@atomic-chat/core 0.5.1 → 0.13.2
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 +22 -0
- package/dist/app-daemon.js +20 -67
- package/dist/app-daemon.js.map +1 -1
- package/dist/backend/advisor/advisor.d.ts +132 -0
- package/dist/backend/advisor/advisor.js +537 -0
- package/dist/backend/advisor/advisor.js.map +1 -0
- package/dist/backend/advisor/index.d.ts +6 -0
- package/dist/backend/advisor/index.js +7 -0
- package/dist/backend/advisor/index.js.map +1 -0
- package/dist/backend/advisor/policy.d.ts +81 -0
- package/dist/backend/advisor/policy.js +154 -0
- package/dist/backend/advisor/policy.js.map +1 -0
- package/dist/backend/catalog/archive.d.ts +21 -7
- package/dist/backend/catalog/archive.js +57 -17
- package/dist/backend/catalog/archive.js.map +1 -1
- package/dist/backend/catalog/bundled-manifest-baseline.js +42 -26
- package/dist/backend/catalog/bundled-manifest-baseline.js.map +1 -1
- package/dist/backend/catalog/cuda-family.d.ts +12 -2
- package/dist/backend/catalog/cuda-family.js +30 -7
- package/dist/backend/catalog/cuda-family.js.map +1 -1
- package/dist/backend/catalog/index.d.ts +5 -0
- package/dist/backend/catalog/index.js +5 -0
- package/dist/backend/catalog/index.js.map +1 -1
- package/dist/backend/catalog/manifest.d.ts +5 -3
- package/dist/backend/catalog/manifest.js +12 -9
- package/dist/backend/catalog/manifest.js.map +1 -1
- package/dist/backend/catalog/migrate.d.ts +8 -6
- package/dist/backend/catalog/migrate.js +26 -15
- package/dist/backend/catalog/migrate.js.map +1 -1
- package/dist/backend/catalog/prism-catalog.d.ts +42 -0
- package/dist/backend/catalog/prism-catalog.js +101 -0
- package/dist/backend/catalog/prism-catalog.js.map +1 -0
- package/dist/backend/catalog/prism-manifest-baseline.d.ts +2 -0
- package/dist/backend/catalog/prism-manifest-baseline.js +11 -0
- package/dist/backend/catalog/prism-manifest-baseline.js.map +1 -0
- package/dist/backend/catalog/prism-manifest-baseline.json +134 -0
- package/dist/backend/catalog/prism-manifest.d.ts +90 -0
- package/dist/backend/catalog/prism-manifest.js +171 -0
- package/dist/backend/catalog/prism-manifest.js.map +1 -0
- package/dist/backend/catalog/turboquant-catalog.d.ts +60 -0
- package/dist/backend/catalog/turboquant-catalog.js +162 -0
- package/dist/backend/catalog/turboquant-catalog.js.map +1 -0
- package/dist/backend/catalog/turboquant-index.d.ts +104 -0
- package/dist/backend/catalog/turboquant-index.js +260 -0
- package/dist/backend/catalog/turboquant-index.js.map +1 -0
- package/dist/backend/index.d.ts +1 -0
- package/dist/backend/index.js +1 -0
- package/dist/backend/index.js.map +1 -1
- package/dist/backend/install/runtime-backend.d.ts +6 -6
- package/dist/backend/install/runtime-backend.js +23 -20
- package/dist/backend/install/runtime-backend.js.map +1 -1
- package/dist/backend/install/service.d.ts +106 -5
- package/dist/backend/install/service.js +247 -21
- package/dist/backend/install/service.js.map +1 -1
- package/dist/backend/installed/bundled.d.ts +19 -0
- package/dist/backend/installed/bundled.js +45 -0
- package/dist/backend/installed/bundled.js.map +1 -0
- package/dist/backend/installed/index.d.ts +1 -0
- package/dist/backend/installed/index.js +1 -0
- package/dist/backend/installed/index.js.map +1 -1
- package/dist/backend/installed/installed.js +3 -0
- package/dist/backend/installed/installed.js.map +1 -1
- package/dist/backend/optimal/optimal-cache.d.ts +27 -8
- package/dist/backend/optimal/optimal-cache.js +38 -19
- package/dist/backend/optimal/optimal-cache.js.map +1 -1
- package/dist/backend/optimal/optimal-store.js +1 -1
- package/dist/backend/optimal/optimal-store.js.map +1 -1
- package/dist/backend/select/categories.d.ts +2 -1
- package/dist/backend/select/categories.js +8 -1
- package/dist/backend/select/categories.js.map +1 -1
- package/dist/backend/select/features.d.ts +13 -5
- package/dist/backend/select/features.js +36 -6
- package/dist/backend/select/features.js.map +1 -1
- package/dist/backend/select/index.d.ts +2 -0
- package/dist/backend/select/index.js +2 -0
- package/dist/backend/select/index.js.map +1 -1
- package/dist/backend/select/prism.d.ts +54 -0
- package/dist/backend/select/prism.js +212 -0
- package/dist/backend/select/prism.js.map +1 -0
- package/dist/backend/select/recovery.d.ts +4 -2
- package/dist/backend/select/recovery.js +17 -5
- package/dist/backend/select/recovery.js.map +1 -1
- package/dist/backend/select/tiers.d.ts +20 -11
- package/dist/backend/select/tiers.js +49 -11
- package/dist/backend/select/tiers.js.map +1 -1
- package/dist/backend/select/turboquant-tiers.d.ts +45 -0
- package/dist/backend/select/turboquant-tiers.js +114 -0
- package/dist/backend/select/turboquant-tiers.js.map +1 -0
- package/dist/backend/types.d.ts +6 -69
- package/dist/cli/commands/backends.d.ts +10 -0
- package/dist/cli/commands/backends.js +142 -0
- package/dist/cli/commands/backends.js.map +1 -0
- package/dist/cli/commands/hardware.d.ts +13 -0
- package/dist/cli/commands/hardware.js +71 -0
- package/dist/cli/commands/hardware.js.map +1 -0
- package/dist/cli/commands/host-step.d.ts +28 -0
- package/dist/cli/commands/host-step.js +67 -0
- package/dist/cli/commands/host-step.js.map +1 -0
- package/dist/cli/commands/index.d.ts +4 -1
- package/dist/cli/commands/index.js +4 -1
- package/dist/cli/commands/index.js.map +1 -1
- package/dist/cli/commands/serve.js +4 -4
- package/dist/cli/commands/serve.js.map +1 -1
- package/dist/cli/commands/shared.d.ts +6 -1
- package/dist/cli/commands/shared.js +7 -0
- package/dist/cli/commands/shared.js.map +1 -1
- package/dist/cli/io.d.ts +11 -0
- package/dist/cli/io.js +28 -3
- package/dist/cli/io.js.map +1 -1
- package/dist/cli/launch.js +2 -2
- package/dist/cli/launch.js.map +1 -1
- package/dist/cli/main.d.ts +1 -1
- package/dist/cli/main.js +11 -1
- package/dist/cli/main.js.map +1 -1
- package/dist/client/control-client.d.ts +128 -7
- package/dist/client/control-client.js +211 -12
- package/dist/client/control-client.js.map +1 -1
- package/dist/cloud/registry.js +9 -1
- package/dist/cloud/registry.js.map +1 -1
- package/dist/config/paths.d.ts +137 -0
- package/dist/config/paths.js +163 -0
- package/dist/config/paths.js.map +1 -1
- package/dist/contracts/backend-advisor.d.ts +225 -0
- package/dist/contracts/backend-advisor.js +14 -0
- package/dist/contracts/backend-advisor.js.map +1 -0
- package/dist/contracts/control-api.d.ts +7 -1
- package/dist/contracts/control-api.js +7 -1
- package/dist/contracts/control-api.js.map +1 -1
- package/dist/contracts/decision.d.ts +394 -0
- package/dist/contracts/decision.js +69 -0
- package/dist/contracts/decision.js.map +1 -0
- package/dist/contracts/diffusion.d.ts +62 -9
- package/dist/contracts/embedding.d.ts +103 -0
- package/dist/contracts/embedding.js +24 -0
- package/dist/contracts/embedding.js.map +1 -0
- package/dist/contracts/engine-builds.d.ts +108 -0
- package/dist/contracts/engine-builds.js +11 -0
- package/dist/contracts/engine-builds.js.map +1 -0
- package/dist/contracts/engines.d.ts +157 -0
- package/dist/contracts/engines.js +50 -0
- package/dist/contracts/engines.js.map +1 -0
- package/dist/contracts/environment.d.ts +774 -0
- package/dist/contracts/environment.js +116 -0
- package/dist/contracts/environment.js.map +1 -0
- package/dist/contracts/errors.d.ts +43 -2
- package/dist/contracts/errors.js +41 -0
- package/dist/contracts/errors.js.map +1 -1
- package/dist/contracts/events.d.ts +82 -4
- package/dist/contracts/hardware.d.ts +79 -6
- package/dist/contracts/hardware.js +9 -5
- package/dist/contracts/hardware.js.map +1 -1
- package/dist/contracts/index.d.ts +8 -0
- package/dist/contracts/index.js +8 -0
- package/dist/contracts/index.js.map +1 -1
- package/dist/contracts/model-compatibility.d.ts +97 -0
