,CAAC,wCAAwC,CAAC,CAAC;IACrI,MAAM,SAAS,GAAG,qBAAqB,CAAC,KAAK,CAAC,WAAW,CAAC,CAAC;IAC3D,IAAI,SAAS,CAAC,QAAQ,CAAC,IAAI,KAAK,WAAW,EAAE,CAAC;QAC5C,MAAM,YAAY,CAAC,yEAAyE,CAAC,CAAC;IAChG,CAAC;IACD,gCAAgC,CAAC,SAAS,CAAC,CAAC;IAC5C,MAAM,SAAS,GAAG,MAAM,uBAAuB,CAAC,KAAK,CAAC,OAAO,CAAC,IAAI,EAAE,CAAC,CAAC;IACtE,MAAM,QAAQ,GAAG,eAAe,CAAC,KAAK,CAAC,QAAQ,IAAI,CAAC,EAAE,UAAU,CAAC,CAAC;IAClE,MAAM,aAAa,GAAG,eAAe,CAAC,KAAK,CAAC,cAAc,IAAI,CAAC,EAAE,gBAAgB,CAAC,CAAC;IACnF,MAAM,qBAAqB,GAAG,kBAAkB,CAC9C,KAAK,CAAC,sBAAsB,IAAI,8BAA8B,EAC9D,wBAAwB,CACzB,CAAC;IACF,MAAM,0BAA0B,GAAG,iBAAiB,CAClD,KAAK,CAAC,+BAA+B,IAAI,uCAAuC,EAChF,iCAAiC,CAClC,CAAC;IACF,MAAM,OAAO,GAAG,iBAAiB,CAAC,KAAK,CAAC,UAAU,IAAI,CAAC,SAAS,CAAC,QAAQ,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,uBAAuB,CAAC,EAAE,YAAY,CAAC,CAAC;IACxH,MAAM,YAAY,GAAG,iBAAiB,CAAC,KAAK,CAAC,gBAAgB,IAAI,6BAA6B,EAAE,kBAAkB,CAAC,CAAC;IACpH,IAAI,KAAK,CAAC,KAAK,KAAK,SAAS,IAAI,OAAO,KAAK,CAAC,KAAK,KAAK,QAAQ;QAAE,MAAM,YAAY,CAAC,wBAAwB,CAAC,CAAC;IAC/G,IAAI,KAAK,CAAC,UAAU,KAAK,SAAS,IAAI,CAAC,CAAC,KAAK,CAAC,OAAO,CAAC,KAAK,CAAC,UAAU,CAAC,IAAI,KAAK,CAAC,UAAU,CAAC,IAAI,CAAC,CAAC,KAAK,EAAE,EAAE,CAAC,OAAO,KAAK,KAAK,QAAQ,CAAC,CAAC,EAAE,CAAC;QACxI,MAAM,YAAY,CAAC,wCAAwC,CAAC,CAAC;IAC/D,CAAC;IACD,IAAI,KAAK,CAAC,QAAQ,KAAK,SAAS,IAAI,CAAC,CAAC,KAAK,CAAC,OAAO,CAAC,KAAK,CAAC,QAAQ,CAAC,IAAI,KAAK,CAAC,QAAQ,CAAC,IAAI,CAAC,CAAC,KAAK,EAAE,EAAE,CAAC,OAAO,KAAK,KAAK,QAAQ,CAAC,CAAC,EAAE,CAAC;QAClI,MAAM,YAAY,CAAC,sCAAsC,CAAC,CAAC;IAC7D,CAAC;IACD,MAAM,SAAS,GAAG,KAAK,CAAC,OAAO,CAAC,KAAK,CAAC,UAAU,CAAC,CAAC,CAAC,CAAC,CAAC,GAAG,KAAK,CAAC,UAAU,CAAa,CAAC,CAAC,CAAC,EAAE,CAAC;IAC3F,IAAI,SAAS,CAAC,QAAQ,EAAE,CAAC;QACvB,MAAM,gCAAgC,CAAC,KAAK,CAAC,OAAO,CAAC,IAAI,EAAE,EAAE,SAAS,CAAC,QAAQ,EAAE,SAAS,CAAC,UAAU,EAAE,SAAS,CAAC,CAAC;IACpH,CAAC;IACD,MAAM,SAAS,GAAG,sBAAsB,CAAC,SAAS,CAAC,UAAU,CAAC,CAAC;IAC/D,IAAI,SAAS,IAAI,CAAC,SAAS,CAAC,QAAQ,CAAC,SAAS,CAAC;QAAE,SAAS,CAAC,OAAO,CAAC,SAAS,CAAC,CAAC;IAC9E,OAAO;QACL,cAAc,EAAE,cAAc;QAC9B,OAAO,EAAE,QAAQ;QACjB,OAAO,EAAE,MAAM,CAAC,KAAK,CAAC,OAAO,CAAC,CAAC,IAAI,EAAE;QACrC,WAAW,EAAE,SAAS,CAAC,SAAS;QAChC,KAAK,EAAE,OAAO,KAAK,CAAC,KAAK,KAAK,QAAQ,CAAC,CAAC,CAAC,KAAK,CAAC,KAAK,CAAC,CAAC,CAAC,EAAE;QACzD,QAAQ;QACR,cAAc,EAAE,aAAa;QAC7B,sBAAsB,EAAE,qBAAqB;QAC7C,+BAA+B,EAAE,0BAA0B;QAC3D,UAAU,EAAE,OAAO;QACnB,gBAAgB,EAAE,YAAY;QAC9B,UAAU,EAAE,SAAS;QACrB,QAAQ,EAAE,CAAC,GAAG,IAAI,GAAG,CAAC,KAAK,CAAC,OAAO,CAAC,KAAK,CAAC,QAAQ,CAAC,CAAC,CAAC,CAAC,KAAK,CAAC,QAAoB,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC;QACvF,YAAY,EAAE,SAAS,CAAC,YAAY;QACpC,kBAAkB,EAAE,SAAS,CAAC,kBAAkB;KACjD,CAAC;AACJ,CAAC;AAED,MAAM,CAAC,KAAK,UAAU,yBAAyB,CAAC,OAAe;IAC7D,MAAM,EAAE,YAAY,EAAE,kBAAkB,EAAE,GAAG,MAAM,uBAAuB,CAAC,OAAO,CAAC,CAAC;IACpF,OAAO,EAAE,YAAY,EAAE,kBAAkB,EAAE,CAAC;AAC9C,CAAC;AAED,KAAK,UAAU,uBAAuB,CAAC,OAAe;IAKpD,MAAM,GAAG,GAAG,OAAO,CAAC,IAAI,EAAE,CAAC;IAC3B,MAAM,KAAK,GAAG,IAAI,CAAC,OAAO,CAAC,GAAG,CAAC,CAAC;IAChC,IAAI,CAAC;QACH,IAAI,CAAC,MAAM,IAAI,CAAC,KAAK,CAAC,CAAC,CAAC,WAAW,EAAE,EAAE,CAAC;YACtC,MAAM,QAAQ,GAAG,MAAM,4BAA4B,CAAC,KAAK,CAAC,CAAC;YAC3D,IAAI,QAAQ,EAAE,CAAC;gBACb,OAAO;oBACL,YAAY,EAAE,QAAQ,CAAC,SAAS,CAAC,EAAE;oBACnC,kBAAkB,EAAE,QAAQ,CAAC,cAAc;oBAC3C,QAAQ;iBACT,CAAC;YACJ,CAAC;YACD,OAAO;gBACL,YAAY,EAAE,SAAS,IAAI,CAAC,QAAQ,CAAC,KAAK,CAAC,EAAE;gBAC7C,kBAAkB,EAAE,MAAM,eAAe,CAAC,KAAK,EAAE,EAAE,gBAAgB,EAAE,IAAI,GAAG,CAAC,CAAC,MAAM,CAAC,CAAC,EAAE,CAAC;gBACzF,QAAQ,EAAE,IAAI;aACf,CAAC;QACJ,CAAC;IACH,CAAC;IAAC,OAAO,KAAK,EAAE,CAAC;QACf,IAAK,KAA+B,CAAC,IAAI,KAAK,QAAQ;YAAE,MAAM,KAAK,CAAC;IACtE,CAAC;IACD,MAAM,SAAS,GAAG,GAAG,CAAC,WAAW,CAAC,GAAG,CAAC,CAAC;IACvC,IAAI,SAAS,IAAI,CAAC,IAAI,SAAS,KAAK,GAAG,CAAC,MAAM,GAAG,CAAC,EAAE,CAAC;QACnD,MAAM,YAAY,CAAC,wEAAwE,CAAC,CAAC;IAC/F,CAAC;IACD,MAAM,WAAW,GAAG,GAAG,CAAC,KAAK,CAAC,CAAC,EAAE,SAAS,CAAC,CAAC;IAC5C,MAAM,QAAQ,GAAG,GAAG,CAAC,KAAK,CAAC,SAAS,GAAG,CAAC,CAAC,CAAC;IAC1C,IAAI,QAAQ,CAAC,WAAW,EAAE,KAAK,QAAQ;QAAE,MAAM,YAAY,CAAC,mCAAmC,CAAC,CAAC;IACjG,OAAO,EAAE,YAAY,EAAE,WAAW,EAAE,kBAAkB,EAAE,QAAQ,EAAE,QAAQ,EAAE,IAAI,EAAE,CAAC;AACrF,CAAC;AAED,MAAM,CAAC,KAAK,UAAU,0BAA0B,CAAC,OAAe;IAC9D,MAAM,KAAK,GAAG,IAAI,CAAC,OAAO,CAAC,OAAO,CAAC,IAAI,EAAE,CAAC,CAAC;IAC3C,IAAI,CAAC;QACH,IAAI,CAAC,CAAC,MAAM,IAAI,CAAC,KAAK,CAAC,CAAC,CAAC,WAAW,EAAE;YAAE,OAAO,IAAI,CAAC;IACtD,CAAC;IAAC,OAAO,KAAK,EAAE,CAAC;QACf,IAAK,KAA+B,CAAC,IAAI,KAAK,QAAQ;YAAE,OAAO,IAAI,CAAC;QACpE,MAAM,KAAK,CAAC;IACd,CAAC;IACD,IAAI,MAAM,aAAa,CAAC,IAAI,CAAC,IAAI,CAAC,KAAK,EAAE,WAAW,CAAC,CAAC;QAAE,OAAO,CAAC,IAAI,CAAC,QAAQ,CAAC,KAAK,CAAC,CAAC,CAAC;IACtF,MAAM,OAAO,GAAG,MAAM,OAAO,CAAC,KAAK,EAAE,EAAE,aAAa,EAAE,IAAI,EAAE,CAAC,CAAC;IAC9D,MAAM,KAAK,GAAa,EAAE,CAAC;IAC3B,KAAK,MAAM,KAAK,IAAI,OAAO,CAAC,IAAI,CAAC,CAAC,IAAI,EAAE,KAAK,EAAE,EAAE,CAAC,IAAI,CAAC,IAAI,CAAC,aAAa,CAAC,KAAK,CAAC,IAAI,CAAC,CAAC,EAAE,CAAC;QACvF,IAAI,KAAK,CAAC,WAAW,EAAE,IAAI,MAAM,aAAa,CAAC,IAAI,CAAC,IAAI,CAAC,KAAK,EAAE,KAAK,CAAC,IAAI,EAAE,WAAW,CAAC,CAAC;YAAE,KAAK,CAAC,IAAI,CAAC,KAAK,CAAC,IAAI,CAAC,CAAC;IACpH,CAAC;IACD,OAAO,KAAK,CAAC;AACf,CAAC;AAED,KAAK,UAAU,aAAa,CAAC,IAAY;IACvC,IAAI,CAAC;QACH,OAAO,CAAC,MAAM,IAAI,CAAC,IAAI,CAAC,CAAC,CAAC,MAAM,EAAE,CAAC;IACrC,CAAC;IAAC,OAAO,KAAK,EAAE,CAAC;QACf,IAAK,KAA+B,CAAC,IAAI,KAAK,QAAQ;YAAE,OAAO,KAAK,CAAC;QACrE,MAAM,KAAK,CAAC;IACd,CAAC;AACH,CAAC;AAED,SAAS,eAAe,CAAC,KAAc,EAAE,IAAY;IACnD,MAAM,MAAM,GAAG,MAAM,CAAC,KAAK,CAAC,CAAC;IAC7B,IAAI,CAAC,MAAM,CAAC,SAAS,CAAC,MAAM,CAAC,IAAI,MAAM,IAAI,CAAC;QAAE,MAAM,YAAY,CAAC,GAAG,IAAI,6BAA6B,CAAC,CAAC;IACvG,OAAO,MAAM,CAAC;AAChB,CAAC;AAED,SAAS,kBAAkB,CAAC,KAAc,EAAE,IAAY;IACtD,MAAM,MAAM,GAAG,MAAM,CAAC,KAAK,CAAC,CAAC;IAC7B,IAAI,CAAC,MAAM,CAAC,aAAa,CAAC,MAAM,CAAC,IAAI,MAAM,GAAG,CAAC;QAAE,MAAM,YAAY,CAAC,GAAG,IAAI,iCAAiC,CAAC,CAAC;IAC9G,OAAO,MAAM,CAAC;AAChB,CAAC;AAED,SAAS,iBAAiB,CAAC,KAAc,EAAE,IAAY;IACrD,MAAM,MAAM,GAAG,MAAM,CAAC,KAAK,CAAC,CAAC;IAC7B,IAAI,CAAC,MAAM,CAAC,QAAQ,CAAC,MAAM,CAAC,IAAI,MAAM,GAAG,CAAC;QAAE,MAAM,YAAY,CAAC,GAAG,IAAI,gCAAgC,CAAC,CAAC;IACxG,OAAO,MAAM,CAAC;AAChB,CAAC;AACD,OAAO,EAAE,UAAU,EAAE,MAAM,aAAa,CAAC;AACzC,OAAO,EAAE,OAAO,EAAE,IAAI,EAAE,MAAM,kBAAkB,CAAC;AACjD,OAAO,IAAI,MAAM,WAAW,CAAC;AAE7B,OAAO,EAAE,cAAc,EAAE,YAAY,EAAE,MAAM,wBAAwB,CAAC;AACtE,OAAO,EAAE,gCAAgC,EAAE,qBAAqB,EAAE,MAAM,uBAAuB,CAAC;AAChG,OAAO,EAAE,eAAe,EAAE,MAAM,wBAAwB,CAAC;AACzD,OAAO,EAAE,4BAA4B,EAAmC,MAAM,iCAAiC,CAAC;AAChH,OAAO,EAAE,gCAAgC,EAAE,MAAM,0BAA0B,CAAC"}
|