- package/dist/contracts/model-compatibility.js +6 -0
- package/dist/contracts/model-compatibility.js.map +1 -0
- package/dist/contracts/model-setup.d.ts +115 -0
- package/dist/contracts/model-setup.js +33 -0
- package/dist/contracts/model-setup.js.map +1 -0
- package/dist/contracts/model-yml.d.ts +14 -0
- package/dist/contracts/model-yml.js +1 -0
- package/dist/contracts/model-yml.js.map +1 -1
- package/dist/contracts/session.d.ts +45 -2
- package/dist/contracts/session.js +14 -1
- package/dist/contracts/session.js.map +1 -1
- package/dist/contracts/settings.d.ts +44 -1
- package/dist/contracts/settings.js.map +1 -1
- package/dist/core/atomic-core.d.ts +44 -5
- package/dist/core/atomic-core.js +24 -0
- package/dist/core/atomic-core.js.map +1 -1
- package/dist/core/create.js +517 -23
- package/dist/core/create.js.map +1 -1
- package/dist/core/gpu/index.d.ts +6 -0
- package/dist/core/gpu/index.js +7 -0
- package/dist/core/gpu/index.js.map +1 -0
- package/dist/core/gpu/policy.d.ts +68 -0
- package/dist/core/gpu/policy.js +99 -0
- package/dist/core/gpu/policy.js.map +1 -0
- package/dist/core/gpu/residency.d.ts +44 -0
- package/dist/core/gpu/residency.js +101 -0
- package/dist/core/gpu/residency.js.map +1 -0
- package/dist/core/gpu-residency.d.ts +39 -0
- package/dist/core/gpu-residency.js +32 -0
- package/dist/core/gpu-residency.js.map +1 -0
- package/dist/core/index.d.ts +1 -1
- package/dist/core/managed-engines.d.ts +207 -0
- package/dist/core/managed-engines.js +579 -0
- package/dist/core/managed-engines.js.map +1 -0
- package/dist/core/managed-environment.d.ts +72 -0
- package/dist/core/managed-environment.js +218 -0
- package/dist/core/managed-environment.js.map +1 -0
- package/dist/core/model-setup/index.d.ts +4 -0
- package/dist/core/model-setup/index.js +5 -0
- package/dist/core/model-setup/index.js.map +1 -0
- package/dist/core/model-setup/plan.d.ts +56 -0
- package/dist/core/model-setup/plan.js +158 -0
- package/dist/core/model-setup/plan.js.map +1 -0
- package/dist/core/model-setup/service.d.ts +92 -0
- package/dist/core/model-setup/service.js +329 -0
- package/dist/core/model-setup/service.js.map +1 -0
- package/dist/core/model-setup/store.d.ts +29 -0
- package/dist/core/model-setup/store.js +116 -0
- package/dist/core/model-setup/store.js.map +1 -0
- package/dist/core/model-setup/wiring.d.ts +59 -0
- package/dist/core/model-setup/wiring.js +132 -0
- package/dist/core/model-setup/wiring.js.map +1 -0
- package/dist/core/prism.d.ts +36 -0
- package/dist/core/prism.js +42 -0
- package/dist/core/prism.js.map +1 -0
- package/dist/core/public-server.d.ts +7 -0
- package/dist/core/public-server.js +2 -0
- package/dist/core/public-server.js.map +1 -1
- package/dist/core/sessions.d.ts +6 -1
- package/dist/core/sessions.js +18 -14
- package/dist/core/sessions.js.map +1 -1
- package/dist/core/types.d.ts +27 -1
- package/dist/core/types.js.map +1 -1
- package/dist/decision/args.d.ts +110 -0
- package/dist/decision/args.js +126 -0
- package/dist/decision/args.js.map +1 -0
- package/dist/decision/backoff.d.ts +25 -0
- package/dist/decision/backoff.js +33 -0
- package/dist/decision/backoff.js.map +1 -0
- package/dist/decision/engine-candidates.d.ts +52 -0
- package/dist/decision/engine-candidates.js +138 -0
- package/dist/decision/engine-candidates.js.map +1 -0
- package/dist/decision/engine-fallback.d.ts +41 -0
- package/dist/decision/engine-fallback.js +57 -0
- package/dist/decision/engine-fallback.js.map +1 -0
- package/dist/decision/engine-version.d.ts +27 -0
- package/dist/decision/engine-version.js +54 -0
- package/dist/decision/engine-version.js.map +1 -0
- package/dist/decision/engine.d.ts +114 -0
- package/dist/decision/engine.js +253 -0
- package/dist/decision/engine.js.map +1 -0
- package/dist/decision/http.d.ts +36 -0
- package/dist/decision/http.js +73 -0
- package/dist/decision/http.js.map +1 -0
- package/dist/decision/index.d.ts +20 -0
- package/dist/decision/index.js +21 -0
- package/dist/decision/index.js.map +1 -0
- package/dist/decision/model-facts.d.ts +32 -0
- package/dist/decision/model-facts.js +40 -0
- package/dist/decision/model-facts.js.map +1 -0
- package/dist/decision/outcome.d.ts +35 -0
- package/dist/decision/outcome.js +105 -0
- package/dist/decision/outcome.js.map +1 -0
- package/dist/decision/process.d.ts +67 -0
- package/dist/decision/process.js +133 -0
- package/dist/decision/process.js.map +1 -0
- package/dist/decision/readiness.d.ts +74 -0
- package/dist/decision/readiness.js +165 -0
- package/dist/decision/readiness.js.map +1 -0
- package/dist/decision/request.d.ts +11 -0
- package/dist/decision/request.js +56 -0
- package/dist/decision/request.js.map +1 -0
- package/dist/decision/service.d.ts +202 -0
- package/dist/decision/service.js +711 -0
- package/dist/decision/service.js.map +1 -0
- package/dist/decision/upstream-version.d.ts +23 -0
- package/dist/decision/upstream-version.js +34 -0
- package/dist/decision/upstream-version.js.map +1 -0
- package/dist/decision/wiring.d.ts +67 -0
- package/dist/decision/wiring.js +101 -0
- package/dist/decision/wiring.js.map +1 -0
- package/dist/diffusion/args.d.ts +4 -2
- package/dist/diffusion/args.js +10 -3
- package/dist/diffusion/args.js.map +1 -1
- package/dist/diffusion/image-job.js +1 -1
- package/dist/diffusion/image-job.js.map +1 -1
- package/dist/diffusion/install.d.ts +35 -11
- package/dist/diffusion/install.js +74 -49
- package/dist/diffusion/install.js.map +1 -1
- package/dist/diffusion/job-kind.d.ts +33 -7
- package/dist/diffusion/jobs.d.ts +6 -1
- package/dist/diffusion/jobs.js +60 -18
- package/dist/diffusion/jobs.js.map +1 -1
- package/dist/diffusion/parse.d.ts +1 -2
- package/dist/diffusion/parse.js +1 -13
- package/dist/diffusion/parse.js.map +1 -1
- package/dist/diffusion/progress.js +3 -0
- package/dist/diffusion/progress.js.map +1 -1
- package/dist/diffusion/server-process.d.ts +6 -0
- package/dist/diffusion/server-process.js +3 -2
- package/dist/diffusion/server-process.js.map +1 -1
- package/dist/diffusion/service.d.ts +52 -6
- package/dist/diffusion/service.js +215 -40
- package/dist/diffusion/service.js.map +1 -1
- package/dist/diffusion/session.d.ts +27 -3
- package/dist/diffusion/session.js +108 -17
- package/dist/diffusion/session.js.map +1 -1
- package/dist/diffusion/state.d.ts +19 -1
- package/dist/diffusion/state.js +10 -0
- package/dist/diffusion/state.js.map +1 -1
- package/dist/diffusion/tracker.d.ts +15 -1
- package/dist/diffusion/tracker.js +32 -6
- package/dist/diffusion/tracker.js.map +1 -1
- package/dist/diffusion/types.d.ts +7 -0
- package/dist/diffusion/types.js +11 -0
- package/dist/diffusion/types.js.map +1 -1
- package/dist/diffusion/video-estimate.d.ts +214 -0
- package/dist/diffusion/video-estimate.js +402 -0
- package/dist/diffusion/video-estimate.js.map +1 -0
- package/dist/diffusion/video-eta.d.ts +90 -0
- package/dist/diffusion/video-eta.js +248 -0
- package/dist/diffusion/video-eta.js.map +1 -0
- package/dist/diffusion/video-history.d.ts +45 -0
- package/dist/diffusion/video-history.js +121 -0
- package/dist/diffusion/video-history.js.map +1 -0
- package/dist/diffusion/video-job.d.ts +13 -5
- package/dist/diffusion/video-job.js +42 -15
- package/dist/diffusion/video-job.js.map +1 -1
- package/dist/diffusion/video-recipe.js +3 -0
- package/dist/diffusion/video-recipe.js.map +1 -1
- package/dist/diffusion/video-tiling.d.ts +54 -0
- package/dist/diffusion/video-tiling.js +70 -0
- package/dist/diffusion/video-tiling.js.map +1 -0
- package/dist/diffusion/wiring.d.ts +14 -1
- package/dist/diffusion/wiring.js +7 -0
- package/dist/diffusion/wiring.js.map +1 -1
- package/dist/downloads/protocol.d.ts +3 -8
- package/dist/downloads/protocol.js.map +1 -1
- package/dist/embedding/args.d.ts +39 -0
- package/dist/embedding/args.js +46 -0
- package/dist/embedding/args.js.map +1 -0
- package/dist/embedding/engine.d.ts +50 -0
- package/dist/embedding/engine.js +100 -0
- package/dist/embedding/engine.js.map +1 -0
- package/dist/embedding/ffmpeg.d.ts +18 -0
- package/dist/embedding/ffmpeg.js +63 -0
- package/dist/embedding/ffmpeg.js.map +1 -0
- package/dist/embedding/index.d.ts +16 -0
- package/dist/embedding/index.js +17 -0
- package/dist/embedding/index.js.map +1 -0
- package/dist/embedding/model-facts.d.ts +51 -0
- package/dist/embedding/model-facts.js +84 -0
- package/dist/embedding/model-facts.js.map +1 -0
- package/dist/embedding/process.d.ts +66 -0
- package/dist/embedding/process.js +131 -0
- package/dist/embedding/process.js.map +1 -0
- package/dist/embedding/readiness.d.ts +56 -0
- package/dist/embedding/readiness.js +114 -0
- package/dist/embedding/readiness.js.map +1 -0
- package/dist/embedding/request.d.ts +43 -0
- package/dist/embedding/request.js +285 -0
- package/dist/embedding/request.js.map +1 -0
- package/dist/embedding/service.d.ts +152 -0
- package/dist/embedding/service.js +602 -0
- package/dist/embedding/service.js.map +1 -0
- package/dist/embedding/wiring.d.ts +51 -0
- package/dist/embedding/wiring.js +79 -0
- package/dist/embedding/wiring.js.map +1 -0
- package/dist/engine-builds/builds.d.ts +54 -0
- package/dist/engine-builds/builds.js +151 -0
- package/dist/engine-builds/builds.js.map +1 -0
- package/dist/engine-builds/index.d.ts +10 -0
- package/dist/engine-builds/index.js +11 -0
- package/dist/engine-builds/index.js.map +1 -0
- package/dist/engine-builds/install.d.ts +42 -0
- package/dist/engine-builds/install.js +100 -0
- package/dist/engine-builds/install.js.map +1 -0
- package/dist/engine-builds/manifest.d.ts +121 -0
- package/dist/engine-builds/manifest.js +254 -0
- package/dist/engine-builds/manifest.js.map +1 -0
- package/dist/engine-builds/order.d.ts +24 -0
- package/dist/engine-builds/order.js +44 -0
- package/dist/engine-builds/order.js.map +1 -0
- package/dist/engine-builds/select.d.ts +123 -0
- package/dist/engine-builds/select.js +223 -0
- package/dist/engine-builds/select.js.map +1 -0
- package/dist/engine-builds/service.d.ts +123 -0
- package/dist/engine-builds/service.js +392 -0
- package/dist/engine-builds/service.js.map +1 -0
- package/dist/engines/descriptor-order.d.ts +28 -0
- package/dist/engines/descriptor-order.js +60 -0
- package/dist/engines/descriptor-order.js.map +1 -0
- package/dist/engines/engine-build.d.ts +23 -0
- package/dist/engines/engine-build.js +58 -0
- package/dist/engines/engine-build.js.map +1 -0
- package/dist/engines/index.d.ts +10 -0
- package/dist/engines/index.js +11 -0
- package/dist/engines/index.js.map +1 -0
- package/dist/engines/llamacpp.d.ts +71 -0
- package/dist/engines/llamacpp.js +176 -0
- package/dist/engines/llamacpp.js.map +1 -0
- package/dist/engines/managed.d.ts +49 -0
- package/dist/engines/managed.js +88 -0
- package/dist/engines/managed.js.map +1 -0
- package/dist/engines/service.d.ts +46 -0
- package/dist/engines/service.js +61 -0
- package/dist/engines/service.js.map +1 -0
- package/dist/engines/versions.d.ts +65 -0
- package/dist/engines/versions.js +259 -0
- package/dist/engines/versions.js.map +1 -0
- package/dist/hardware/cpu-flags.d.ts +44 -0
- package/dist/hardware/cpu-flags.js +206 -0
- package/dist/hardware/cpu-flags.js.map +1 -0
- package/dist/hardware/drm-sysfs.d.ts +51 -0
- package/dist/hardware/drm-sysfs.js +67 -0
- package/dist/hardware/drm-sysfs.js.map +1 -0
- package/dist/hardware/facts.d.ts +42 -0
- package/dist/hardware/facts.js +92 -0
- package/dist/hardware/facts.js.map +1 -0
- package/dist/hardware/index.d.ts +19 -3
- package/dist/hardware/index.js +19 -3
- package/dist/hardware/index.js.map +1 -1
- package/dist/hardware/merge.d.ts +55 -0
- package/dist/hardware/merge.js +141 -0
- package/dist/hardware/merge.js.map +1 -0
- package/dist/hardware/nvidia-smi.d.ts +43 -0
- package/dist/hardware/nvidia-smi.js +118 -0
- package/dist/hardware/nvidia-smi.js.map +1 -0
- package/dist/hardware/probe-common.d.ts +72 -0
- package/dist/hardware/probe-common.js +106 -0
- package/dist/hardware/probe-common.js.map +1 -0
- package/dist/hardware/probe-darwin.d.ts +20 -0
- package/dist/hardware/probe-darwin.js +76 -0
- package/dist/hardware/probe-darwin.js.map +1 -0
- package/dist/hardware/probe-linux.d.ts +9 -0
- package/dist/hardware/probe-linux.js +133 -0
- package/dist/hardware/probe-linux.js.map +1 -0
- package/dist/hardware/probe-windows.d.ts +7 -0
- package/dist/hardware/probe-windows.js +50 -0
- package/dist/hardware/probe-windows.js.map +1 -0
- package/dist/hardware/probe.d.ts +16 -0
- package/dist/hardware/probe.js +65 -0
- package/dist/hardware/probe.js.map +1 -0
- package/dist/hardware/service.d.ts +60 -0
- package/dist/hardware/service.js +114 -0
- package/dist/hardware/service.js.map +1 -0
- package/dist/hardware/vulkan.d.ts +47 -0
- package/dist/hardware/vulkan.js +148 -0
- package/dist/hardware/vulkan.js.map +1 -0
- package/dist/hardware/windows-video.d.ts +97 -0
- package/dist/hardware/windows-video.js +214 -0
- package/dist/hardware/windows-video.js.map +1 -0
- package/dist/host/app-daemon.d.ts +47 -0
- package/dist/host/app-daemon.js +146 -0
- package/dist/host/app-daemon.js.map +1 -0
- package/dist/host/index.d.ts +5 -3
- package/dist/host/index.js +5 -3
- package/dist/host/index.js.map +1 -1
- package/dist/host/log-file.d.ts +74 -0
- package/dist/host/log-file.js +214 -0
- package/dist/host/log-file.js.map +1 -0
- package/dist/host/recipes/enable-wsl.d.ts +63 -0
- package/dist/host/recipes/enable-wsl.js +73 -0
- package/dist/host/recipes/enable-wsl.js.map +1 -0
- package/dist/host/recipes/executor-io-windows.d.ts +69 -0
- package/dist/host/recipes/executor-io-windows.js +235 -0
- package/dist/host/recipes/executor-io-windows.js.map +1 -0
- package/dist/host/recipes/executor-io.d.ts +80 -0
- package/dist/host/recipes/executor-io.js +211 -0
- package/dist/host/recipes/executor-io.js.map +1 -0
- package/dist/host/recipes/executor.d.ts +81 -0
- package/dist/host/recipes/executor.js +678 -0
- package/dist/host/recipes/executor.js.map +1 -0
- package/dist/host/recipes/guest-executor.d.ts +29 -0
- package/dist/host/recipes/guest-executor.js +76 -0
- package/dist/host/recipes/guest-executor.js.map +1 -0
- package/dist/host/recipes/index.d.ts +16 -0
- package/dist/host/recipes/index.js +17 -0
- package/dist/host/recipes/index.js.map +1 -0
- package/dist/host/recipes/install-container-runtime.d.ts +389 -0
- package/dist/host/recipes/install-container-runtime.js +796 -0
- package/dist/host/recipes/install-container-runtime.js.map +1 -0
- package/dist/host/recipes/openpgp.d.ts +23 -0
- package/dist/host/recipes/openpgp.js +153 -0
- package/dist/host/recipes/openpgp.js.map +1 -0
- package/dist/host/recipes/request-file.d.ts +96 -0
- package/dist/host/recipes/request-file.js +103 -0
- package/dist/host/recipes/request-file.js.map +1 -0
- package/dist/lock/process-journal.d.ts +5 -2
- package/dist/lock/process-journal.js.map +1 -1
- package/dist/models/capabilities.d.ts +7 -0
- package/dist/models/capabilities.js +17 -1
- package/dist/models/capabilities.js.map +1 -1
- package/dist/models/compatibility/index.d.ts +10 -0
- package/dist/models/compatibility/index.js +11 -0
- package/dist/models/compatibility/index.js.map +1 -0
- package/dist/models/compatibility/inspect.d.ts +19 -0
- package/dist/models/compatibility/inspect.js +59 -0
- package/dist/models/compatibility/inspect.js.map +1 -0
- package/dist/models/compatibility/resolve.d.ts +59 -0
- package/dist/models/compatibility/resolve.js +219 -0
- package/dist/models/compatibility/resolve.js.map +1 -0
- package/dist/models/compatibility/rules-baseline.d.ts +2 -0
- package/dist/models/compatibility/rules-baseline.js +10 -0