|
1
|
+
{"version":3,"file":"request.js","sourceRoot":"","sources":["../../../src/evals/request.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,qBAAqB,EAAE,MAAM,oBAAoB,CAAC;AAC3D,MAAM,CAAC,MAAM,uBAAuB,GAAG,EAAE,GAAG,EAAE,GAAG,KAAK,CAAC;AACvD,MAAM,CAAC,MAAM,6BAA6B,GAAG,EAAE,GAAG,EAAE,GAAG,KAAK,CAAC;AAC7D,MAAM,CAAC,MAAM,8BAA8B,GAAG,CAAC,CAAC;AAChD,MAAM,CAAC,MAAM,uCAAuC,GAAG,KAAK,CAAC;AAE7D,MAAM,sBAAsB,GAAqC;IAC/D,KAAK,EAAE,4CAA4C;IACnD,QAAQ,EAAE,gCAAgC;CAC3C,CAAC;AAEF,MAAM,UAAU,SAAS;IACvB,OAAO,QAAQ,UAAU,EAAE,CAAC,UAAU,CAAC,GAAG,EAAE,EAAE,CAAC,EAAY,CAAC;AAC9D,CAAC;AAED,MAAM,UAAU,cAAc,CAAC,KAAa;IAC1C,IAAI,CAAC,qBAAqB,CAAC,IAAI,CAAC,KAAK,CAAC;QAAE,MAAM,YAAY,CAAC,oBAAoB,KAAK,EAAE,CAAC,CAAC;IACxF,OAAO,KAAe,CAAC;AACzB,CAAC;AAoBD,MAAM,CAAC,KAAK,UAAU,mBAAmB,CAAC,KAAuB;IAC/D,IAAI,CAAC,KAAK,IAAI,OAAO,KAAK,KAAK,QAAQ,IAAI,KAAK,CAAC,OAAO,CAAC,KAAK,CAAC;QAAE,MAAM,YAAY,CAAC,oCAAoC,CAAC,CAAC;IAC1H,MAAM,OAAO,GAAG,IAAI,GAAG,CAAC;QACtB,gBAAgB,EAAE,SAAS,EAAE,SAAS,EAAE,aAAa,EAAE,OAAO,EAAE,UAAU;QAC1E,gBAAgB,EAAE,wBAAwB,EAAE,iCAAiC;QAC7E,YAAY,EAAE,kBAAkB,EAAE,YAAY,EAAE,UAAU,EAAE,iBAAiB,EAAE,kBAAkB;KAClG,CAAC,CAAC;IACH,MAAM,UAAU,GAAG,MAAM,CAAC,IAAI,CAAC,KAAK,CAAC,CAAC,IAAI,CAAC,CAAC,KAAK,EAAE,EAAE,CAAC,CAAC,OAAO,CAAC,GAAG,CAAC,KAAK,CAAC,CAAC,CAAC;IAC3E,IAAI,UAAU;QAAE,MAAM,YAAY,CAAC,+BAA+B,UAAU,EAAE,CAAC,CAAC;IAChF,IAAI,KAAK,CAAC,cAAc,KAAK,SAAS,IAAI,KAAK,CAAC,cAAc,KAAK,cAAc;QAAE,MAAM,YAAY,CAAC,+BAA+B,KAAK,CAAC,cAAc,EAAE,CAAC,CAAC;IAC7J,IAAI,CAAC,KAAK,CAAC,OAAO,IAAI,QAAQ,CAAC,KAAK,QAAQ;QAAE,MAAM,YAAY,CAAC,qDAAqD,CAAC,CAAC;IACxH,IAAI,OAAO,KAAK,CAAC,OAAO,KAAK,QAAQ,IAAI,CAAC,KAAK,CAAC,OAAO,CAAC,IAAI,EAAE;QAAE,MAAM,YAAY,CAAC,oCAAoC,CAAC,CAAC;IACzH,IAAI,OAAO,KAAK,CAAC,WAAW,KAAK,QAAQ,IAAI,CAAC,KAAK,CAAC,WAAW,CAAC,IAAI,EAAE;QAAE,MAAM,YAAY,CAAC,wCAAwC,CAAC,CAAC;IACrI,MAAM,SAAS,GAAG,qBAAqB,CAAC,KAAK,CAAC,WAAW,CAAC,CAAC;IAC3D,IAAI,SAAS,CAAC,QAAQ,CAAC,IAAI,KAAK,WAAW,EAAE,CAAC;QAC5C,MAAM,YAAY,CAAC,yEAAyE,CAAC,CAAC;IAChG,CAAC;IACD,gCAAgC,CAAC,SAAS,CAAC,CAAC;IAC5C,MAAM,SAAS,GAAG,MAAM,uBAAuB,CAAC,KAAK,CAAC,OAAO,CAAC,IAAI,EAAE,CAAC,CAAC;IACtE,MAAM,QAAQ,GAAG,eAAe,CAAC,KAAK,CAAC,QAAQ,IAAI,CAAC,EAAE,UAAU,CAAC,CAAC;IAClE,MAAM,aAAa,GAAG,eAAe,CAAC,KAAK,CAAC,cAAc,IAAI,CAAC,EAAE,gBAAgB,CAAC,CAAC;IACnF,MAAM,qBAAqB,GAAG,kBAAkB,CAC9C,KAAK,CAAC,sBAAsB,IAAI,8BAA8B,EAC9D,wBAAwB,CACzB,CAAC;IACF,MAAM,0BAA0B,GAAG,iBAAiB,CAClD,KAAK,CAAC,+BAA+B,IAAI,uCAAuC,EAChF,iCAAiC,CAClC,CAAC;IACF,MAAM,OAAO,GAAG,iBAAiB,CAAC,KAAK,CAAC,UAAU,IAAI,CAAC,SAAS,CAAC,QAAQ,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,uBAAuB,CAAC,EAAE,YAAY,CAAC,CAAC;IACxH,MAAM,YAAY,GAAG,iBAAiB,CAAC,KAAK,CAAC,gBAAgB,IAAI,6BAA6B,EAAE,kBAAkB,CAAC,CAAC;IACpH,IAAI,KAAK,CAAC,KAAK,KAAK,SAAS,IAAI,OAAO,KAAK,CAAC,KAAK,KAAK,QAAQ;QAAE,MAAM,YAAY,CAAC,wBAAwB,CAAC,CAAC;IAC/G,IAAI,KAAK,CAAC,UAAU,KAAK,SAAS,IAAI,CAAC,CAAC,KAAK,CAAC,OAAO,CAAC,KAAK,CAAC,UAAU,CAAC,IAAI,KAAK,CAAC,UAAU,CAAC,IAAI,CAAC,CAAC,KAAK,EAAE,EAAE,CAAC,OAAO,KAAK,KAAK,QAAQ,CAAC,CAAC,EAAE,CAAC;QACxI,MAAM,YAAY,CAAC,wCAAwC,CAAC,CAAC;IAC/D,CAAC;IACD,IAAI,KAAK,CAAC,QAAQ,KAAK,SAAS,IAAI,CAAC,CAAC,KAAK,CAAC,OAAO,CAAC,KAAK,CAAC,QAAQ,CAAC,IAAI,KAAK,CAAC,QAAQ,CAAC,IAAI,CAAC,CAAC,KAAK,EAAE,EAAE,CAAC,OAAO,KAAK,KAAK,QAAQ,CAAC,CAAC,EAAE,CAAC;QAClI,MAAM,YAAY,CAAC,sCAAsC,CAAC,CAAC;IAC7D,CAAC;IACD,MAAM,cAAc,GAAG,0BAA0B,CAAC,KAAK,CAAC,eAAe,EAAE,OAAO,KAAK,CAAC,KAAK,KAAK,QAAQ,CAAC,CAAC,CAAC,KAAK,CAAC,KAAK,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC;IAC7H,MAAM,QAAQ,GAAG,KAAK,CAAC,gBAAgB,KAAK,SAAS,CAAC,CAAC,CAAC,SAAS,CAAC,CAAC,CAAC,oBAAoB,CAAC,KAAK,CAAC,gBAAgB,CAAC,CAAC;IACjH,IAAI,QAAQ,IAAI,CAAC,cAAc,IAAI,KAAK,CAAC,KAAK,KAAK,YAAY,QAAQ,CAAC,SAAS,EAAE,IAAI,QAAQ,KAAK,CAAC,IAAI,qBAAqB,KAAK,CAAC;WAC/H,SAAS,CAAC,UAAU,KAAK,eAAe,IAAK,KAAK,CAAC,UAAoC,EAAE,MAAM,IAAK,KAAK,CAAC,QAAkC,EAAE,MAAM,CAAC,EAAE,CAAC;QAC3J,MAAM,YAAY,CAAC,4JAA4J,CAAC,CAAC;IACnL,CAAC;IACD,IAAI,CAAC,QAAQ,IAAI,MAAM,CAAC,KAAK,CAAC,KAAK,CAAC,CAAC,UAAU,CAAC,WAAW,CAAC;QAAE,MAAM,YAAY,CAAC,sDAAsD,CAAC,CAAC;IACzI,MAAM,SAAS,GAAG,KAAK,CAAC,OAAO,CAAC,KAAK,CAAC,UAAU,CAAC,CAAC,CAAC,CAAC,CAAC,GAAG,KAAK,CAAC,UAAU,CAAa,CAAC,CAAC,CAAC,EAAE,CAAC;IAC3F,IAAI,SAAS,CAAC,QAAQ,EAAE,CAAC;QACvB,MAAM,gCAAgC,CAAC,KAAK,CAAC,OAAO,CAAC,IAAI,EAAE,EAAE,SAAS,CAAC,QAAQ,EAAE,SAAS,CAAC,UAAU,EAAE,SAAS,CAAC,CAAC;IACpH,CAAC;IACD,MAAM,SAAS,GAAG,sBAAsB,CAAC,SAAS,CAAC,UAAU,CAAC,CAAC;IAC/D,IAAI,SAAS,IAAI,CAAC,SAAS,CAAC,QAAQ,CAAC,SAAS,CAAC;QAAE,SAAS,CAAC,OAAO,CAAC,SAAS,CAAC,CAAC;IAC9E,OAAO;QACL,cAAc,EAAE,cAAc;QAC9B,OAAO,EAAE,QAAQ;QACjB,OAAO,EAAE,MAAM,CAAC,KAAK,CAAC,OAAO,CAAC,CAAC,IAAI,EAAE;QACrC,WAAW,EAAE,SAAS,CAAC,SAAS;QAChC,KAAK,EAAE,OAAO,KAAK,CAAC,KAAK,KAAK,QAAQ,CAAC,CAAC,CAAC,KAAK,CAAC,KAAK,CAAC,CAAC,CAAC,EAAE;QACzD,QAAQ;QACR,cAAc,EAAE,aAAa;QAC7B,sBAAsB,EAAE,qBAAqB;QAC7C,+BAA+B,EAAE,0BAA0B;QAC3D,UAAU,EAAE,OAAO;QACnB,gBAAgB,EAAE,YAAY;QAC9B,UAAU,EAAE,SAAS;QACrB,QAAQ,EAAE,CAAC,GAAG,IAAI,GAAG,CAAC,KAAK,CAAC,OAAO,CAAC,KAAK,CAAC,QAAQ,CAAC,CAAC,CAAC,CAAC,KAAK,CAAC,QAAoB,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC;QACvF,YAAY,EAAE,SAAS,CAAC,YAAY;QACpC,kBAAkB,EAAE,SAAS,CAAC,kBAAkB;QAChD,GAAG,CAAC,cAAc,CAAC,CAAC,CAAC,EAAE,eAAe,EAAE,cAAc,EAAE,CAAC,CAAC,CAAC,EAAE,CAAC;QAC9D,GAAG,CAAC,QAAQ,CAAC,CAAC,CAAC,EAAE,gBAAgB,EAAE,QAAQ,EAAE,CAAC,CAAC,CAAC,EAAE,CAAC;KACpD,CAAC;AACJ,CAAC;AAED,SAAS,0BAA0B,CAAC,KAAc,EAAE,KAAa;IAC/D,MAAM,UAAU,GAAG,KAAK,CAAC,UAAU,CAAC,QAAQ,CAAC,CAAC;IAC9C,IAAI,CAAC,UAAU,EAAE,CAAC;QAChB,IAAI,KAAK,KAAK,SAAS;YAAE,MAAM,YAAY,CAAC,mFAAmF,CAAC,CAAC;QACjI,OAAO,SAAS,CAAC;IACnB,CAAC;IACD,IAAI,KAAK,KAAK,SAAS;QAAE,OAAO,EAAE,KAAK,EAAE,MAAM,EAAE,MAAM,EAAE,OAAO,EAAE,UAAU,EAAE,OAAO,EAAE,KAAK,EAAE,CAAC;IAC/F,IAAI,CAAC,KAAK,IAAI,OAAO,KAAK,KAAK,QAAQ,IAAI,KAAK,CAAC,OAAO,CAAC,KAAK,CAAC;QAAE,MAAM,YAAY,CAAC,mCAAmC,CAAC,CAAC;IACzH,MAAM,MAAM,GAAG,KAAgC,CAAC;IAChD,MAAM,OAAO,GAAG,IAAI,GAAG,CAAC,CAAC,OAAO,EAAE,QAAQ,EAAE,SAAS,EAAE,cAAc,EAAE,SAAS,EAAE,YAAY,CAAC,CAAC,CAAC;IACjG,MAAM,UAAU,GAAG,MAAM,CAAC,IAAI,CAAC,MAAM,CAAC,CAAC,IAAI,CAAC,CAAC,KAAK,EAAE,EAAE,CAAC,CAAC,OAAO,CAAC,GAAG,CAAC,KAAK,CAAC,CAAC,CAAC;IAC5E,IAAI,UAAU;QAAE,MAAM,YAAY,CAAC,sCAAsC,UAAU,EAAE,CAAC,CAAC;IACvF,IAAI,MAAM,CAAC,KAAK,KAAK,SAAS,IAAI,MAAM,CAAC,KAAK,KAAK,KAAK;QAAE,MAAM,YAAY,CAAC,wCAAwC,CAAC,CAAC;IACvH,MAAM,SAAS,GAAG,MAAM,CAAC,UAAU,KAAK,SAAS,CAAC,CAAC,CAAC,SAAS,CAAC,CAAC,CAAC,qBAAqB,CAAC,MAAM,CAAC,UAAU,CAAC,CAAC;IACzG,IAAI,SAAS,IAAI,MAAM,CAAC,YAAY,KAAK,SAAS;QAAE,MAAM,IAAI,SAAS,CAAC,wDAAwD,CAAC,CAAC;IAClI,MAAM,uBAAuB,GAAG,SAAS,IAAI,CAAC,MAAM,CAAC,MAAM,KAAK,SAAS,IAAI,MAAM,CAAC,MAAM,KAAK,MAAM,CAAC,IAAI,CAAC,MAAM,CAAC,OAAO,KAAK,SAAS,IAAI,MAAM,CAAC,OAAO,KAAK,UAAU,CAAC,CAAC;IAC1K,IAAI,MAAM,CAAC,YAAY,KAAK,SAAS,IAAI,CAAC,uBAAuB,IAAI,CAAC,MAAM,CAAC,MAAM,KAAK,SAAS,IAAI,MAAM,CAAC,OAAO,KAAK,SAAS,CAAC,EAAE,CAAC;QACnI,MAAM,YAAY,CAAC,iEAAiE,CAAC,CAAC;IACxF,CAAC;IACD,MAAM,MAAM,GAAG,MAAM,CAAC,MAAM,IAAI,MAAM,CAAC;IACvC,MAAM,OAAO,GAAG,MAAM,CAAC,OAAO,IAAI,UAAU,CAAC;IAC7C,IAAI,CAAC,IAAI,GAAG,CAAC,CAAC,MAAM,EAAE,KAAK,EAAE,MAAM,EAAE,OAAO,CAAC,CAAC,CAAC,GAAG,CAAC,MAAM,CAAC,MAAM,CAAC,CAAC;QAAE,MAAM,YAAY,CAAC,0DAA0D,CAAC,CAAC;IACnJ,IAAI,OAAO,KAAK,UAAU,IAAI,OAAO,KAAK,YAAY;QAAE,MAAM,YAAY,CAAC,wDAAwD,CAAC,CAAC;IACrI,IAAI,MAAM,CAAC,OAAO,KAAK,SAAS,IAAI,OAAO,MAAM,CAAC,OAAO,KAAK,SAAS;QAAE,MAAM,YAAY,CAAC,2CAA2C,CAAC,CAAC;IACzI,IAAI,MAAM,CAAC,YAAY,KAAK,SAAS,IAAI,CAAC,OAAO,MAAM,CAAC,YAAY,KAAK,QAAQ,IAAI,CAAC,uBAAuB,CAAC,IAAI,CAAC,MAAM,CAAC,YAAY,CAAC,CAAC,EAAE,CAAC;QACzI,MAAM,YAAY,CAAC,uDAAuD,CAAC,CAAC;IAC9E,CAAC;IACD,OAAO;QACL,KAAK;QACL,MAAM,EAAE,MAA2D;QACnE,OAAO;QACP,OAAO,EAAE,MAAM,CAAC,OAAO,KAAK,IAAI;QAChC,GAAG,CAAC,SAAS,CAAC,CAAC,CAAC,EAAE,UAAU,EAAE,SAAS,EAAE,CAAC,CAAC,CAAC,EAAE,CAAC;QAC/C,GAAG,CAAC,MAAM,CAAC,YAAY,KAAK,SAAS,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,EAAE,YAAY,EAAE,MAAM,CAAC,YAAmD,EAAE,CAAC;KAC3H,CAAC;AACJ,CAAC;AAED,MAAM,CAAC,KAAK,UAAU,yBAAyB,CAAC,OAAe;IAC7D,MAAM,EAAE,YAAY,EAAE,kBAAkB,EAAE,GAAG,MAAM,uBAAuB,CAAC,OAAO,CAAC,CAAC;IACpF,OAAO,EAAE,YAAY,EAAE,kBAAkB,EAAE,CAAC;AAC9C,CAAC;AAED,KAAK,UAAU,uBAAuB,CAAC,OAAe;IAKpD,MAAM,GAAG,GAAG,OAAO,CAAC,IAAI,EAAE,CAAC;IAC3B,MAAM,KAAK,GAAG,IAAI,CAAC,OAAO,CAAC,GAAG,CAAC,CAAC;IAChC,IAAI,CAAC;QACH,IAAI,CAAC,MAAM,IAAI,CAAC,KAAK,CAAC,CAAC,CAAC,WAAW,EAAE,EAAE,CAAC;YACtC,MAAM,QAAQ,GAAG,MAAM,4BAA4B,CAAC,KAAK,CAAC,CAAC;YAC3D,IAAI,QAAQ,EAAE,CAAC;gBACb,OAAO;oBACL,YAAY,EAAE,QAAQ,CAAC,SAAS,CAAC,EAAE;oBACnC,kBAAkB,EAAE,QAAQ,CAAC,cAAc;oBAC3C,QAAQ;iBACT,CAAC;YACJ,CAAC;YACD,OAAO;gBACL,YAAY,EAAE,SAAS,IAAI,CAAC,QAAQ,CAAC,KAAK,CAAC,EAAE;gBAC7C,kBAAkB,EAAE,MAAM,eAAe,CAAC,KAAK,EAAE,EAAE,gBAAgB,EAAE,IAAI,GAAG,CAAC,CAAC,MAAM,CAAC,CAAC,EAAE,CAAC;gBACzF,QAAQ,EAAE,IAAI;aACf,CAAC;QACJ,CAAC;IACH,CAAC;IAAC,OAAO,KAAK,EAAE,CAAC;QACf,IAAK,KAA+B,CAAC,IAAI,KAAK,QAAQ;YAAE,MAAM,KAAK,CAAC;IACtE,CAAC;IACD,MAAM,SAAS,GAAG,GAAG,CAAC,WAAW,CAAC,GAAG,CAAC,CAAC;IACvC,IAAI,SAAS,IAAI,CAAC,IAAI,SAAS,KAAK,GAAG,CAAC,MAAM,GAAG,CAAC,EAAE,CAAC;QACnD,MAAM,YAAY,CAAC,wEAAwE,CAAC,CAAC;IAC/F,CAAC;IACD,MAAM,WAAW,GAAG,GAAG,CAAC,KAAK,CAAC,CAAC,EAAE,SAAS,CAAC,CAAC;IAC5C,MAAM,QAAQ,GAAG,GAAG,CAAC,KAAK,CAAC,SAAS,GAAG,CAAC,CAAC,CAAC;IAC1C,IAAI,QAAQ,CAAC,WAAW,EAAE,KAAK,QAAQ;QAAE,MAAM,YAAY,CAAC,mCAAmC,CAAC,CAAC;IACjG,OAAO,EAAE,YAAY,EAAE,WAAW,EAAE,kBAAkB,EAAE,QAAQ,EAAE,QAAQ,EAAE,IAAI,EAAE,CAAC;AACrF,CAAC;AAED,MAAM,CAAC,KAAK,UAAU,0BAA0B,CAAC,OAAe;IAC9D,MAAM,KAAK,GAAG,IAAI,CAAC,OAAO,CAAC,OAAO,CAAC,IAAI,EAAE,CAAC,CAAC;IAC3C,IAAI,CAAC;QACH,IAAI,CAAC,CAAC,MAAM,IAAI,CAAC,KAAK,CAAC,CAAC,CAAC,WAAW,EAAE;YAAE,OAAO,IAAI,CAAC;IACtD,CAAC;IAAC,OAAO,KAAK,EAAE,CAAC;QACf,IAAK,KAA+B,CAAC,IAAI,KAAK,QAAQ;YAAE,OAAO,IAAI,CAAC;QACpE,MAAM,KAAK,CAAC;IACd,CAAC;IACD,IAAI,MAAM,aAAa,CAAC,IAAI,CAAC,IAAI,CAAC,KAAK,EAAE,WAAW,CAAC,CAAC;QAAE,OAAO,CAAC,IAAI,CAAC,QAAQ,CAAC,KAAK,CAAC,CAAC,CAAC;IACtF,MAAM,OAAO,GAAG,MAAM,OAAO,CAAC,KAAK,EAAE,EAAE,aAAa,EAAE,IAAI,EAAE,CAAC,CAAC;IAC9D,MAAM,KAAK,GAAa,EAAE,CAAC;IAC3B,KAAK,MAAM,KAAK,IAAI,OAAO,CAAC,IAAI,CAAC,CAAC,IAAI,EAAE,KAAK,EAAE,EAAE,CAAC,IAAI,CAAC,IAAI,CAAC,aAAa,CAAC,KAAK,CAAC,IAAI,CAAC,CAAC,EAAE,CAAC;QACvF,IAAI,KAAK,CAAC,WAAW,EAAE,IAAI,MAAM,aAAa,CAAC,IAAI,CAAC,IAAI,CAAC,KAAK,EAAE,KAAK,CAAC,IAAI,EAAE,WAAW,CAAC,CAAC;YAAE,KAAK,CAAC,IAAI,CAAC,KAAK,CAAC,IAAI,CAAC,CAAC;IACpH,CAAC;IACD,OAAO,KAAK,CAAC;AACf,CAAC;AAED,KAAK,UAAU,aAAa,CAAC,IAAY;IACvC,IAAI,CAAC;QACH,OAAO,CAAC,MAAM,IAAI,CAAC,IAAI,CAAC,CAAC,CAAC,MAAM,EAAE,CAAC;IACrC,CAAC;IAAC,OAAO,KAAK,EAAE,CAAC;QACf,IAAK,KAA+B,CAAC,IAAI,KAAK,QAAQ;YAAE,OAAO,KAAK,CAAC;QACrE,MAAM,KAAK,CAAC;IACd,CAAC;AACH,CAAC;AAED,SAAS,eAAe,CAAC,KAAc,EAAE,IAAY;IACnD,MAAM,MAAM,GAAG,MAAM,CAAC,KAAK,CAAC,CAAC;IAC7B,IAAI,CAAC,MAAM,CAAC,SAAS,CAAC,MAAM,CAAC,IAAI,MAAM,IAAI,CAAC;QAAE,MAAM,YAAY,CAAC,GAAG,IAAI,6BAA6B,CAAC,CAAC;IACvG,OAAO,MAAM,CAAC;AAChB,CAAC;AAED,SAAS,kBAAkB,CAAC,KAAc,EAAE,IAAY;IACtD,MAAM,MAAM,GAAG,MAAM,CAAC,KAAK,CAAC,CAAC;IAC7B,IAAI,CAAC,MAAM,CAAC,aAAa,CAAC,MAAM,CAAC,IAAI,MAAM,GAAG,CAAC;QAAE,MAAM,YAAY,CAAC,GAAG,IAAI,iCAAiC,CAAC,CAAC;IAC9G,OAAO,MAAM,CAAC;AAChB,CAAC;AAED,SAAS,iBAAiB,CAAC,KAAc,EAAE,IAAY;IACrD,MAAM,MAAM,GAAG,MAAM,CAAC,KAAK,CAAC,CAAC;IAC7B,IAAI,CAAC,MAAM,CAAC,QAAQ,CAAC,MAAM,CAAC,IAAI,MAAM,GAAG,CAAC;QAAE,MAAM,YAAY,CAAC,GAAG,IAAI,gCAAgC,CAAC,CAAC;IACxG,OAAO,MAAM,CAAC;AAChB,CAAC;AACD,OAAO,EAAE,UAAU,EAAE,MAAM,aAAa,CAAC;AACzC,OAAO,EAAE,OAAO,EAAE,IAAI,EAAE,MAAM,kBAAkB,CAAC;AACjD,OAAO,IAAI,MAAM,WAAW,CAAC;AAE7B,OAAO,EAAE,cAAc,EAAE,YAAY,EAAE,MAAM,wBAAwB,CAAC;AACtE,OAAO,EAAE,gCAAgC,EAAE,qBAAqB,EAAE,MAAM,uBAAuB,CAAC;AAChG,OAAO,EAAE,eAAe,EAAE,MAAM,wBAAwB,CAAC;AACzD,OAAO,EAAE,4BAA4B,EAAmC,MAAM,iCAAiC,CAAC;AAChH,OAAO,EAAE,gCAAgC,EAAE,MAAM,0BAA0B,CAAC;AAC5E,OAAO,EAAE,oBAAoB,EAAE,MAAM,0BAA0B,CAAC"}
|
|
@@ -0,0 +1,39 @@
|
|
|
1
|
+
import { rm } from "node:fs/promises";
|
|
2
|
+
import path from "node:path";
|
|
3
|
+
import { HitchError, atomicWriteJSON, readJSON, sha256JSON } from "../foundation/index.js";
|
|
4
|
+
import { evalRerunSemantics } from "./rerun-types.js";
|
|
5
|
+
const file = (directory) => path.join(directory, "completion-pending.json");
|
|
6
|
+
export async function stageRemoteRerunCompletion(evalDirectory, directory, result) {
|
|
7
|
+
const source = await readJSON(path.join(evalDirectory, "result.json"));
|
|
8
|
+
if (source.status !== result.eval_status)
|
|
9
|
+
throw invalid();
|
|
10
|
+
await atomicWriteJSON(file(directory), { schema_version: "2", result, result_digest: sha256JSON(result), source_result_digest: sha256JSON(source) });
|
|
11
|
+
}
|
|
12
|
+
export async function readRemoteRerunCompletion(directory, evalId, rerunId) {
|
|
13
|
+
const record = await readJSON(file(directory), null);
|
|
14
|
+
if (!record)
|
|
15
|
+
return null;
|
|
16
|
+
const result = record.result;
|
|
17
|
+