- package/dist/models/compatibility/rules-baseline.js.map +1 -0
- package/dist/models/compatibility/rules-baseline.json +292 -0
- package/dist/models/compatibility/rules-service.d.ts +36 -0
- package/dist/models/compatibility/rules-service.js +109 -0
- package/dist/models/compatibility/rules-service.js.map +1 -0
- package/dist/models/compatibility/rules.d.ts +81 -0
- package/dist/models/compatibility/rules.js +164 -0
- package/dist/models/compatibility/rules.js.map +1 -0
- package/dist/models/compatibility/service.d.ts +59 -0
- package/dist/models/compatibility/service.js +137 -0
- package/dist/models/compatibility/service.js.map +1 -0
- package/dist/models/decision-checkpoint.d.ts +22 -0
- package/dist/models/decision-checkpoint.js +33 -0
- package/dist/models/decision-checkpoint.js.map +1 -0
- package/dist/models/gguf/classify.d.ts +27 -1
- package/dist/models/gguf/classify.js +78 -2
- package/dist/models/gguf/classify.js.map +1 -1
- package/dist/models/gguf/index.d.ts +1 -0
- package/dist/models/gguf/index.js +1 -0
- package/dist/models/gguf/index.js.map +1 -1
- package/dist/models/gguf/tensors.d.ts +56 -0
- package/dist/models/gguf/tensors.js +236 -0
- package/dist/models/gguf/tensors.js.map +1 -0
- package/dist/models/index.d.ts +2 -0
- package/dist/models/index.js +2 -0
- package/dist/models/index.js.map +1 -1
- package/dist/router/resolve.d.ts +7 -1
- package/dist/router/resolve.js +14 -1
- package/dist/router/resolve.js.map +1 -1
- package/dist/runtime/container/argv.d.ts +88 -0
- package/dist/runtime/container/argv.js +406 -0
- package/dist/runtime/container/argv.js.map +1 -0
- package/dist/runtime/container/docker-binary.d.ts +4 -0
- package/dist/runtime/container/docker-binary.js +34 -0
- package/dist/runtime/container/docker-binary.js.map +1 -0
- package/dist/runtime/container/env.d.ts +17 -0
- package/dist/runtime/container/env.js +52 -0
- package/dist/runtime/container/env.js.map +1 -0
- package/dist/runtime/container/exec.d.ts +27 -0
- package/dist/runtime/container/exec.js +79 -0
- package/dist/runtime/container/exec.js.map +1 -0
- package/dist/runtime/container/execution-journal.d.ts +55 -0
- package/dist/runtime/container/execution-journal.js +167 -0
- package/dist/runtime/container/execution-journal.js.map +1 -0
- package/dist/runtime/container/heartbeat.d.ts +47 -0
- package/dist/runtime/container/heartbeat.js +78 -0
- package/dist/runtime/container/heartbeat.js.map +1 -0
- package/dist/runtime/container/index.d.ts +30 -0
- package/dist/runtime/container/index.js +31 -0
- package/dist/runtime/container/index.js.map +1 -0
- package/dist/runtime/container/operations.d.ts +40 -0
- package/dist/runtime/container/operations.js +233 -0
- package/dist/runtime/container/operations.js.map +1 -0
- package/dist/runtime/container/pull.d.ts +48 -0
- package/dist/runtime/container/pull.js +280 -0
- package/dist/runtime/container/pull.js.map +1 -0
- package/dist/runtime/container/reconcile.d.ts +34 -0
- package/dist/runtime/container/reconcile.js +79 -0
- package/dist/runtime/container/reconcile.js.map +1 -0
- package/dist/runtime/container/types.d.ts +124 -0
- package/dist/runtime/container/types.js +2 -0
- package/dist/runtime/container/types.js.map +1 -0
- package/dist/runtime/container/watchdog.d.ts +114 -0
- package/dist/runtime/container/watchdog.js +418 -0
- package/dist/runtime/container/watchdog.js.map +1 -0
- package/dist/runtime/container/wiring.d.ts +78 -0
- package/dist/runtime/container/wiring.js +92 -0
- package/dist/runtime/container/wiring.js.map +1 -0
- package/dist/runtime/container/wsl-exec.d.ts +13 -0
- package/dist/runtime/container/wsl-exec.js +18 -0
- package/dist/runtime/container/wsl-exec.js.map +1 -0
- package/dist/runtime/environment/cached-document.d.ts +128 -0
- package/dist/runtime/environment/cached-document.js +218 -0
- package/dist/runtime/environment/cached-document.js.map +1 -0
- package/dist/runtime/environment/canonical-json.d.ts +78 -0
- package/dist/runtime/environment/canonical-json.js +86 -0
- package/dist/runtime/environment/canonical-json.js.map +1 -0
- package/dist/runtime/environment/descriptor-provider.d.ts +121 -0
- package/dist/runtime/environment/descriptor-provider.js +159 -0
- package/dist/runtime/environment/descriptor-provider.js.map +1 -0
- package/dist/runtime/environment/descriptor.d.ts +35 -0
- package/dist/runtime/environment/descriptor.js +239 -0
- package/dist/runtime/environment/descriptor.js.map +1 -0
- package/dist/runtime/environment/diagnostics.d.ts +50 -0
- package/dist/runtime/environment/diagnostics.js +114 -0
- package/dist/runtime/environment/diagnostics.js.map +1 -0
- package/dist/runtime/environment/document-fields.d.ts +39 -0
- package/dist/runtime/environment/document-fields.js +96 -0
- package/dist/runtime/environment/document-fields.js.map +1 -0
- package/dist/runtime/environment/environment-manifest-provider.d.ts +65 -0
- package/dist/runtime/environment/environment-manifest-provider.js +89 -0
- package/dist/runtime/environment/environment-manifest-provider.js.map +1 -0
- package/dist/runtime/environment/environment-manifest.d.ts +35 -0
- package/dist/runtime/environment/environment-manifest.js +131 -0
- package/dist/runtime/environment/environment-manifest.js.map +1 -0
- package/dist/runtime/environment/guest-host.d.ts +43 -0
- package/dist/runtime/environment/guest-host.js +88 -0
- package/dist/runtime/environment/guest-host.js.map +1 -0
- package/dist/runtime/environment/host-exec.d.ts +38 -0
- package/dist/runtime/environment/host-exec.js +145 -0
- package/dist/runtime/environment/host-exec.js.map +1 -0
- package/dist/runtime/environment/host-step-failure.d.ts +33 -0
- package/dist/runtime/environment/host-step-failure.js +34 -0
- package/dist/runtime/environment/host-step-failure.js.map +1 -0
- package/dist/runtime/environment/index.d.ts +56 -0
- package/dist/runtime/environment/index.js +57 -0
- package/dist/runtime/environment/index.js.map +1 -0
- package/dist/runtime/environment/installations.d.ts +25 -0
- package/dist/runtime/environment/installations.js +70 -0
- package/dist/runtime/environment/installations.js.map +1 -0
- package/dist/runtime/environment/inventory.d.ts +42 -0
- package/dist/runtime/environment/inventory.js +57 -0
- package/dist/runtime/environment/inventory.js.map +1 -0
- package/dist/runtime/environment/linux-blockers.d.ts +130 -0
- package/dist/runtime/environment/linux-blockers.js +251 -0
- package/dist/runtime/environment/linux-blockers.js.map +1 -0
- package/dist/runtime/environment/linux-docker-facts.d.ts +200 -0
- package/dist/runtime/environment/linux-docker-facts.js +342 -0
- package/dist/runtime/environment/linux-docker-facts.js.map +1 -0
- package/dist/runtime/environment/linux-docker-network.d.ts +80 -0
- package/dist/runtime/environment/linux-docker-network.js +195 -0
- package/dist/runtime/environment/linux-docker-network.js.map +1 -0
- package/dist/runtime/environment/linux-host.d.ts +22 -0
- package/dist/runtime/environment/linux-host.js +98 -0
- package/dist/runtime/environment/linux-host.js.map +1 -0
- package/dist/runtime/environment/linux-plan.d.ts +126 -0
- package/dist/runtime/environment/linux-plan.js +392 -0
- package/dist/runtime/environment/linux-plan.js.map +1 -0
- package/dist/runtime/environment/linux-probe.d.ts +245 -0
- package/dist/runtime/environment/linux-probe.js +315 -0
- package/dist/runtime/environment/linux-probe.js.map +1 -0
- package/dist/runtime/environment/linux-provisioner.d.ts +102 -0
- package/dist/runtime/environment/linux-provisioner.js +734 -0
- package/dist/runtime/environment/linux-provisioner.js.map +1 -0
- package/dist/runtime/environment/provisioner-descriptors.d.ts +17 -0
- package/dist/runtime/environment/provisioner-descriptors.js +62 -0
- package/dist/runtime/environment/provisioner-descriptors.js.map +1 -0
- package/dist/runtime/environment/recovery.d.ts +74 -0
- package/dist/runtime/environment/recovery.js +121 -0
- package/dist/runtime/environment/recovery.js.map +1 -0
- package/dist/runtime/environment/rootfs-download.d.ts +2 -0
- package/dist/runtime/environment/rootfs-download.js +47 -0
- package/dist/runtime/environment/rootfs-download.js.map +1 -0
- package/dist/runtime/environment/service.d.ts +280 -0
- package/dist/runtime/environment/service.js +776 -0
- package/dist/runtime/environment/service.js.map +1 -0
- package/dist/runtime/environment/state.d.ts +163 -0
- package/dist/runtime/environment/state.js +479 -0
- package/dist/runtime/environment/state.js.map +1 -0
- package/dist/runtime/environment/store-models.d.ts +19 -0
- package/dist/runtime/environment/store-models.js +30 -0
- package/dist/runtime/environment/store-models.js.map +1 -0
- package/dist/runtime/environment/store.d.ts +210 -0
- package/dist/runtime/environment/store.js +438 -0
- package/dist/runtime/environment/store.js.map +1 -0
- package/dist/runtime/environment/windows-environment-record.d.ts +48 -0
- package/dist/runtime/environment/windows-environment-record.js +89 -0
- package/dist/runtime/environment/windows-environment-record.js.map +1 -0
- package/dist/runtime/environment/windows-forwarding.d.ts +27 -0
- package/dist/runtime/environment/windows-forwarding.js +84 -0
- package/dist/runtime/environment/windows-forwarding.js.map +1 -0
- package/dist/runtime/environment/windows-guest-setup.d.ts +48 -0
- package/dist/runtime/environment/windows-guest-setup.js +157 -0
- package/dist/runtime/environment/windows-guest-setup.js.map +1 -0
- package/dist/runtime/environment/windows-host.d.ts +46 -0
- package/dist/runtime/environment/windows-host.js +163 -0
- package/dist/runtime/environment/windows-host.js.map +1 -0
- package/dist/runtime/environment/windows-plan.d.ts +78 -0
- package/dist/runtime/environment/windows-plan.js +216 -0
- package/dist/runtime/environment/windows-plan.js.map +1 -0
- package/dist/runtime/environment/windows-probe.d.ts +135 -0
- package/dist/runtime/environment/windows-probe.js +261 -0
- package/dist/runtime/environment/windows-probe.js.map +1 -0
- package/dist/runtime/environment/windows-provisioner.d.ts +113 -0
- package/dist/runtime/environment/windows-provisioner.js +782 -0
- package/dist/runtime/environment/windows-provisioner.js.map +1 -0
- package/dist/runtime/environment/wiring.d.ts +117 -0
- package/dist/runtime/environment/wiring.js +354 -0
- package/dist/runtime/environment/wiring.js.map +1 -0
- package/dist/runtime/foundation-models/runtime.d.ts +8 -0
- package/dist/runtime/foundation-models/runtime.js +5 -2
- package/dist/runtime/foundation-models/runtime.js.map +1 -1
- package/dist/runtime/llamacpp/args.d.ts +10 -1
- package/dist/runtime/llamacpp/args.js +43 -7
- package/dist/runtime/llamacpp/args.js.map +1 -1
- package/dist/runtime/llamacpp/fit-margin.d.ts +46 -0
- package/dist/runtime/llamacpp/fit-margin.js +47 -0
- package/dist/runtime/llamacpp/fit-margin.js.map +1 -0
- package/dist/runtime/llamacpp/index.d.ts +1 -0
- package/dist/runtime/llamacpp/index.js +1 -0
- package/dist/runtime/llamacpp/index.js.map +1 -1
- package/dist/runtime/llamacpp/load-plan.d.ts +23 -1
- package/dist/runtime/llamacpp/load-plan.js +82 -1
- package/dist/runtime/llamacpp/load-plan.js.map +1 -1
- package/dist/runtime/llamacpp/policy.d.ts +16 -0
- package/dist/runtime/llamacpp/policy.js +20 -0
- package/dist/runtime/llamacpp/policy.js.map +1 -1
- package/dist/runtime/llamacpp/probe.d.ts +21 -1
- package/dist/runtime/llamacpp/probe.js +32 -6
- package/dist/runtime/llamacpp/probe.js.map +1 -1
- package/dist/runtime/llamacpp/runtime.d.ts +56 -1
- package/dist/runtime/llamacpp/runtime.js +142 -15
- package/dist/runtime/llamacpp/runtime.js.map +1 -1
- package/dist/runtime/managed-engines/index.d.ts +8 -0
- package/dist/runtime/managed-engines/index.js +9 -0
- package/dist/runtime/managed-engines/index.js.map +1 -0
- package/dist/runtime/managed-engines/runtime.d.ts +140 -0
- package/dist/runtime/managed-engines/runtime.js +402 -0
- package/dist/runtime/managed-engines/runtime.js.map +1 -0
- package/dist/runtime/managed-engines/spec.d.ts +70 -0
- package/dist/runtime/managed-engines/spec.js +39 -0
- package/dist/runtime/managed-engines/spec.js.map +1 -0
- package/dist/runtime/managed-models/check.d.ts +50 -0
- package/dist/runtime/managed-models/check.js +157 -0
- package/dist/runtime/managed-models/check.js.map +1 -0
- package/dist/runtime/managed-models/compatibility.d.ts +256 -0
- package/dist/runtime/managed-models/compatibility.js +594 -0
- package/dist/runtime/managed-models/compatibility.js.map +1 -0
- package/dist/runtime/managed-models/delete.d.ts +15 -0
- package/dist/runtime/managed-models/delete.js +60 -0
- package/dist/runtime/managed-models/delete.js.map +1 -0
- package/dist/runtime/managed-models/guest-files.d.ts +8 -0
- package/dist/runtime/managed-models/guest-files.js +49 -0
- package/dist/runtime/managed-models/guest-files.js.map +1 -0
- package/dist/runtime/managed-models/index.d.ts +17 -0
- package/dist/runtime/managed-models/index.js +18 -0
- package/dist/runtime/managed-models/index.js.map +1 -0
- package/dist/runtime/managed-models/location.d.ts +22 -0
- package/dist/runtime/managed-models/location.js +53 -0
- package/dist/runtime/managed-models/location.js.map +1 -0
- package/dist/runtime/managed-models/migrate.d.ts +30 -0
- package/dist/runtime/managed-models/migrate.js +119 -0
- package/dist/runtime/managed-models/migrate.js.map +1 -0
- package/dist/runtime/managed-models/model-dir.d.ts +36 -0
- package/dist/runtime/managed-models/model-dir.js +136 -0
- package/dist/runtime/managed-models/model-dir.js.map +1 -0
- package/dist/runtime/managed-models/quant-format.d.ts +71 -0
- package/dist/runtime/managed-models/quant-format.js +299 -0
- package/dist/runtime/managed-models/quant-format.js.map +1 -0
- package/dist/runtime/managed-models/registry.d.ts +26 -0
- package/dist/runtime/managed-models/registry.js +66 -0
- package/dist/runtime/managed-models/registry.js.map +1 -0
- package/dist/runtime/managed-text/adapter.d.ts +266 -0
- package/dist/runtime/managed-text/adapter.js +96 -0
- package/dist/runtime/managed-text/adapter.js.map +1 -0
- package/dist/runtime/managed-text/backend-target.d.ts +3 -0
- package/dist/runtime/managed-text/backend-target.js +52 -0
- package/dist/runtime/managed-text/backend-target.js.map +1 -0
- package/dist/runtime/managed-text/deployment.d.ts +14 -0
- package/dist/runtime/managed-text/deployment.js +32 -0
- package/dist/runtime/managed-text/deployment.js.map +1 -0
- package/dist/runtime/managed-text/engine-cache.d.ts +15 -0
- package/dist/runtime/managed-text/engine-cache.js +64 -0
- package/dist/runtime/managed-text/engine-cache.js.map +1 -0
- package/dist/runtime/managed-text/gateway.d.ts +158 -0
- package/dist/runtime/managed-text/gateway.js +535 -0
- package/dist/runtime/managed-text/gateway.js.map +1 -0
- package/dist/runtime/managed-text/generation-defaults.d.ts +23 -0
- package/dist/runtime/managed-text/generation-defaults.js +51 -0
- package/dist/runtime/managed-text/generation-defaults.js.map +1 -0
- package/dist/runtime/managed-text/index.d.ts +26 -0
- package/dist/runtime/managed-text/index.js +27 -0
- package/dist/runtime/managed-text/index.js.map +1 -0
- package/dist/runtime/managed-text/lifecycle.d.ts +276 -0
- package/dist/runtime/managed-text/lifecycle.js +859 -0
- package/dist/runtime/managed-text/lifecycle.js.map +1 -0
- package/dist/runtime/managed-text/load-policy.d.ts +47 -0
- package/dist/runtime/managed-text/load-policy.js +111 -0
- package/dist/runtime/managed-text/load-policy.js.map +1 -0
- package/dist/runtime/managed-text/mount-source.d.ts +14 -0