const request = await readJSON(path.join(directory, "request.json"), null);
|
|
18
|
+
if (record.schema_version !== "2" || !result || record.result_digest !== sha256JSON(result)
|
|
19
|
+
|| typeof record.source_result_digest !== "string" || !/^sha256:[a-f0-9]{64}$/.test(record.source_result_digest)
|
|
20
|
+
|| result.schema_version !== "1" || result.kind !== "eval-rerun" || result.eval_id !== evalId || result.rerun_id !== rerunId
|
|
21
|
+
|| !["candidate-restart", "verifier-only"].includes(result.rerun_type) || result.status !== "completed" || !["succeeded", "failed"].includes(result.eval_status)
|
|
22
|
+
|| sha256JSON(result.semantics) !== sha256JSON(evalRerunSemantics(result.rerun_type))
|
|
23
|
+
|| !Number.isFinite(Date.parse(result.completed_at)) || !request || request.schema_version !== "1"
|
|
24
|
+
|| request.eval_id !== evalId || request.rerun_id !== rerunId || request.rerun_type !== result.rerun_type
|
|
25
|
+
|| sha256JSON(request.semantics) !== sha256JSON(result.semantics)
|
|
26
|
+
|| sha256JSON(request.trials) !== sha256JSON(result.selected_trials) || sha256JSON(request.tasks) !== sha256JSON(result.selected_tasks))
|
|
27
|
+
throw invalid();
|
|
28
|
+
return record;
|
|
29
|
+
}
|
|
30
|
+
export async function acknowledgeRemoteRerunCompletion(directory, result) {
|
|
31
|
+
const record = await readRemoteRerunCompletion(directory, result.eval_id, result.rerun_id);
|
|
32
|
+
if (!record)
|
|
33
|
+
return;
|
|
34
|
+
if (record.result_digest !== sha256JSON(result))
|
|
35
|
+
throw invalid();
|
|
36
|
+
await rm(file(directory));
|
|
37
|
+
}
|
|
38
|
+
function invalid() { return new HitchError("remote rerun completion differs from its frozen result or selection", { code: "execution_state_ambiguous", exitCode: 12 }); }
|
|
39
|
+
//# sourceMappingURL=rerun-completion.js.map
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"rerun-completion.js","sourceRoot":"","sources":["../../../src/evals/rerun-completion.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,EAAE,EAAE,MAAM,kBAAkB,CAAC;AACtC,OAAO,IAAI,MAAM,WAAW,CAAC;AAC7B,OAAO,EAAE,UAAU,EAAE,eAAe,EAAE,QAAQ,EAAE,UAAU,EAAE,MAAM,wBAAwB,CAAC;AAC3F,OAAO,EAAE,kBAAkB,EAAE,MAAM,kBAAkB,CAAC;AAItD,MAAM,IAAI,GAAG,CAAC,SAAiB,EAAE,EAAE,CAAC,IAAI,CAAC,IAAI,CAAC,SAAS,EAAE,yBAAyB,CAAC,CAAC;AAEpF,MAAM,CAAC,KAAK,UAAU,0BAA0B,CAAC,aAAqB,EAAE,SAAiB,EAAE,MAAuB;IAChH,MAAM,MAAM,GAAG,MAAM,QAAQ,CAA0B,IAAI,CAAC,IAAI,CAAC,aAAa,EAAE,aAAa,CAAC,CAAC,CAAC;IAChG,IAAI,MAAM,CAAC,MAAM,KAAK,MAAM,CAAC,WAAW;QAAE,MAAM,OAAO,EAAE,CAAC;IAC1D,MAAM,eAAe,CAAC,IAAI,CAAC,SAAS,CAAC,EAAE,EAAE,cAAc,EAAE,GAAG,EAAE,MAAM,EAAE,aAAa,EAAE,UAAU,CAAC,MAAM,CAAC,EAAE,oBAAoB,EAAE,UAAU,CAAC,MAAM,CAAC,EAAE,CAAC,CAAC;AACvJ,CAAC;AAED,MAAM,CAAC,KAAK,UAAU,yBAAyB,CAAC,SAAiB,EAAE,MAAc,EAAE,OAAe;IAChG,MAAM,MAAM,GAAG,MAAM,QAAQ,CAAoB,IAAI,CAAC,SAAS,CAAC,EAAE,IAAI,CAAC,CAAC;IACxE,IAAI,CAAC,MAAM;QAAE,OAAO,IAAI,CAAC;IACzB,MAAM,MAAM,GAAG,MAAM,CAAC,MAAM,CAAC;IAC7B,MAAM,OAAO,GAAG,MAAM,QAAQ,CAAiC,IAAI,CAAC,IAAI,CAAC,SAAS,EAAE,cAAc,CAAC,EAAE,IAAI,CAAC,CAAC;IAC3G,IAAI,MAAM,CAAC,cAAc,KAAK,GAAG,IAAI,CAAC,MAAM,IAAI,MAAM,CAAC,aAAa,KAAK,UAAU,CAAC,MAAM,CAAC;WACtF,OAAO,MAAM,CAAC,oBAAoB,KAAK,QAAQ,IAAI,CAAC,uBAAuB,CAAC,IAAI,CAAC,MAAM,CAAC,oBAAoB,CAAC;WAC7G,MAAM,CAAC,cAAc,KAAK,GAAG,IAAI,MAAM,CAAC,IAAI,KAAK,YAAY,IAAI,MAAM,CAAC,OAAO,KAAK,MAAM,IAAI,MAAM,CAAC,QAAQ,KAAK,OAAO;WACzH,CAAC,CAAC,mBAAmB,EAAE,eAAe,CAAC,CAAC,QAAQ,CAAC,MAAM,CAAC,UAAU,CAAC,IAAI,MAAM,CAAC,MAAM,KAAK,WAAW,IAAI,CAAC,CAAC,WAAW,EAAE,QAAQ,CAAC,CAAC,QAAQ,CAAC,MAAM,CAAC,WAAW,CAAC;WAC7J,UAAU,CAAC,MAAM,CAAC,SAAS,CAAC,KAAK,UAAU,CAAC,kBAAkB,CAAC,MAAM,CAAC,UAAU,CAAC,CAAC;WAClF,CAAC,MAAM,CAAC,QAAQ,CAAC,IAAI,CAAC,KAAK,CAAC,MAAM,CAAC,YAAY,CAAC,CAAC,IAAI,CAAC,OAAO,IAAI,OAAO,CAAC,cAAc,KAAK,GAAG;WAC/F,OAAO,CAAC,OAAO,KAAK,MAAM,IAAI,OAAO,CAAC,QAAQ,KAAK,OAAO,IAAI,OAAO,CAAC,UAAU,KAAK,MAAM,CAAC,UAAU;WACtG,UAAU,CAAC,OAAO,CAAC,SAAS,CAAC,KAAK,UAAU,CAAC,MAAM,CAAC,SAAS,CAAC;WAC9D,UAAU,CAAC,OAAO,CAAC,MAAM,CAAC,KAAK,UAAU,CAAC,MAAM,CAAC,eAAe,CAAC,IAAI,UAAU,CAAC,OAAO,CAAC,KAAK,CAAC,KAAK,UAAU,CAAC,MAAM,CAAC,cAAc,CAAC;QAAE,MAAM,OAAO,EAAE,CAAC;IAC3J,OAAO,MAAM,CAAC;AAChB,CAAC;AAED,MAAM,CAAC,KAAK,UAAU,gCAAgC,CAAC,SAAiB,EAAE,MAAuB;IAC/F,MAAM,MAAM,GAAG,MAAM,yBAAyB,CAAC,SAAS,EAAE,MAAM,CAAC,OAAO,EAAE,MAAM,CAAC,QAAQ,CAAC,CAAC;IAC3F,IAAI,CAAC,MAAM;QAAE,OAAO;IACpB,IAAI,MAAM,CAAC,aAAa,KAAK,UAAU,CAAC,MAAM,CAAC;QAAE,MAAM,OAAO,EAAE,CAAC;IACjE,MAAM,EAAE,CAAC,IAAI,CAAC,SAAS,CAAC,CAAC,CAAC;AAC5B,CAAC;AACD,SAAS,OAAO,KAAiB,OAAO,IAAI,UAAU,CAAC,qEAAqE,EAAE,EAAE,IAAI,EAAE,2BAA2B,EAAE,QAAQ,EAAE,EAAE,EAAE,CAAC,CAAC,CAAC,CAAC"}
|
|
@@ -0,0 +1,70 @@
|
|
|
1
|
+
import path from "node:path";
|
|
2
|
+
import { HitchError, readJSON, sha256JSON } from "../foundation/index.js";
|
|
3
|
+
import { candidateRestartWorkItem } from "./physical-work-plan.js";
|
|
4
|
+
import { evalRerunSemantics } from "./rerun-types.js";
|
|
5
|
+
import { slotKey, sortSlots, uniqueTasks } from "./rerun-slots.js";
|
|
6
|
+
import { readExecutionLeases } from "./execution-leases.js";
|
|
7
|
+
/** Restore the original selection; recovered publications must not select a new batch. */
|
|
8
|
+
export async function restoreRemoteRerunSelection(input) {
|
|
9
|
+
const type = input.rerunType ?? "candidate-restart";
|
|
10
|
+
const saved = await readJSON(path.join(input.directory, "request.json"), null);
|
|
11
|
+
if (!saved || saved.schema_version !== "1" || saved.eval_id !== input.evalId || saved.rerun_id !== input.rerunId
|
|
12
|
+
|| !["candidate-restart", "verifier-only"].includes(type) || saved.rerun_type !== type
|
|
13
|
+
|| saved.verifier_runtime_id !== input.verifierRuntimeId || saved.mode !== input.selector.mode
|
|
14
|
+
|| sha256JSON(saved.semantics) !== sha256JSON(evalRerunSemantics(type))
|
|
15
|
+
|| !Number.isSafeInteger(saved.base_generation) || saved.base_generation < 0
|
|
16
|
+
|| saved.base_generation > input.progress.generation
|
|
17
|
+
|| typeof saved.created_at !== "string" || !Number.isFinite(Date.parse(saved.created_at))
|
|
18
|
+
|| !Array.isArray(saved.trials))
|
|
19
|
+
throw ambiguous("remote rerun has no valid frozen selection");
|
|
20
|
+
const trials = saved.trials.map(value => {
|
|
21
|
+
if (!value || typeof value !== "object" || Array.isArray(value))