- package/dist/runtime/managed-text/mount-source.js +18 -0
- package/dist/runtime/managed-text/mount-source.js.map +1 -0
- package/dist/runtime/managed-text/readiness.d.ts +13 -0
- package/dist/runtime/managed-text/readiness.js +24 -0
- package/dist/runtime/managed-text/readiness.js.map +1 -0
- package/dist/runtime/managed-text/request-rules.d.ts +18 -0
- package/dist/runtime/managed-text/request-rules.js +62 -0
- package/dist/runtime/managed-text/request-rules.js.map +1 -0
- package/dist/runtime/managed-text/types.d.ts +88 -0
- package/dist/runtime/managed-text/types.js +12 -0
- package/dist/runtime/managed-text/types.js.map +1 -0
- package/dist/runtime/managed-text/wsl-deployment.d.ts +24 -0
- package/dist/runtime/managed-text/wsl-deployment.js +70 -0
- package/dist/runtime/managed-text/wsl-deployment.js.map +1 -0
- package/dist/runtime/mlx/errors.d.ts +6 -1
- package/dist/runtime/mlx/errors.js +7 -2
- package/dist/runtime/mlx/errors.js.map +1 -1
- package/dist/runtime/mlx/runtime.d.ts +36 -1
- package/dist/runtime/mlx/runtime.js +85 -4
- package/dist/runtime/mlx/runtime.js.map +1 -1
- package/dist/runtime/shared/backend-output.d.ts +23 -0
- package/dist/runtime/shared/backend-output.js +32 -0
- package/dist/runtime/shared/backend-output.js.map +1 -0
- package/dist/runtime/shared/bounded-output.d.ts +24 -0
- package/dist/runtime/shared/bounded-output.js +63 -0
- package/dist/runtime/shared/bounded-output.js.map +1 -0
- package/dist/runtime/shared/external-sessions.js +9 -1
- package/dist/runtime/shared/external-sessions.js.map +1 -1
- package/dist/runtime/shared/gpu-occupancy.d.ts +60 -0
- package/dist/runtime/shared/gpu-occupancy.js +10 -0
- package/dist/runtime/shared/gpu-occupancy.js.map +1 -0
- package/dist/runtime/shared/index.d.ts +4 -0
- package/dist/runtime/shared/index.js +4 -0
- package/dist/runtime/shared/index.js.map +1 -1
- package/dist/runtime/shared/local-runtime.d.ts +35 -0
- package/dist/runtime/shared/ports.d.ts +6 -0
- package/dist/runtime/shared/ports.js +12 -1
- package/dist/runtime/shared/ports.js.map +1 -1
- package/dist/runtime/shared/process.d.ts +11 -0
- package/dist/runtime/shared/process.js +20 -0
- package/dist/runtime/shared/process.js.map +1 -1
- package/dist/runtime/shared/redact-args.d.ts +7 -0
- package/dist/runtime/shared/redact-args.js +26 -0
- package/dist/runtime/shared/redact-args.js.map +1 -0
- package/dist/runtime/shared/sidecar.d.ts +15 -0
- package/dist/runtime/shared/sidecar.js +43 -8
- package/dist/runtime/shared/sidecar.js.map +1 -1
- package/dist/runtime/tensorrt-llm/adapter.d.ts +250 -0
- package/dist/runtime/tensorrt-llm/adapter.js +1011 -0
- package/dist/runtime/tensorrt-llm/adapter.js.map +1 -0
- package/dist/runtime/tensorrt-llm/check.d.ts +12 -0
- package/dist/runtime/tensorrt-llm/check.js +21 -0
- package/dist/runtime/tensorrt-llm/check.js.map +1 -0
- package/dist/runtime/tensorrt-llm/compatibility.d.ts +100 -0
- package/dist/runtime/tensorrt-llm/compatibility.js +181 -0
- package/dist/runtime/tensorrt-llm/compatibility.js.map +1 -0
- package/dist/runtime/tensorrt-llm/engine.d.ts +3 -0
- package/dist/runtime/tensorrt-llm/engine.js +24 -0
- package/dist/runtime/tensorrt-llm/engine.js.map +1 -0
- package/dist/runtime/tensorrt-llm/host-facts.d.ts +55 -0
- package/dist/runtime/tensorrt-llm/host-facts.js +87 -0
- package/dist/runtime/tensorrt-llm/host-facts.js.map +1 -0
- package/dist/runtime/tensorrt-llm/index.d.ts +30 -0
- package/dist/runtime/tensorrt-llm/index.js +31 -0
- package/dist/runtime/tensorrt-llm/index.js.map +1 -0
- package/dist/runtime/tensorrt-llm/installation.d.ts +23 -0
- package/dist/runtime/tensorrt-llm/installation.js +53 -0
- package/dist/runtime/tensorrt-llm/installation.js.map +1 -0
- package/dist/runtime/tensorrt-llm/kv-cache.d.ts +25 -0
- package/dist/runtime/tensorrt-llm/kv-cache.js +28 -0
- package/dist/runtime/tensorrt-llm/kv-cache.js.map +1 -0
- package/dist/runtime/tensorrt-llm/prelaunch.d.ts +20 -0
- package/dist/runtime/tensorrt-llm/prelaunch.js +168 -0
- package/dist/runtime/tensorrt-llm/prelaunch.js.map +1 -0
- package/dist/runtime/tensorrt-llm/route-policy.d.ts +19 -0
- package/dist/runtime/tensorrt-llm/route-policy.js +17 -0
- package/dist/runtime/tensorrt-llm/route-policy.js.map +1 -0
- package/dist/runtime/tensorrt-llm/settings.d.ts +7 -0
- package/dist/runtime/tensorrt-llm/settings.js +48 -0
- package/dist/runtime/tensorrt-llm/settings.js.map +1 -0
- package/dist/runtime/vllm/adapter.d.ts +46 -0
- package/dist/runtime/vllm/adapter.js +425 -0
- package/dist/runtime/vllm/adapter.js.map +1 -0
- package/dist/runtime/vllm/engine.d.ts +16 -0
- package/dist/runtime/vllm/engine.js +37 -0
- package/dist/runtime/vllm/engine.js.map +1 -0
- package/dist/runtime/vllm/index.d.ts +11 -0
- package/dist/runtime/vllm/index.js +12 -0
- package/dist/runtime/vllm/index.js.map +1 -0
- package/dist/runtime/vllm/memory.d.ts +69 -0
- package/dist/runtime/vllm/memory.js +123 -0
- package/dist/runtime/vllm/memory.js.map +1 -0
- package/dist/runtime/vllm/settings.d.ts +52 -0
- package/dist/runtime/vllm/settings.js +151 -0
- package/dist/runtime/vllm/settings.js.map +1 -0
- package/dist/runtime/wsl/guest-paths.d.ts +33 -0
- package/dist/runtime/wsl/guest-paths.js +83 -0
- package/dist/runtime/wsl/guest-paths.js.map +1 -0
- package/dist/runtime/wsl/guest-scope.d.ts +14 -0
- package/dist/runtime/wsl/guest-scope.js +98 -0
- package/dist/runtime/wsl/guest-scope.js.map +1 -0
- package/dist/runtime/wsl/index.d.ts +12 -0
- package/dist/runtime/wsl/index.js +13 -0
- package/dist/runtime/wsl/index.js.map +1 -0
- package/dist/runtime/wsl/keeper.d.ts +29 -0
- package/dist/runtime/wsl/keeper.js +48 -0
- package/dist/runtime/wsl/keeper.js.map +1 -0
- package/dist/runtime/wsl/transport.d.ts +106 -0
- package/dist/runtime/wsl/transport.js +181 -0
- package/dist/runtime/wsl/transport.js.map +1 -0
- package/dist/server/control/index.d.ts +1 -1
- package/dist/server/control/router.js +15 -0
- package/dist/server/control/router.js.map +1 -1
- package/dist/server/control/routes/backends.js +26 -0
- package/dist/server/control/routes/backends.js.map +1 -1
- package/dist/server/control/routes/decision.d.ts +10 -0
- package/dist/server/control/routes/decision.js +43 -0
- package/dist/server/control/routes/decision.js.map +1 -0
- package/dist/server/control/routes/diffusion-video.js +4 -0
- package/dist/server/control/routes/diffusion-video.js.map +1 -1
- package/dist/server/control/routes/diffusion.d.ts +1 -1
- package/dist/server/control/routes/diffusion.js +4 -9
- package/dist/server/control/routes/diffusion.js.map +1 -1
- package/dist/server/control/routes/embedding.d.ts +9 -0
- package/dist/server/control/routes/embedding.js +43 -0
- package/dist/server/control/routes/embedding.js.map +1 -0
- package/dist/server/control/routes/engine-builds.d.ts +14 -0
- package/dist/server/control/routes/engine-builds.js +82 -0
- package/dist/server/control/routes/engine-builds.js.map +1 -0
- package/dist/server/control/routes/engines.d.ts +17 -0
- package/dist/server/control/routes/engines.js +152 -0
- package/dist/server/control/routes/engines.js.map +1 -0
- package/dist/server/control/routes/environments.d.ts +16 -0
- package/dist/server/control/routes/environments.js +207 -0
- package/dist/server/control/routes/environments.js.map +1 -0
- package/dist/server/control/routes/hardware.d.ts +1 -1
- package/dist/server/control/routes/hardware.js +11 -6
- package/dist/server/control/routes/hardware.js.map +1 -1
- package/dist/server/control/routes/model-setups.d.ts +11 -0
- package/dist/server/control/routes/model-setups.js +55 -0
- package/dist/server/control/routes/model-setups.js.map +1 -0
- package/dist/server/control/routes/models.d.ts +3 -2
- package/dist/server/control/routes/models.js +45 -2
- package/dist/server/control/routes/models.js.map +1 -1
- package/dist/server/control/types.d.ts +167 -9
- package/dist/server/control/types.js.map +1 -1
- package/dist/server/http.d.ts +9 -1
- package/dist/server/http.js +51 -1
- package/dist/server/http.js.map +1 -1
- package/dist/server/public/decision-preview.d.ts +17 -0
- package/dist/server/public/decision-preview.js +119 -0
- package/dist/server/public/decision-preview.js.map +1 -0
- package/dist/server/public/decision.d.ts +32 -0
- package/dist/server/public/decision.js +135 -0
- package/dist/server/public/decision.js.map +1 -0
- package/dist/server/public/embedding.d.ts +31 -0
- package/dist/server/public/embedding.js +126 -0
- package/dist/server/public/embedding.js.map +1 -0
- package/dist/server/public/errors.d.ts +2 -1
- package/dist/server/public/errors.js +5 -2
- package/dist/server/public/errors.js.map +1 -1
- package/dist/server/public/forward.js +75 -3
- package/dist/server/public/forward.js.map +1 -1
- package/dist/server/public/gates.d.ts +13 -0
- package/dist/server/public/gates.js +37 -14
- package/dist/server/public/gates.js.map +1 -1
- package/dist/server/public/index.d.ts +5 -2
- package/dist/server/public/index.js +14 -1
- package/dist/server/public/index.js.map +1 -1
- package/dist/server/public/listing.d.ts +13 -2
- package/dist/server/public/listing.js +33 -9
- package/dist/server/public/listing.js.map +1 -1
- package/dist/server/public/policy.d.ts +9 -0
- package/dist/server/public/policy.js +68 -0
- package/dist/server/public/policy.js.map +1 -0
- package/dist/server/public/responses.js +10 -0
- package/dist/server/public/responses.js.map +1 -1
- package/dist/server/public/trace.d.ts +2 -2
- package/dist/server/public/trace.js +6 -2
- package/dist/server/public/trace.js.map +1 -1
- package/dist/server/public/types.d.ts +84 -1
- package/dist/settings/decision.d.ts +15 -0
- package/dist/settings/decision.js +88 -0
- package/dist/settings/decision.js.map +1 -0
- package/dist/settings/embedding.d.ts +12 -0
- package/dist/settings/embedding.js +82 -0
- package/dist/settings/embedding.js.map +1 -0
- package/dist/settings/index.d.ts +2 -0
- package/dist/settings/index.js +2 -0
- package/dist/settings/index.js.map +1 -1
- package/dist/settings/schema/atomic-prism.json +433 -0
- package/dist/settings/schema/tensorrt-llm.json +253 -0
- package/dist/settings/schema/vllm.json +273 -0
- package/dist/settings/schema.js +58 -0
- package/dist/settings/schema.js.map +1 -1
- package/dist/settings/store.d.ts +13 -2
- package/dist/settings/store.js +44 -0
- package/dist/settings/store.js.map +1 -1
- package/dist/util/index.d.ts +1 -0
- package/dist/util/index.js +1 -0
- package/dist/util/index.js.map +1 -1
- package/dist/util/result.d.ts +19 -0
- package/dist/util/result.js +15 -0
- package/dist/util/result.js.map +1 -0
- package/dist/version.d.ts +1 -1
- package/dist/version.js +1 -1
- package/dist/version.js.map +1 -1
- package/package.json +1 -1
- package/src/settings/schema/atomic-prism.json +433 -0
- package/src/settings/schema/tensorrt-llm.json +253 -0
- package/src/settings/schema/vllm.json +273 -0
|
@@ -1 +1 @@
|
|
|
1
|
-
{"version":3,"file":"policy.js","sourceRoot":"","sources":["../../../src/runtime/llamacpp/policy.ts"],"names":[],"mappings":"AAAA;;;;GAIG;AAEH,OAAO,EAAE,eAAe,EAAE,MAAM,0BAA0B,CAAA;AAG1D,MAAM,CAAC,MAAM,qBAAqB,GAAG,YAAqB,CAAA;AAC1D,MAAM,CAAC,MAAM,kBAAkB,GAC7B,wKAAwK,CAAA;AAC1K,MAAM,CAAC,MAAM,8BAA8B,GACzC,4JAA4J,CAAA;AAE9J,4FAA4F;AAC5F,MAAM,CAAC,MAAM,mCAAmC,GAAG,IAAI,CAAA;AAEvD,MAAM,UAAU,QAAQ,CAAC,CAAS;IAChC,OAAO,CAAC,CAAC,OAAO,CAAC,SAAS,EAAE,EAAE,CAAC,CAAC,IAAI,EAAE,CAAA;AACxC,CAAC;AAED,mGAAmG;AACnG,MAAM,UAAU,sBAAsB,CAAC,OAAe;IACpD,MAAM,CAAC,GAAG,UAAU,CAAC,IAAI,CAAC,OAAO,CAAC,CAAA;IAClC,OAAO,CAAC,CAAC,CAAC,CAAC,QAAQ,CAAC,CAAC,CAAC,CAAC,CAAW,EAAE,EAAE,CAAC,CAAC,CAAC,CAAC,IAAI,CAAA;AAChD,CAAC;AAED,sGAAsG;AACtG,MAAM,UAAU,wBAAwB,CAAC,EAA6B;IACpE,MAAM,CAAC,GAAG,QAAQ,CAAC,EAAE,IAAI,EAAE,CAAC,CAAA;IAC5B,IAAI,CAAC,CAAC,IAAI,CAAC,KAAK,MAAM;QAAE,OAAO,KAAK,CAAA;IACpC,IAAI,CAAC,CAAC,CAAC,QAAQ,CAAC,GAAG,CAAC;QAAE,OAAO,KAAK,CAAA;IAClC,IAAI,CAAC,CAAC,UAAU,CAAC,SAAS,CAAC;QAAE,OAAO,KAAK,CAAA;IACzC,OAAO,IAAI,CAAA;AACb,CAAC;AAED,iEAAiE;AACjE,MAAM,UAAU,YAAY,CAAC,OAAkC;IAC7D,OAAO,QAAQ,CAAC,OAAO,IAAI,EAAE,CAAC;SAC3B,WAAW,EAAE;SACb,QAAQ,CAAC,OAAO,CAAC,CAAA;AACtB,CAAC;AAED,MAAM,UAAU,SAAS,CAAC,UAAuC;IAC/D,IAAI,CAAC,UAAU,IAAI,UAAU,CAAC,MAAM,KAAK,CAAC;QAAE,OAAO,KAAK,CAAA;IACxD,OAAO,UAAU,CAAC,IAAI,CAAC,CAAC,CAAC,EAAE,EAAE;QAC3B,MAAM,CAAC,GAAG,CAAC,CAAC,WAAW,EAAE,CAAA;QACzB,OAAO,CAAC,KAAK,KAAK,IAAI,CAAC,KAAK,MAAM,IAAI,CAAC,CAAC,UAAU,CAAC,QAAQ,CAAC,CAAA;IAC9D,CAAC,CAAC,CAAA;AACJ,CAAC;AAED,gGAAgG;AAChG,MAAM,UAAU,qBAAqB,CACnC,IAA+B,EAC/B,OAAkC,EAClC,UAAuC;IAEvC,MAAM,CAAC,GAAG,CAAC,IAAI,IAAI,EAAE,CAAC,CAAC,IAAI,EAAE,CAAC,WAAW,EAAE,CAAA;IAC3C,MAAM,KAAK,GAAG,CAAC,KAAK,QAAQ,IAAI,CAAC,KAAK,KAAK,IAAI,CAAC,KAAK,OAAO,CAAA;IAC5D,IAAI,CAAC,KAAK;QAAE,OAAO,KAAK,CAAA;IACxB,IAAI,CAAC,YAAY,CAAC,OAAO,CAAC;QAAE,OAAO,KAAK,CAAA;IACxC,IAAI,CAAC,UAAU,IAAI,UAAU,CAAC,MAAM,KAAK,CAAC;QAAE,OAAO,KAAK,CAAA;IACxD,OAAO,CAAC,SAAS,CAAC,UAAU,CAAC,CAAA;AAC/B,CAAC;AAED,MAAM,sBAAsB,GAAG,IAAI,GAAG,CAAC;IACrC,WAAW;IACX,aAAa;IACb,aAAa;IACb,aAAa;IACb,aAAa;IACb,QAAQ;CACT,CAAC,CAAA;AAEF,MAAM,UAAU,oBAAoB,CAAC,QAAgB;IACnD,OAAO,sBAAsB,CAAC,GAAG,CAAC,QAAQ,CAAC,CAAA;AAC7C,CAAC;AAID,MAAM,UAAU,SAAS,CAAC,QAAgB;IACxC,IAAI,QAAQ,CAAC,UAAU,CAAC,OAAO,CAAC;QAAE,OAAO,MAAM,CAAA;IAC/C,IAAI,QAAQ,KAAK,QAAQ;QAAE,OAAO,QAAQ,CAAA;IAC1C,OAAO,OAAO,CAAA;AAChB,CAAC;AAED,mFAAmF;AACnF,MAAM,UAAU,eAAe,CAAC,aAA4D;IAC1F,IAAI,CAAC,aAAa;QAAE,OAAO,KAAK,CAAA;IAChC,IAAI,aAAa,CAAC,oBAAoB,KAAK,CAAC;QAAE,OAAO,IAAI,CAAA;IACzD,MAAM,OAAO,GAAG,CAAC,aAAa,CAAC,cAAc,IAAI,EAAE,CAAC,CAAC,IAAI,EAAE,CAAA;IAC3D,IAAI,CAAC,OAAO;QAAE,OAAO,KAAK,CAAA;IAC1B,OAAO,OAAO,KAAK,KAAK,IAAI,OAAO,CAAC,UAAU,CAAC,MAAM,CAAC,CAAA;AACxD,CAAC;AAiBD;;;GAGG;AACH,MAAM,UAAU,uBAAuB,CAAC,KAOvC;IACC,MAAM,UAAU,GAAG,QAAQ,CAAC,KAAK,CAAC,iBAAiB,IAAI,EAAE,CAAC,CAAA;IAC1D,MAAM,SAAS,GAAG,QAAQ,CAAC,KAAK,CAAC,gBAAgB,IAAI,EAAE,CAAC,IAAI,UAAU,CAAA;IACtE,IAAI,CAAC,UAAU;QAAE,OAAO,EAAE,IAAI,EAAE,IAAI,EAAE,CAAA;IACtC,IAAI,SAAS,IAAI,SAAS,KAAK,UAAU;QAAE,OAAO,EAAE,IAAI,EAAE,iBAAiB,EAAE,UAAU,EAAE,SAAS,EAAE,CAAA;IACpG,MAAM,iBAAiB,GAAG,KAAK,CAAC,UAAU,CAAC,SAAS,CAAC,CAAA;IACrD,MAAM,gBAAgB,GAAG,KAAK,CAAC,kBAAkB,KAAK,CAAC,CAAA;IACvD,IAAI,oBAAoB,CAAC,iBAAiB,CAAC,IAAI,CAAC,gBAAgB,IAAI,eAAe,CAAC,KAAK,CAAC,aAAa,CAAC,EAAE,CAAC;QACzG,OAAO;YACL,IAAI,EAAE,aAAa;YACnB,UAAU,EAAE,SAAS;YACrB,aAAa,EAAE,CAAC,KAAK,CAAC,aAAa,EAAE,cAAc,IAAI,EAAE,CAAC,CAAC,IAAI,EAAE,IAAI,KAAK;YAC1E,SAAS,EAAE,KAAK,CAAC,aAAa,EAAE,oBAAoB,IAAI,IAAI;YAC5D,KAAK,EAAE,KAAK,CAAC,aAAa,EAAE,YAAY,IAAI,IAAI;YAChD,OAAO,EAAE,SAAS,CAAC,iBAAiB,CAAC;YACrC,kBAAkB,EAAE,KAAK,CAAC,aAAa,EAAE,oBAAoB,KAAK,IAAI;YACtE,eAAe,EAAE,KAAK,CAAC,aAAa,EAAE,iBAAiB,IAAI,IAAI;SAChE,CAAA;IACH,CAAC;IACD,MAAM,KAAK,GAAG,QAAQ,CAAC,KAAK,CAAC,YAAY,IAAI,EAAE,CAAC,CAAA;IAChD,IAAI,KAAK,EAAE,CAAC;QACV,MAAM,aAAa,GAAG,KAAK,CAAC,UAAU,CAAC,KAAK,CAAC,CAAA;QAC7C,IAAI,oBAAoB,CAAC,aAAa,CAAC,IAAI,aAAa,KAAK,iBAAiB,EAAE,CAAC;YAC/E,OAAO,EAAE,IAAI,EAAE,mBAAmB,EAAE,UAAU,EAAE,SAAS,EAAE,KAAK,EAAE,CAAA;QACpE,CAAC;IACH,CAAC;IACD,OAAO,EAAE,IAAI,EAAE,IAAI,EAAE,CAAA;AACvB,CAAC;AAED,qFAAqF;AACrF,MAAM,UAAU,cAAc,CAC5B,SAAiB,EACjB,SAAiC,EAAE;IAEnC,SAAS;SACN,KAAK,CAAC,GAAG,CAAC;SACV,MAAM,CAAC,CAAC,IAAI,EAAE,EAAE,CAAC,IAAI,CAAC,IAAI,EAAE,CAAC;SAC7B,OAAO,CAAC,CAAC,IAAI,EAAE,EAAE;QAChB,MAAM,CAAC,GAAG,EAAE,GAAG,UAAU,CAAC,GAAG,IAAI,CAAC,KAAK,CAAC,GAAG,CAAC,CAAA;QAC5C,MAAM,QAAQ,GAAG,GAAG,EAAE,IAAI,EAAE,CAAA;QAC5B,IAAI,QAAQ,IAAI,UAAU,CAAC,MAAM,GAAG,CAAC,IAAI,CAAC,QAAQ,CAAC,UAAU,CAAC,OAAO,CAAC,EAAE,CAAC;YACvE,MAAM,CAAC,QAAQ,CAAC,GAAG,UAAU,CAAC,IAAI,CAAC,GAAG,CAAC,CAAC,IAAI,EAAE,CAAA;QAChD,CAAC;IACH,CAAC,CAAC,CAAA;IACJ,OAAO,MAAM,CAAA;AACf,CAAC;AAED,yFAAyF;AACzF,MAAM,UAAU,yBAAyB,CAAC,qBAAsC;IAC9E,MAAM,UAAU,GAAG,MAAM,CAAC