|
|
22
|
+
throw ambiguous("remote rerun selected trial is invalid");
|
|
23
|
+
const slot = value;
|
|
24
|
+
if (Object.keys(slot).length !== 2 || typeof slot.task_id !== "string" || !input.tasks.includes(slot.task_id)
|
|
25
|
+
|| !Number.isSafeInteger(slot.attempt) || slot.attempt < 1 || slot.attempt > input.attempts)
|
|
26
|
+
throw ambiguous("remote rerun selected trial differs from the frozen plan");
|
|
27
|
+
return { task_id: slot.task_id, attempt: slot.attempt };
|
|
28
|
+
});
|
|
29
|
+
if (new Set(trials.map(slotKey)).size !== trials.length || sha256JSON(trials) !== sha256JSON(sortSlots(trials))
|
|
30
|
+
|| sha256JSON(saved.tasks) !== sha256JSON(uniqueTasks(trials))
|
|
31
|
+
|| input.selector.mode === "tasks" && sha256JSON([...new Set(input.selector.taskNames)].sort()) !== sha256JSON(uniqueTasks(trials))) {
|
|
32
|
+
throw ambiguous("remote rerun selected tasks do not match its original request");
|
|
33
|
+
}
|
|
34
|
+
return { trials, startedAt: saved.created_at };
|
|
35
|
+
}
|
|
36
|
+
/** Inspect every work journal before starting a model service or dispatching anything. */
|
|
37
|
+
export async function remoteRerunNeedsExecution(input) {
|
|
38
|
+
const leases = await readExecutionLeases(input.evalDirectory);
|
|
39
|
+
let needsExecution = false;
|
|
40
|
+
for (const slot of input.slots) {
|
|
41
|
+
const logical = input.plan.slots.find(item => item.task_id === slot.task_id && item.attempt === slot.attempt);
|
|
42
|
+
const source = logical && input.plan.work_items.find(item => item.slots.includes(logical.slot_id));
|
|
43
|
+
if (!source || source.slots.length !== 1 || source.task_ids.length !== 1)
|
|
44
|
+
throw ambiguous("remote rerun selection has no single-task physical source");
|
|
45
|
+
const work = candidateRestartWorkItem(source, input.rerunId);
|
|
46
|
+
const identity = { schema_version: "1", rerun_id: input.rerunId, source_work_id: source.work_id,
|
|
47
|
+
work_id: work.work_id, plan_digest: sha256JSON(input.plan), request_digest: sha256JSON(input.request) };
|
|
48
|
+
const journal = await readJSON(path.join(input.directory, "remote-work", `${work.work_id}.json`), null);
|
|
49
|
+
if (!journal) {
|
|
50
|
+
needsExecution = true;
|
|
51
|
+
continue;
|
|
52
|
+
}
|
|
53
|
+
if (journal.state === "not-started" && sha256JSON(journal.identity) === sha256JSON(identity) && !journal.result
|
|
54
|
+
&& leases.some(lease => lease.lease_id === journal.lease_id && lease.work_id === work.work_id
|
|
55
|
+
&& lease.provider === input.plan.provider && lease.state === "released")) {
|
|
56
|
+
needsExecution = true;
|
|
57
|
+
continue;
|
|
58
|
+
}
|
|
59
|
+
if (sha256JSON(journal.identity) !== sha256JSON(identity) || journal.state !== "completed" || !journal.result
|
|
60
|
+
|| !leases.some(lease => lease.lease_id === journal.result.leaseId && lease.work_id === work.work_id
|
|
61
|
+
&& lease.provider === input.plan.provider && lease.state === "released"))
|
|
62
|
+
throw ambiguous("remote rerun work needs a matching collected result and confirmed lease release");
|
|
63
|
+
if (journal.result.run.backend.process_exit_code !== 0 || journal.result.run.rawResult === null) {
|
|
64
|
+
throw new HitchError("recovered remote Harbor rerun failed before producing a complete result", { code: "eval_rerun_harbor_failed", exitCode: 13 });
|
|
65
|
+
}
|
|
66
|
+
}
|
|
67
|
+
return needsExecution;
|
|
68
|
+
}
|
|
69
|
+
function ambiguous(message) { return new HitchError(message, { code: "execution_state_ambiguous", exitCode: 12 }); }
|
|
70
|
+
//# sourceMappingURL=rerun-resume.js.map
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"rerun-resume.js","sourceRoot":"","sources":["../../../src/evals/rerun-resume.ts"],"names":[],"mappings":"AAAA,OAAO,IAAI,MAAM,WAAW,CAAC;AAE7B,OAAO,EAAE,UAAU,EAAE,QAAQ,EAAE,UAAU,EAAE,MAAM,wBAAwB,CAAC;AAC1E,OAAO,EAAE,wBAAwB,EAAE,MAAM,yBAAyB,CAAC;AAEnE,OAAO,EAAE,kBAAkB,EAAE,MAAM,kBAAkB,CAAC;AACtD,OAAO,EAAE,OAAO,EAAE,SAAS,EAAE,WAAW,EAAE,MAAM,kBAAkB,CAAC;AAGnE,OAAO,EAAE,mBAAmB,EAAE,MAAM,uBAAuB,CAAC;AAE5D,0FAA0F;AAC1F,MAAM,CAAC,KAAK,UAAU,2BAA2B,CAAC,KAGjD;IACC,MAAM,IAAI,GAAG,KAAK,CAAC,SAAS,IAAI,mBAAmB,CAAC;IACpD,MAAM,KAAK,GAAG,MAAM,QAAQ,CAAiC,IAAI,CAAC,IAAI,CAAC,KAAK,CAAC,SAAS,EAAE,cAAc,CAAC,EAAE,IAAI,CAAC,CAAC;IAC/G,IAAI,CAAC,KAAK,IAAI,KAAK,CAAC,cAAc,KAAK,GAAG,IAAI,KAAK,CAAC,OAAO,KAAK,KAAK,CAAC,MAAM,IAAI,KAAK,CAAC,QAAQ,KAAK,KAAK,CAAC,OAAO;WAC3G,CAAC,CAAC,mBAAmB,EAAE,eAAe,CAAC,CAAC,QAAQ,CAAC,IAAI,CAAC,IAAI,KAAK,CAAC,UAAU,KAAK,IAAI;WACnF,KAAK,CAAC,mBAAmB,KAAK,KAAK,CAAC,iBAAiB,IAAI,KAAK,CAAC,IAAI,KAAK,KAAK,CAAC,QAAQ,CAAC,IAAI;WAC3F,UAAU,CAAC,KAAK,CAAC,SAAS,CAAC,KAAK,UAAU,CAAC,kBAAkB,CAAC,IAAI,CAAC,CAAC;WACpE,CAAC,MAAM,CAAC,aAAa,CAAC,KAAK,CAAC,eAAe,CAAC,IAAK,KAAK,CAAC,eAA0B,GAAG,CAAC;WACpF,KAAK,CAAC,eAA0B,GAAG,KAAK,CAAC,QAAQ,CAAC,UAAU;WAC7D,OAAO,KAAK,CAAC,UAAU,KAAK,QAAQ,IAAI,CAAC,MAAM,CAAC,QAAQ,CAAC,IAAI,CAAC,KAAK,CAAC,KAAK,CAAC,UAAU,CAAC,CAAC;WACtF,CAAC,KAAK,CAAC,OAAO,CAAC,KAAK,CAAC,MAAM,CAAC;QAAE,MAAM,SAAS,CAAC,4CAA4C,CAAC,CAAC;IACjG,MAAM,MAAM,GAAG,KAAK,CAAC,MAAM,CAAC,GAAG,CAAC,KAAK,CAAC,EAAE;QACtC,IAAI,CAAC,KAAK,IAAI,OAAO,KAAK,KAAK,QAAQ,IAAI,KAAK,CAAC,OAAO,CAAC,KAAK,CAAC;YAAE,MAAM,SAAS,CAAC,wCAAwC,CAAC,CAAC;QAC3H,MAAM,IAAI,GAAG,KAAgC,CAAC;QAC9C,IAAI,MAAM,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC,MAAM,KAAK,CAAC,IAAI,OAAO,IAAI,CAAC,OAAO,KAAK,QAAQ,IAAI,CAAC,KAAK,CAAC,KAAK,CAAC,QAAQ,CAAC,IAAI,CAAC,OAAO,CAAC;eACxG,CAAC,MAAM,CAAC,aAAa,CAAC,IAAI,CAAC,OAAO,CAAC,IAAK,IAAI,CAAC,OAAkB,GAAG,CAAC,IAAK,IAAI,CAAC,OAAkB,GAAG,KAAK,CAAC,QAAQ;YAAE,MAAM,SAAS,CAAC,0DAA0D,CAAC,CAAC;QACnM,OAAO,EAAE,OAAO,EAAE,IAAI,CAAC,OAAO,EAAE,OAAO,EAAE,IAAI,CAAC,OAAiB,EAAE,CAAC;IACpE,CAAC,CAAC,CAAC;IACH,IAAI,IAAI,GAAG,CAAC,MAAM,CAAC,GAAG,CAAC,OAAO,CAAC,CAAC,CAAC,IAAI,KAAK,MAAM,CAAC,MAAM,IAAI,UAAU,CAAC,MAAM,CAAC,KAAK,UAAU,CAAC,SAAS,CAAC,MAAM,CAAC,CAAC;WAC1G,UAAU,CAAC,KAAK,CAAC,KAAK,CAAC,KAAK,UAAU,CAAC,WAAW,CAAC,MAAM,CAAC,CAAC;WAC3D,KAAK,CAAC,QAAQ,CAAC,IAAI,KAAK,OAAO,IAAI,UAAU,CAAC,CAAC,GAAG,IAAI,GAAG,CAAC,KAAK,CAAC,QAAQ,CAAC,SAAS,CAAC,CAAC,CAAC,IAAI,EAAE,CAAC,KAAK,UAAU,CAAC,WAAW,CAAC,MAAM,CAAC,CAAC,EAAE,CAAC;QACtI,MAAM,SAAS,CAAC,+DAA+D,CAAC,CAAC;IACnF,CAAC;IACD,OAAO,EAAE,MAAM,EAAE,SAAS,EAAE,KAAK,CAAC,UAAU,EAAE,CAAC;AACjD,CAAC;AAED,0FAA0F;AAC1F,MAAM,CAAC,KAAK,UAAU,yBAAyB,CAAC,KAG/C;IACC,MAAM,MAAM,GAAG,MAAM,mBAAmB,CAAC,KAAK,CAAC,aAAa,CAAC,CAAC;IAC9D,IAAI,cAAc,GAAG,KAAK,CAAC;IAC3B,KAAK,MAAM,IAAI,IAAI,KAAK,CAAC,KAAK,EAAE,CAAC;QAC/B,MAAM,OAAO,GAAG,KAAK,CAAC,IAAI,CAAC,KAAK,CAAC,IAAI,CAAC,IAAI,CAAC,EAAE,CAAC,IAAI,CAAC,OAAO,KAAK,IAAI,CAAC,OAAO,IAAI,IAAI,CAAC,OAAO,KAAK,IAAI,CAAC,OAAO,CAAC,CAAC;QAC9G,MAAM,MAAM,GAAG,OAAO,IAAI,KAAK,CAAC,IAAI,CAAC,UAAU,CAAC,IAAI,CAAC,IAAI,CAAC,EAAE,CAAC,IAAI,CAAC,KAAK,CAAC,QAAQ,CAAC,OAAO,CAAC,OAAO,CAAC,CAAC,CAAC;QACnG,IAAI,CAAC,MAAM,IAAI,MAAM,CAAC,KAAK,CAAC,MAAM,KAAK,CAAC,IAAI,MAAM,CAAC,QAAQ,CAAC,MAAM,KAAK,CAAC;YAAE,MAAM,SAAS,CAAC,2DAA2D,CAAC,CAAC;QACvJ,MAAM,IAAI,GAAG,wBAAwB,CAAC,MAAM,EAAE,KAAK,CAAC,OAAO,CAAC,CAAC;QAC7D,MAAM,QAAQ,GAAG,EAAE,cAAc,EAAE,GAAG,EAAE,QAAQ,EAAE,KAAK,CAAC,OAAO,EAAE,cAAc,EAAE,MAAM,CAAC,OAAO;YAC7F,OAAO,EAAE,IAAI,CAAC,OAAO,EAAE,WAAW,EAAE,UAAU,CAAC,KAAK,CAAC,IAAI,CAAC,EAAE,cAAc,EAAE,UAAU,CAAC,KAAK,CAAC,OAAO,CAAC,EAAE,CAAC;QAC1G,MAAM,OAAO,GAAG,MAAM,QAAQ,CAC5B,IAAI,CAAC,IAAI,CAAC,KAAK,CAAC,SAAS,EAAE,aAAa,EAAE,GAAG,IAAI,CAAC,OAAO,OAAO,CAAC,EAAE,IAAI,CAAC,CAAC;QAC3E,IAAI,CAAC,OAAO,EAAE,CAAC;YAAC,cAAc,GAAG,IAAI,CAAC;YAAC,SAAS;QAAC,CAAC;QAClD,IAAI,OAAO,CAAC,KAAK,KAAK,aAAa,IAAI,UAAU,CAAC,OAAO,CAAC,QAAQ,CAAC,KAAK,UAAU,CAAC,QAAQ,CAAC,IAAI,CAAC,OAAO,CAAC,MAAM;eAC1G,MAAM,CAAC,IAAI,CAAC,KAAK,CAAC,EAAE,CAAC,KAAK,CAAC,QAAQ,KAAK,OAAO,CAAC,QAAQ,IAAI,KAAK,CAAC,OAAO,KAAK,IAAI,CAAC,OAAO;mBACxF,KAAK,CAAC,QAAQ,KAAK,KAAK,CAAC,IAAI,CAAC,QAAQ,IAAI,KAAK,CAAC,KAAK,KAAK,UAAU,CAAC,EAAE,CAAC;YAAC,cAAc,GAAG,IAAI,CAAC;YAAC,SAAS;QAAC,CAAC;QAClH,IAAI,UAAU,CAAC,OAAO,CAAC,QAAQ,CAAC,KAAK,UAAU,CAAC,QAAQ,CAAC,IAAI,OAAO,CAAC,KAAK,KAAK,WAAW,IAAI,CAAC,OAAO,CAAC,MAAM;eACxG,CAAC,MAAM,CAAC,IAAI,CAAC,KAAK,CAAC,EAAE,CAAC,KAAK,CAAC,QAAQ,KAAK,OAAO,CAAC,MAAO,CAAC,OAAO,IAAI,KAAK,CAAC,OAAO,KAAK,IAAI,CAAC,OAAO;mBAChG,KAAK,CAAC,QAAQ,KAAK,KAAK,CAAC,IAAI,CAAC,QAAQ,IAAI,KAAK,CAAC,KAAK,KAAK,UAAU,CAAC;YAAE,MAAM,SAAS,CAAC,iFAAiF,CAAC,CAAC;QACjL,IAAI,OAAO,CAAC,MAAM,CAAC,GAAG,CAAC,OAAO,CAAC,iBAAiB,KAAK,CAAC,IAAI,OAAO,CAAC,MAAM,CAAC,GAAG,CAAC,SAAS,KAAK,IAAI,EAAE,CAAC;YAChG,MAAM,IAAI,UAAU,CAAC,yEAAyE,EAAE,EAAE,IAAI,EAAE,0BAA0B,EAAE,QAAQ,EAAE,EAAE,EAAE,CAAC,CAAC;QACtJ,CAAC;IACH,CAAC;IACD,OAAO,cAAc,CAAC;AACxB,CAAC;AAED,SAAS,SAAS,CAAC,OAAe,IAAgB,OAAO,IAAI,UAAU,CAAC,OAAO,EAAE,EAAE,IAAI,EAAE,2BAA2B,EAAE,QAAQ,EAAE,EAAE,EAAE,CAAC,CAAC,CAAC,CAAC"}
|
|
@@ -0,0 +1,25 @@
|
|
|
1
|
+
import { SCHEMA_VERSION, atomicWriteJSON } from "../foundation/index.js";
|
|
2
|
+
import { evalRerunSemantics } from "./rerun-types.js";
|
|
3
|
+
import { sortSlots } from "./rerun-slots.js";
|
|
4
|
+
export async function writeRerunState(file, input) {
|
|
5
|
+
await atomicWriteJSON(file, {
|
|
6
|
+
schema_version: SCHEMA_VERSION,
|
|
7
|
+
rerun_id: input.rerunId,
|
|
8
|
+
eval_id: input.evalId,
|
|
9
|
+
rerun_type: input.rerunType,
|
|
10
|
+
semantics: evalRerunSemantics(input.rerunType),
|
|
11
|
+
status: input.status,
|
|
12
|
+
tasks: [...input.tasks],
|
|
13
|
+
...(input.trials ? { trials: sortSlots(input.trials) } : {}),
|
|
14
|
+
repaired_tasks: [...input.repairedTasks],
|
|
15
|
+
...(input.repairedTrials ? { repaired_trials: sortSlots(input.repairedTrials) } : {}),
|
|
16
|
+
...(input.evalStatus ? { eval_status: input.evalStatus } : {}),
|
|
17
|
+
...(input.remainingInvalidTasks ? { remaining_invalid_tasks: [...input.remainingInvalidTasks] } : {}),
|
|
18
|
+
...(input.remainingInvalidTrials ? { remaining_invalid_trials: sortSlots(input.remainingInvalidTrials) } : {}),
|
|
19
|
+
...(input.errorCode ? { error: { code: input.errorCode } } : {}),
|
|
20
|
+
started_at: input.startedAt,
|
|
21
|
+
...(input.completedAt ? { completed_at: input.completedAt } : {}),
|
|
22
|
+
updated_at: input.completedAt ?? new Date().toISOString(),
|
|
23
|
+
});
|
|
24
|
+
}
|
|
25
|
+
//# sourceMappingURL=rerun-state.js.map
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"rerun-state.js","sourceRoot":"","sources":["../../../src/evals/rerun-state.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,cAAc,EAAE,eAAe,EAAE,MAAM,wBAAwB,CAAC;AACzE,OAAO,EAAE,kBAAkB,EAAE,MAAM,kBAAkB,CAAC;AAEtD,OAAO,EAAE,SAAS,EAAE,MAAM,kBAAkB,CAAC;AAG7C,MAAM,CAAC,KAAK,UAAU,eAAe,CAAC,IAAY,EAAE,KAenD;IACC,MAAM,eAAe,CAAC,IAAI,EAAE;QAC1B,cAAc,EAAE,cAAc;QAC9B,QAAQ,EAAE,KAAK,CAAC,OAAO;QACvB,OAAO,EAAE,KAAK,CAAC,MAAM;QACrB,UAAU,EAAE,KAAK,CAAC,SAAS;QAC3B,SAAS,EAAE,kBAAkB,CAAC,KAAK,CAAC,SAAS,CAAC;QAC9C,MAAM,EAAE,KAAK,CAAC,MAAM;QACpB,KAAK,EAAE,CAAC,GAAG,KAAK,CAAC,KAAK,CAAC;QACvB,GAAG,CAAC,KAAK,CAAC,MAAM,CAAC,CAAC,CAAC,EAAE,MAAM,EAAE,SAAS,CAAC,KAAK,CAAC,MAAM,CAAC,EAAE,CAAC,CAAC,CAAC,EAAE,CAAC;QAC5D,cAAc,EAAE,CAAC,GAAG,KAAK,CAAC,aAAa,CAAC;QACxC,GAAG,CAAC,KAAK,CAAC,cAAc,CAAC,CAAC,CAAC,EAAE,eAAe,EAAE,SAAS,CAAC,KAAK,CAAC,cAAc,CAAC,EAAE,CAAC,CAAC,CAAC,EAAE,CAAC;QACrF,GAAG,CAAC,KAAK,CAAC,UAAU,CAAC,CAAC,CAAC,EAAE,WAAW,EAAE,KAAK,CAAC,UAAU,EAAE,CAAC,CAAC,CAAC,EAAE,CAAC;QAC9D,GAAG,CAAC,KAAK,CAAC,qBAAqB,CAAC,CAAC,CAAC,EAAE,uBAAuB,EAAE,CAAC,GAAG,KAAK,CAAC,qBAAqB,CAAC,EAAE,CAAC,CAAC,CAAC,EAAE,CAAC;QACrG,GAAG,CAAC,KAAK,CAAC,sBAAsB,CAAC,CAAC,CAAC,EAAE,wBAAwB,EAAE,SAAS,CAAC,KAAK,CAAC,sBAAsB,CAAC,EAAE,CAAC,CAAC,CAAC,EAAE,CAAC;QAC9G,GAAG,CAAC,KAAK,CAAC,SAAS,CAAC,CAAC,CAAC,EAAE,KAAK,EAAE,EAAE,IAAI,EAAE,KAAK,CAAC,SAAS,EAAE,EAAE,CAAC,CAAC,CAAC,EAAE,CAAC;QAChE,UAAU,EAAE,KAAK,CAAC,SAAS;QAC3B,GAAG,CAAC,KAAK,CAAC,WAAW,CAAC,CAAC,CAAC,EAAE,YAAY,EAAE,KAAK,CAAC,WAAW,EAAE,CAAC,CAAC,CAAC,EAAE,CAAC;QACjE,UAAU,EAAE,KAAK,CAAC,WAAW,IAAI,IAAI,IAAI,EAAE,CAAC,WAAW,EAAE;KAC1D,CAAC,CAAC;AACL,CAAC"}
|
|
@@ -1 +1 @@
|
|
|
1
|
-
{"version":3,"file":"rerun-types.js","sourceRoot":"","sources":["../../../src/evals/rerun-types.ts"],"names":[],"mappings":"AAGA,OAAO,EAAE,UAAU,EAAE,YAAY,EAAE,MAAM,wBAAwB,CAAC;AAElE,MAAM,CAAC,MAAM,gBAAgB,GAAG;IAC9B,mBAAmB;IACnB,kBAAkB;IAClB,mBAAmB;IACnB,eAAe;IACf,cAAc;CACN,CAAC;
|
|
1
|
+
{"version":3,"file":"rerun-types.js","sourceRoot":"","sources":["../../../src/evals/rerun-types.ts"],"names":[],"mappings":"AAGA,OAAO,EAAE,UAAU,EAAE,YAAY,EAAE,MAAM,wBAAwB,CAAC;AAElE,MAAM,CAAC,MAAM,gBAAgB,GAAG;IAC9B,mBAAmB;IACnB,kBAAkB;IAClB,mBAAmB;IACnB,eAAe;IACf,cAAc;CACN,CAAC;AA8DX,MAAM,UAAU,kBAAkB,CAAC,KAAc;IAC/C,IAAI,OAAO,KAAK,KAAK,QAAQ,IAAI,CAAC,gBAAgB,CAAC,QAAQ,CAAC,KAAsB,CAAC,EAAE,CAAC;QACpF,MAAM,YAAY,CAAC,qCAAqC,gBAAgB,CAAC,IAAI,CAAC,IAAI,CAAC,EAAE,CAAC,CAAC;IACzF,CAAC;IACD,OAAO,KAAsB,CAAC;AAChC,CAAC;AAED,MAAM,UAAU,kBAAkB,CAAC,IAAmB;IACpD,QAAQ,IAAI,EAAE,CAAC;QACb,KAAK,mBAAmB;YACtB,OAAO,EAAE,gBAAgB,EAAE,SAAS,EAAE,mBAAmB,EAAE,sBAAsB,EAAE,cAAc,EAAE,OAAO,EAAE,kBAAkB,EAAE,IAAI,EAAE,CAAC;QACzI,KAAK,kBAAkB;YACrB,OAAO,EAAE,gBAAgB,EAAE,QAAQ,EAAE,mBAAmB,EAAE,gBAAgB,EAAE,cAAc,EAAE,YAAY,EAAE,kBAAkB,EAAE,IAAI,EAAE,CAAC;QACvI,KAAK,mBAAmB;YACtB,OAAO,EAAE,gBAAgB,EAAE,QAAQ,EAAE,mBAAmB,EAAE,sBAAsB,EAAE,cAAc,EAAE,YAAY,EAAE,kBAAkB,EAAE,IAAI,EAAE,CAAC;QAC7I,KAAK,eAAe;YAClB,OAAO,EAAE,gBAAgB,EAAE,MAAM,EAAE,mBAAmB,EAAE,MAAM,EAAE,cAAc,EAAE,oBAAoB,EAAE,kBAAkB,EAAE,KAAK,EAAE,CAAC;QACpI,KAAK,cAAc;YACjB,OAAO,EAAE,gBAAgB,EAAE,MAAM,EAAE,mBAAmB,EAAE,MAAM,EAAE,cAAc,EAAE,MAAM,EAAE,kBAAkB,EAAE,KAAK,EAAE,CAAC;IACxH,CAAC;AACH,CAAC;AAED,MAAM,UAAU,4BAA4B,CAAC,IAAmB;IAC9D,IAAI,IAAI,KAAK,mBAAmB,IAAI,IAAI,KAAK,cAAc,IAAI,IAAI,KAAK,eAAe;QAAE,OAAO;IAChG,MAAM,WAAW,GAA8H;QAC7I,kBAAkB,EAAE;YAClB,IAAI,EAAE,mCAAmC;YACzC,OAAO,EAAE,qJAAqJ;SAC/J;QACD,mBAAmB,EAAE;YACnB,IAAI,EAAE,oCAAoC;YAC1C,OAAO,EAAE,wLAAwL;SAClM;KACF,CAAC;IACF,MAAM,OAAO,GAAG,WAAW,CAAC,IAAI,CAAC,CAAC;IAClC,MAAM,IAAI,UAAU,CAAC,OAAO,CAAC,OAAO,EAAE,EAAE,IAAI,EAAE,OAAO,CAAC,IAAI,EAAE,QAAQ,EAAE,CAAC,EAAE,CAAC,CAAC;AAC7E,CAAC"}
|
package/dist/src/evals/rerun.js
CHANGED
|
@@ -14,7 +14,12 @@ import { loadPersistedRerunRequest, loadRerunLocalTransport, loadRerunPreparedAr
|
|
|
14
14
|
import { parseEvalExecutionPlan } from "./execution-plan.js";
|
|
15
15
|
import { startEvalModelCaptureRuntime } from "./model-capture-runtime.js";
|
|
16
16
|
import { verifierOnlyEvalRerun } from "./verifier-only-rerun.js";
|
|
17
|
+
import { remoteVerifierOnlyEvalRerun } from "./remote-verifier-rerun.js";
|
|
17
18
|
import { restartIncompleteEval } from "./preparation-rerun.js";
|
|
19
|
+
import { writeRerunState } from "./rerun-state.js";
|
|
20
|
+
import { assertRemoteRerunQuiescent, executeRemoteRerunGroup } from "./remote-rerun-execution.js";
|
|
21
|
+
import { remoteRerunNeedsExecution, restoreRemoteRerunSelection } from "./rerun-resume.js";
|
|
22
|
+
import { stageRemoteRerunCompletion } from "./rerun-completion.js";
|
|
18
23
|
export { selectRerunTasks, selectRerunTrialSlots } from "./rerun-slots.js";
|
|
19
24
|
export async function rerunEval(options) {
|
|
20
25
|
if (!options.root)
|
|
@@ -32,7 +37,7 @@ export async function rerunEval(options) {
|
|
|
32
37
|
return withFileLock(path.join(evalDirectory, "reruns"), "active", () => rerunEvalLocked({ ...options, rerunId, rerunType, evalId, evalDirectory }), { timeoutCode: "eval_rerun_active", timeoutExitCode: 2, ...(options.signal ? { signal: options.signal } : {}) });
|
|
33
38
|
}
|
|
34
39
|
async function rerunEvalLocked(options) {
|
|
35
|
-
|
|
40
|
+
let startedAt = new Date().toISOString();
|
|
36
41
|
const rerunId = options.rerunId;
|
|
37
42
|
const rerunDirectory = path.join(options.evalDirectory, "reruns", rerunId);
|
|
38
43
|
const statePath = path.join(rerunDirectory, "state.json");
|
|
@@ -44,6 +49,19 @@ async function rerunEvalLocked(options) {
|
|
|
44
49
|
if (requestValue === null)
|
|
45
50
|
throw new HitchError(`eval not found: ${options.evalId}`, { code: "eval_not_found", exitCode: 3 });
|
|
46
51
|
const request = await loadPersistedRerunRequest(requestValue, options.evalDirectory);
|
|
52
|
+
if (request.training_binding)
|
|
53
|
+
throw new HitchError("training slots are repaired by the frozen Gear batch coordinator, not eval rerun", { code: "training_rerun_fenced", exitCode: 12 });
|
|
54
|
+
const submission = await readJSON(path.join(options.evalDirectory, "submission.json"), null);
|
|
55
|
+
if (options.resumeRemoteRerun && (!["candidate-restart", "verifier-only"].includes(options.rerunType) || !submission?.execution?.provider
|
|
56
|
+
|| submission.execution.provider === "local-docker"))
|
|
57
|
+
throw unavailable("only a reconciled remote rerun can restore its selection");
|
|
58
|
+
if (submission?.execution?.provider && submission.execution.provider !== "local-docker" && options.rerunType !== "collect-only") {
|
|
59
|
+
await assertRemoteRerunQuiescent(options.evalDirectory, submission.execution.provider);
|
|
60
|
+
}
|
|
61
|
+
if (submission?.execution?.provider && submission.execution.provider !== "local-docker" && options.rerunType !== "collect-only"
|
|
62
|
+
&& (!["candidate-restart", "verifier-only"].includes(options.rerunType) || !options.remoteWorkExecutor || options.executionWorker?.provider !== submission.execution.provider)) {
|
|
63
|
+
throw new HitchError("remote candidate/verifier rerun requires worker dispatch; local fallback is not supported", { code: "remote_rerun_unavailable", exitCode: 12 });
|
|
64
|
+
}
|
|
47
65
|
if (options.maxConcurrentOverride !== undefined && (!Number.isSafeInteger(options.maxConcurrentOverride)
|
|
48
66
|
|| options.maxConcurrentOverride < 1 || options.maxConcurrentOverride > request.max_concurrent)) {
|
|
49
67
|
throw invalidInput("eval rerun concurrency override is invalid");
|
|
@@ -60,24 +78,30 @@ async function rerunEvalLocked(options) {
|
|
|
60
78
|
const previousResult = await readJSON(path.join(options.evalDirectory, "result.json"), null);
|
|
61
79
|
if (previousResult?.status === "cancelled")
|
|
62
80
|
throw new HitchError("cancelled eval cannot be rerun", { code: "eval_rerun_cancelled", exitCode: 2 });
|
|
63
|
-
const
|
|
81
|
+
const restored = options.resumeRemoteRerun ? await restoreRemoteRerunSelection({ directory: rerunDirectory,
|
|
82
|
+
evalId: options.evalId, rerunId, selector: options.selector, tasks: plan.tasks, attempts: plan.attempts, progress,
|
|
83
|
+
rerunType: options.rerunType, ...(options.verifierRuntimeId ? { verifierRuntimeId: options.verifierRuntimeId } : {}) }) : undefined;
|
|
84
|
+
if (restored)
|
|
85
|
+
startedAt = restored.startedAt;
|
|
86
|
+
const selectedTrials = restored?.trials ?? selectRerunTrialSlots(plan.tasks, plan.attempts, progress, options.selector, {
|
|
64
87
|
// Each executable recovery mode validates its own source prerequisites.
|
|
65
88
|
allowVerifierFailures: options.rerunType === "candidate-restart" || options.rerunType === "collect-only" || options.rerunType === "verifier-only",
|
|
66
89
|
});
|
|
67
90
|
const selectedTasks = uniqueTasks(selectedTrials);
|
|
68
|
-
|
|
69
|
-
|
|
70
|
-
|
|
71
|
-
|
|
72
|
-
|
|
73
|
-
|
|
74
|
-
|
|
75
|
-
|
|
76
|
-
|
|
77
|
-
|
|
78
|
-
|
|
79
|
-
|
|
80
|
-
|
|
91
|
+
if (!restored)
|
|
92
|
+
await atomicWriteJSON(path.join(rerunDirectory, "request.json"), {
|
|
93
|
+
schema_version: SCHEMA_VERSION,
|
|
94
|
+
rerun_id: rerunId,
|
|
95
|
+
eval_id: options.evalId,
|
|
96
|
+
rerun_type: options.rerunType,
|
|
97
|
+
...(options.verifierRuntimeId ? { verifier_runtime_id: options.verifierRuntimeId } : {}),
|
|
98
|
+
semantics: evalRerunSemantics(options.rerunType),
|
|
99
|
+
mode: options.selector.mode,
|
|
100
|
+
tasks: selectedTasks,
|
|
101
|
+
trials: selectedTrials,
|
|
102
|
+
base_generation: progress.generation,
|
|
103
|
+
created_at: startedAt,
|
|
104
|
+
});
|
|
81
105
|
await writeRerunState(statePath, {
|
|
82
106
|
rerunId,
|
|
83
107
|
evalId: options.evalId,
|
|
@@ -92,21 +116,49 @@ async function rerunEvalLocked(options) {
|
|
|
92
116
|
const repaired = new Map();
|
|
93
117
|
const backendRuns = [];
|
|
94
118
|
let captureRuntime;
|
|
119
|
+
let inferenceLease;
|
|
95
120
|
try {
|
|
96
|
-
if (options.rerunType === "verifier-only")
|
|
97
|
-
|
|
121
|
+
if (options.rerunType === "verifier-only") {
|
|
122
|
+
const execute = submission?.execution?.provider && submission.execution.provider !== "local-docker" ? remoteVerifierOnlyEvalRerun : verifierOnlyEvalRerun;
|
|
123
|
+
return await execute({ ...options, rerunDirectory, startedAt, request, plan, progress, previousResult, selectedTrials });
|
|
124
|
+
}
|
|
98
125
|
if (options.rerunType === "collect-only")
|
|
99
126
|
return collectOnlyEvalRerun({ root: options.root, evalId: options.evalId, evalDirectory: options.evalDirectory, rerunId, rerunDirectory, startedAt, request, plan, progress, previousResult, selectedTrials, env: options.env ?? process.env, ...(options.signal ? { signal: options.signal } : {}) });
|
|
100
127
|
if (selectedTrials.length > 0) {
|
|
101
128
|
const executionPlan = parseEvalExecutionPlan(await readJSON(path.join(options.evalDirectory, "execution-plan.json")));
|
|
102
129
|
if (executionPlan.eval_id !== options.evalId)
|
|
103
130
|
throw unavailable("eval execution plan identity changed");
|
|
104
|
-
|
|
105
|
-
|
|
106
|
-
|
|
107
|
-
|
|
108
|
-
|
|
109
|
-
|
|
131
|
+
const remote = executionPlan.provider !== "local-docker";
|
|
132
|
+
if (remote && (!options.remoteWorkExecutor || options.executionWorker?.provider !== executionPlan.provider))
|
|
133
|
+
throw unavailable("remote rerun has no executor for its frozen provider");
|
|
134
|
+
const needsExecution = !remote || await remoteRerunNeedsExecution({ evalDirectory: options.evalDirectory, directory: rerunDirectory,
|
|
135
|
+
rerunId, plan: executionPlan, request, slots: selectedTrials });
|
|
136
|
+
if (request.local_inference && needsExecution) {
|
|
137
|
+
if (!options.inferenceCoordinator)
|
|
138
|
+
throw new HitchError("local eval rerun inference requires the Hitch daemon", { code: "inference_route_unavailable", exitCode: 12 });
|
|
139
|
+
const inferenceId = plan.candidate.inference_id;
|
|
140
|
+
if (typeof inferenceId !== "string" || !/^sha256:[a-f0-9]{64}$/.test(inferenceId)) {
|
|
141
|
+
throw new HitchError("source eval has no valid local inference identity", { code: "inference_lock_mismatch", exitCode: 12 });
|
|
142
|
+
}
|
|
143
|
+
inferenceLease = await options.inferenceCoordinator.acquire({
|
|
144
|
+
run_id: `run_${rerunId.slice("rerun_".length)}`,
|
|
145
|
+
harness_ref: request.harness_ref,
|
|
146
|
+
selection: { ...request.local_inference, inference_id: inferenceId },
|
|
147
|
+
cache_scope_owner: `${options.evalId}:${rerunId}`,
|
|
148
|
+
evidence_owner: { kind: "eval", eval_id: options.evalId, rerun_id: rerunId },
|
|
149
|
+
...(options.signal ? { signal: options.signal } : {}),
|
|
150
|
+
});
|
|
151
|
+
}
|
|
152
|
+
if (needsExecution)
|
|
153
|
+
captureRuntime = await startEvalModelCaptureRuntime({
|
|
154
|
+
plan: executionPlan.model_capture ?? defaultModelCapturePlan(),
|
|
155
|
+
evalId: options.evalId,
|
|
156
|
+
evalDirectory: rerunDirectory,
|
|
157
|
+
env: options.env ?? process.env,
|
|
158
|
+
...(inferenceLease ? { managedInference: {
|
|
159
|
+
binding: inferenceLease.binding, credential: inferenceLease.credential, modelId: inferenceLease.lock.model_id,
|
|
160
|
+
} } : {}),
|
|
161
|
+
});
|
|
110
162
|
const activeCaptureRuntime = captureRuntime;
|
|
111
163
|
const resolvedRevision = await loadRerunResolvedRevision(options.evalDirectory, plan);
|
|
112
164
|
const preparedArtifacts = await loadRerunPreparedArtifacts(options.root, plan);
|
|
@@ -116,7 +168,7 @@ async function rerunEvalLocked(options) {
|
|
|
116
168
|
|| runtime.manifest_digest !== requiredString(plan.controllerRuntime.manifest_digest, "plan controller runtime digest")) {
|
|
117
169
|
throw unavailable("controller runtime identity changed");
|
|
118
170
|
}
|
|
119
|
-
const localTransport = await loadRerunLocalTransport(options.evalDirectory, plan, resolvedRevision, options.env, options.signal);
|
|
171
|
+
const localTransport = remote ? undefined : await loadRerunLocalTransport(options.evalDirectory, plan, resolvedRevision, options.env, options.signal);
|
|
120
172
|
const groups = groupRerunSlotsByArtifact(selectedTrials, executionPlan, [...preparedArtifacts.keys()]);
|
|
121
173
|
for (const { logicalAttempt, slots, artifactId, splitAttempt } of groups) {
|
|
122
174
|
if (options.signal?.aborted)
|
|
@@ -148,9 +200,9 @@ async function rerunEvalLocked(options) {
|
|
|
148
200
|
});
|
|
149
201
|
const previousGeneration = progress.generation;
|
|
150
202
|
progress = replaceInvalidEvalProgressTrial(progress, ref);
|
|
203
|
+
repaired.set(key, slot);
|
|
151
204
|
if (progress.generation === previousGeneration)
|
|
152
205
|
return;
|
|
153
|
-
repaired.set(key, slot);
|
|
154
206
|
await writeEvalProgress(options.evalDirectory, progress);
|
|
155
207
|
await writeRerunState(statePath, {
|
|
156
208
|
rerunId,
|
|
@@ -164,6 +216,16 @@ async function rerunEvalLocked(options) {
|
|
|
164
216
|
startedAt,
|
|
165
217
|
});
|
|
166
218
|
};
|
|
219
|
+
if (remote) {
|
|
220
|
+
backendRuns.push(...await executeRemoteRerunGroup({
|
|
221
|
+
root: options.root, evalId: validateEvalId(options.evalId), evalDirectory: options.evalDirectory, rerunId, rerunDirectory,
|
|
222
|
+
request, plan: executionPlan, slots, resolvedRevision,
|
|
223
|
+
artifact: preparedArtifacts.get(artifactId) ?? (() => { throw unavailable(`rerun artifact is unavailable: ${artifactId}`); })(),
|
|
224
|
+
runtimeDirectory: runtime.directory, runtimeId: runtime.runtime_id, executor: options.remoteWorkExecutor,
|
|
225
|
+
...(inferenceLease ? { inferenceLease } : {}), ...(options.signal ? { signal: options.signal } : {}), publish,
|
|
226
|
+
}));
|
|
227
|
+
continue;
|
|
228
|
+
}
|
|
167
229
|
const backendRun = await runHarborBackend({
|
|
168
230
|
evalId: options.evalId,
|
|
169
231
|
evalDirectory: options.evalDirectory,
|
|
@@ -262,6 +324,8 @@ async function rerunEvalLocked(options) {
|
|
|
262
324
|
started_at: startedAt,
|
|
263
325
|
completed_at: completedAt,
|
|
264
326
|
};
|
|
327
|
+
if (submission?.execution?.provider && submission.execution.provider !== "local-docker")
|
|
328
|
+
await stageRemoteRerunCompletion(options.evalDirectory, rerunDirectory, output);
|
|
265
329
|
await writeRerunState(statePath, {
|
|
266
330
|
rerunId,
|
|
267
331
|
evalId: options.evalId,
|
|
@@ -306,6 +370,7 @@ async function rerunEvalLocked(options) {
|
|
|
306
370
|
}
|
|
307
371
|
finally {
|
|
308
372
|
await captureRuntime?.close().catch(() => undefined);
|
|
373
|
+
await inferenceLease?.release().catch(() => undefined);
|
|
309
374
|
}
|
|
310
375
|
}
|
|
311
376
|
function defaultModelCapturePlan() {
|
|
@@ -379,27 +444,6 @@ async function finalizeRerun(evalDirectory, request, plan, progress, previousRes
|
|
|
379
444
|
await atomicWriteJSON(path.join(evalDirectory, "result.json"), result);
|
|
380
445
|
return result;
|
|
381
446
|
}
|
|
382
|
-
async function writeRerunState(file, input) {
|
|
383
|
-
await atomicWriteJSON(file, {
|
|
384
|
-
schema_version: SCHEMA_VERSION,
|
|
385
|
-
rerun_id: input.rerunId,
|
|
386
|
-
eval_id: input.evalId,
|
|
387
|
-
rerun_type: input.rerunType,
|
|
388
|
-
semantics: evalRerunSemantics(input.rerunType),
|
|
389
|
-
status: input.status,
|
|
390
|
-
tasks: [...input.tasks],
|
|
391
|
-
...(input.trials ? { trials: sortSlots(input.trials) } : {}),
|
|
392
|
-
repaired_tasks: [...input.repairedTasks],
|
|
393
|
-
...(input.repairedTrials ? { repaired_trials: sortSlots(input.repairedTrials) } : {}),
|
|
394
|
-
...(input.evalStatus ? { eval_status: input.evalStatus } : {}),
|
|
395
|
-
...(input.remainingInvalidTasks ? { remaining_invalid_tasks: [...input.remainingInvalidTasks] } : {}),
|
|
396
|
-
...(input.remainingInvalidTrials ? { remaining_invalid_trials: sortSlots(input.remainingInvalidTrials) } : {}),
|
|
397
|
-
...(input.errorCode ? { error: { code: input.errorCode } } : {}),
|
|
398
|
-
started_at: input.startedAt,
|
|
399
|
-
...(input.completedAt ? { completed_at: input.completedAt } : {}),
|
|
400
|
-
updated_at: input.completedAt ?? new Date().toISOString(),
|
|
401
|
-
});
|
|
402
|
-
}
|
|
403
447
|
function requiredString(value, label) {
|
|
404
448
|
if (typeof value !== "string" || value.length === 0)
|
|
405
449
|
throw unavailable(`${label} is missing`);
|