,qBAAqB,CAAC,CAAA;IAChD,MAAM,IAAI,GAAG,MAAM,CAAC,QAAQ,CAAC,UAAU,CAAC,IAAI,UAAU,GAAG,CAAC,CAAC,CAAC,CAAC,UAAU,CAAC,CAAC,CAAC,GAAG,CAAA;IAC7E,OAAO,IAAI,CAAC,GAAG,CAAC,IAAI,EAAE,mCAAmC,CAAC,CAAA;AAC5D,CAAC;AAED,8FAA8F;AAC9F,MAAM,UAAU,eAAe,CAAC,GAAY;IAC1C,IAAI,GAAG,YAAY,KAAK,EAAE,CAAC;QACzB,MAAM,CAAC,GAAG,GAAoD,CAAA;QAC9D,MAAM,KAAK,GAAG,CAAC,GAAG,CAAC,OAAO,CAAC,CAAA;QAC3B,IAAI,OAAO,CAAC,CAAC,OAAO,KAAK,QAAQ,IAAI,CAAC,CAAC,OAAO,CAAC,IAAI,EAAE;YAAE,KAAK,CAAC,IAAI,CAAC,CAAC,CAAC,OAAO,CAAC,IAAI,EAAE,CAAC,CAAA;QACnF,MAAM,IAAI,GAAG,OAAO,CAAC,CAAC,IAAI,KAAK,QAAQ,IAAI,CAAC,CAAC,IAAI,CAAC,CAAC,CAAC,KAAK,CAAC,CAAC,IAAI,GAAG,CAAC,CAAC,CAAC,EAAE,CAAA;QACvE,OAAO,GAAG,CAAC,OAAO,CAAC,CAAC,CAAC,GAAG,KAAK,CAAC,IAAI,CAAC,IAAI,CAAC,GAAG,IAAI,EAAE,CAAC,CAAC,CAAC,MAAM,CAAC,GAAG,CAAC,CAAA;IACjE,CAAC;IACD,IAAI,GAAG,IAAI,OAAO,GAAG,KAAK,QAAQ,EAAE,CAAC;QACnC,MAAM,CAAC,GAAG,GAA+D,CAAA;QACzE,MAAM,KAAK,GAAa,EAAE,CAAA;QAC1B,IAAI,OAAO,CAAC,CAAC,OAAO,KAAK,QAAQ,IAAI,CAAC,CAAC,OAAO,CAAC,IAAI,EAAE;YAAE,KAAK,CAAC,IAAI,CAAC,CAAC,CAAC,OAAO,CAAC,IAAI,EAAE,CAAC,CAAA;QACnF,IAAI,OAAO,CAAC,CAAC,OAAO,KAAK,QAAQ,IAAI,CAAC,CAAC,OAAO,CAAC,IAAI,EAAE;YAAE,KAAK,CAAC,IAAI,CAAC,CAAC,CAAC,OAAO,CAAC,IAAI,EAAE,CAAC,CAAA;QACnF,IAAI,KAAK,CAAC,MAAM,GAAG,CAAC,EAAE,CAAC;YACrB,MAAM,IAAI,GAAG,OAAO,CAAC,CAAC,IAAI,KAAK,QAAQ,IAAI,CAAC,CAAC,IAAI,CAAC,CAAC,CAAC,KAAK,CAAC,CAAC,IAAI,GAAG,CAAC,CAAC,CAAC,EAAE,CAAA;YACvE,OAAO,GAAG,KAAK,CAAC,IAAI,CAAC,IAAI,CAAC,GAAG,IAAI,EAAE,CAAA;QACrC,CAAC;QACD,IAAI,CAAC;YACH,MAAM,IAAI,GAAG,IAAI,CAAC,SAAS,CAAC,GAAG,CAAC,CAAA;YAChC,IAAI,IAAI,IAAI,IAAI,KAAK,IAAI,IAAI,IAAI,KAAK,MAAM;gBAAE,OAAO,IAAI,CAAA;QAC3D,CAAC;QAAC,MAAM,CAAC;YACP,kBAAkB;QACpB,CAAC;IACH,CAAC;IACD,OAAO,MAAM,CAAC,GAAG,CAAC,CAAA;AACpB,CAAC;AAED,sGAAsG;AACtG,MAAM,CAAC,MAAM,4BAA4B,GAAwB,IAAI,GAAG,CAAC;IACvE,sBAAsB;IACtB,oBAAoB;IACpB,yBAAyB;IACzB,kCAAkC;IAClC,kBAAkB;IAClB,0BAA0B;IAC1B,wBAAwB;IACxB,YAAY;CACb,CAAC,CAAA;AAEF,MAAM,UAAU,sBAAsB,CAAC,GAAY;IACjD,MAAM,IAAI,GAAI,GAA6C,EAAE,IAAI,CAAA;IACjE,OAAO,OAAO,IAAI,KAAK,QAAQ,IAAI,4BAA4B,CAAC,GAAG,CAAC,IAAI,CAAC,CAAA;AAC3E,CAAC;AAED,iEAAiE;AACjE,MAAM,UAAU,cAAc,CAAC,IAAe,EAAE,OAAe,EAAE,OAAgB;IAC/E,OAAO,IAAI,eAAe,CAAC,IAAI,EAAE,OAAO,EAAE,OAAO,CAAC,CAAA;AACpD,CAAC"}
|
|
1
|
+
{"version":3,"file":"policy.js","sourceRoot":"","sources":["../../../src/runtime/llamacpp/policy.ts"],"names":[],"mappings":"AAAA;;;;GAIG;AAEH,OAAO,EAAE,QAAQ,EAAE,OAAO,EAAE,MAAM,WAAW,CAAA;AAC7C,OAAO,EAAE,eAAe,EAAE,MAAM,0BAA0B,CAAA;AAG1D,MAAM,CAAC,MAAM,qBAAqB,GAAG,YAAqB,CAAA;AAC1D,MAAM,CAAC,MAAM,kBAAkB,GAC7B,wKAAwK,CAAA;AAC1K,MAAM,CAAC,MAAM,8BAA8B,GACzC,4JAA4J,CAAA;AAE9J,4FAA4F;AAC5F,MAAM,CAAC,MAAM,mCAAmC,GAAG,IAAI,CAAA;AAEvD,MAAM,UAAU,QAAQ,CAAC,CAAS;IAChC,OAAO,CAAC,CAAC,OAAO,CAAC,SAAS,EAAE,EAAE,CAAC,CAAC,IAAI,EAAE,CAAA;AACxC,CAAC;AAED,mGAAmG;AACnG,MAAM,UAAU,sBAAsB,CAAC,OAAe;IACpD,MAAM,CAAC,GAAG,UAAU,CAAC,IAAI,CAAC,OAAO,CAAC,CAAA;IAClC,OAAO,CAAC,CAAC,CAAC,CAAC,QAAQ,CAAC,CAAC,CAAC,CAAC,CAAW,EAAE,EAAE,CAAC,CAAC,CAAC,CAAC,IAAI,CAAA;AAChD,CAAC;AAED,sGAAsG;AACtG,MAAM,UAAU,wBAAwB,CAAC,EAA6B;IACpE,MAAM,CAAC,GAAG,QAAQ,CAAC,EAAE,IAAI,EAAE,CAAC,CAAA;IAC5B,IAAI,CAAC,CAAC,IAAI,CAAC,KAAK,MAAM;QAAE,OAAO,KAAK,CAAA;IACpC,IAAI,CAAC,CAAC,CAAC,QAAQ,CAAC,GAAG,CAAC;QAAE,OAAO,KAAK,CAAA;IAClC,IAAI,CAAC,CAAC,UAAU,CAAC,SAAS,CAAC;QAAE,OAAO,KAAK,CAAA;IACzC,OAAO,IAAI,CAAA;AACb,CAAC;AAED,iEAAiE;AACjE,MAAM,UAAU,YAAY,CAAC,OAAkC;IAC7D,OAAO,QAAQ,CAAC,OAAO,IAAI,EAAE,CAAC;SAC3B,WAAW,EAAE;SACb,QAAQ,CAAC,OAAO,CAAC,CAAA;AACtB,CAAC;AAED,MAAM,UAAU,SAAS,CAAC,UAAuC;IAC/D,IAAI,CAAC,UAAU,IAAI,UAAU,CAAC,MAAM,KAAK,CAAC;QAAE,OAAO,KAAK,CAAA;IACxD,OAAO,UAAU,CAAC,IAAI,CAAC,CAAC,CAAC,EAAE,EAAE;QAC3B,MAAM,CAAC,GAAG,CAAC,CAAC,WAAW,EAAE,CAAA;QACzB,OAAO,CAAC,KAAK,KAAK,IAAI,CAAC,KAAK,MAAM,IAAI,CAAC,CAAC,UAAU,CAAC,QAAQ,CAAC,CAAA;IAC9D,CAAC,CAAC,CAAA;AACJ,CAAC;AAED,gGAAgG;AAChG,MAAM,UAAU,qBAAqB,CACnC,IAA+B,EAC/B,OAAkC,EAClC,UAAuC;IAEvC,MAAM,CAAC,GAAG,CAAC,IAAI,IAAI,EAAE,CAAC,CAAC,IAAI,EAAE,CAAC,WAAW,EAAE,CAAA;IAC3C,MAAM,KAAK,GAAG,CAAC,KAAK,QAAQ,IAAI,CAAC,KAAK,KAAK,IAAI,CAAC,KAAK,OAAO,CAAA;IAC5D,IAAI,CAAC,KAAK;QAAE,OAAO,KAAK,CAAA;IACxB,IAAI,CAAC,YAAY,CAAC,OAAO,CAAC;QAAE,OAAO,KAAK,CAAA;IACxC,IAAI,CAAC,UAAU,IAAI,UAAU,CAAC,MAAM,KAAK,CAAC;QAAE,OAAO,KAAK,CAAA;IACxD,OAAO,CAAC,SAAS,CAAC,UAAU,CAAC,CAAA;AAC/B,CAAC;AAED,MAAM,sBAAsB,GAAG,IAAI,GAAG,CAAC;IACrC,WAAW;IACX,aAAa;IACb,aAAa;IACb,aAAa;IACb,aAAa;IACb,QAAQ;CACT,CAAC,CAAA;AAEF,MAAM,UAAU,oBAAoB,CAAC,QAAgB;IACnD,OAAO,sBAAsB,CAAC,GAAG,CAAC,QAAQ,CAAC,CAAA;AAC7C,CAAC;AAID,MAAM,UAAU,SAAS,CAAC,QAAgB;IACxC,IAAI,QAAQ,CAAC,UAAU,CAAC,OAAO,CAAC;QAAE,OAAO,MAAM,CAAA;IAC/C,IAAI,QAAQ,KAAK,QAAQ;QAAE,OAAO,QAAQ,CAAA;IAC1C,OAAO,OAAO,CAAA;AAChB,CAAC;AAED,mFAAmF;AACnF,MAAM,UAAU,eAAe,CAAC,aAA4D;IAC1F,IAAI,CAAC,aAAa;QAAE,OAAO,KAAK,CAAA;IAChC,IAAI,aAAa,CAAC,oBAAoB,KAAK,CAAC;QAAE,OAAO,IAAI,CAAA;IACzD,MAAM,OAAO,GAAG,CAAC,aAAa,CAAC,cAAc,IAAI,EAAE,CAAC,CAAC,IAAI,EAAE,CAAA;IAC3D,IAAI,CAAC,OAAO;QAAE,OAAO,KAAK,CAAA;IAC1B,OAAO,OAAO,KAAK,KAAK,IAAI,OAAO,CAAC,UAAU,CAAC,MAAM,CAAC,CAAA;AACxD,CAAC;AAiBD;;;GAGG;AACH,MAAM,UAAU,uBAAuB,CAAC,KAOvC;IACC,MAAM,UAAU,GAAG,QAAQ,CAAC,KAAK,CAAC,iBAAiB,IAAI,EAAE,CAAC,CAAA;IAC1D,MAAM,SAAS,GAAG,QAAQ,CAAC,KAAK,CAAC,gBAAgB,IAAI,EAAE,CAAC,IAAI,UAAU,CAAA;IACtE,IAAI,CAAC,UAAU;QAAE,OAAO,EAAE,IAAI,EAAE,IAAI,EAAE,CAAA;IACtC,IAAI,SAAS,IAAI,SAAS,KAAK,UAAU;QAAE,OAAO,EAAE,IAAI,EAAE,iBAAiB,EAAE,UAAU,EAAE,SAAS,EAAE,CAAA;IACpG,MAAM,iBAAiB,GAAG,KAAK,CAAC,UAAU,CAAC,SAAS,CAAC,CAAA;IACrD,MAAM,gBAAgB,GAAG,KAAK,CAAC,kBAAkB,KAAK,CAAC,CAAA;IACvD,IAAI,oBAAoB,CAAC,iBAAiB,CAAC,IAAI,CAAC,gBAAgB,IAAI,eAAe,CAAC,KAAK,CAAC,aAAa,CAAC,EAAE,CAAC;QACzG,OAAO;YACL,IAAI,EAAE,aAAa;YACnB,UAAU,EAAE,SAAS;YACrB,aAAa,EAAE,CAAC,KAAK,CAAC,aAAa,EAAE,cAAc,IAAI,EAAE,CAAC,CAAC,IAAI,EAAE,IAAI,KAAK;YAC1E,SAAS,EAAE,KAAK,CAAC,aAAa,EAAE,oBAAoB,IAAI,IAAI;YAC5D,KAAK,EAAE,KAAK,CAAC,aAAa,EAAE,YAAY,IAAI,IAAI;YAChD,OAAO,EAAE,SAAS,CAAC,iBAAiB,CAAC;YACrC,kBAAkB,EAAE,KAAK,CAAC,aAAa,EAAE,oBAAoB,KAAK,IAAI;YACtE,eAAe,EAAE,KAAK,CAAC,aAAa,EAAE,iBAAiB,IAAI,IAAI;SAChE,CAAA;IACH,CAAC;IACD,MAAM,KAAK,GAAG,QAAQ,CAAC,KAAK,CAAC,YAAY,IAAI,EAAE,CAAC,CAAA;IAChD,IAAI,KAAK,EAAE,CAAC;QACV,MAAM,aAAa,GAAG,KAAK,CAAC,UAAU,CAAC,KAAK,CAAC,CAAA;QAC7C,IAAI,oBAAoB,CAAC,aAAa,CAAC,IAAI,aAAa,KAAK,iBAAiB,EAAE,CAAC;YAC/E,OAAO,EAAE,IAAI,EAAE,mBAAmB,EAAE,UAAU,EAAE,SAAS,EAAE,KAAK,EAAE,CAAA;QACpE,CAAC;IACH,CAAC;IACD,OAAO,EAAE,IAAI,EAAE,IAAI,EAAE,CAAA;AACvB,CAAC;AAED,qFAAqF;AACrF,MAAM,UAAU,cAAc,CAC5B,SAAiB,EACjB,SAAiC,EAAE;IAEnC,SAAS;SACN,KAAK,CAAC,GAAG,CAAC;SACV,MAAM,CAAC,CAAC,IAAI,EAAE,EAAE,CAAC,IAAI,CAAC,IAAI,EAAE,CAAC;SAC7B,OAAO,CAAC,CAAC,IAAI,EAAE,EAAE;QAChB,MAAM,CAAC,GAAG,EAAE,GAAG,UAAU,CAAC,GAAG,IAAI,CAAC,KAAK,CAAC,GAAG,CAAC,CAAA;QAC5C,MAAM,QAAQ,GAAG,GAAG,EAAE,IAAI,EAAE,CAAA;QAC5B,IAAI,QAAQ,IAAI,UAAU,CAAC,MAAM,GAAG,CAAC,IAAI,CAAC,QAAQ,CAAC,UAAU,CAAC,OAAO,CAAC,EAAE,CAAC;YACvE,MAAM,CAAC,QAAQ,CAAC,GAAG,UAAU,CAAC,IAAI,CAAC,GAAG,CAAC,CAAC,IAAI,EAAE,CAAA;QAChD,CAAC;IACH,CAAC,CAAC,CAAA;IACJ,OAAO,MAAM,CAAA;AACf,CAAC;AAED,yFAAyF;AACzF,MAAM,UAAU,yBAAyB,CAAC,qBAAsC;IAC9E,MAAM,UAAU,GAAG,MAAM,CAAC,qBAAqB,CAAC,CAAA;IAChD,MAAM,IAAI,GAAG,MAAM,CAAC,QAAQ,CAAC,UAAU,CAAC,IAAI,UAAU,GAAG,CAAC,CAAC,CAAC,CAAC,UAAU,CAAC,CAAC,CAAC,GAAG,CAAA;IAC7E,OAAO,IAAI,CAAC,GAAG,CAAC,IAAI,EAAE,mCAAmC,CAAC,CAAA;AAC5D,CAAC;AAED,8FAA8F;AAC9F,MAAM,UAAU,eAAe,CAAC,GAAY;IAC1C,IAAI,GAAG,YAAY,KAAK,EAAE,CAAC;QACzB,MAAM,CAAC,GAAG,GAAoD,CAAA;QAC9D,MAAM,KAAK,GAAG,CAAC,GAAG,CAAC,OAAO,CAAC,CAAA;QAC3B,IAAI,OAAO,CAAC,CAAC,OAAO,KAAK,QAAQ,IAAI,CAAC,CAAC,OAAO,CAAC,IAAI,EAAE;YAAE,KAAK,CAAC,IAAI,CAAC,CAAC,CAAC,OAAO,CAAC,IAAI,EAAE,CAAC,CAAA;QACnF,MAAM,IAAI,GAAG,OAAO,CAAC,CAAC,IAAI,KAAK,QAAQ,IAAI,CAAC,CAAC,IAAI,CAAC,CAAC,CAAC,KAAK,CAAC,CAAC,IAAI,GAAG,CAAC,CAAC,CAAC,EAAE,CAAA;QACvE,OAAO,GAAG,CAAC,OAAO,CAAC,CAAC,CAAC,GAAG,KAAK,CAAC,IAAI,CAAC,IAAI,CAAC,GAAG,IAAI,EAAE,CAAC,CAAC,CAAC,MAAM,CAAC,GAAG,CAAC,CAAA;IACjE,CAAC;IACD,IAAI,GAAG,IAAI,OAAO,GAAG,KAAK,QAAQ,EAAE,CAAC;QACnC,MAAM,CAAC,GAAG,GAA+D,CAAA;QACzE,MAAM,KAAK,GAAa,EAAE,CAAA;QAC1B,IAAI,OAAO,CAAC,CAAC,OAAO,KAAK,QAAQ,IAAI,CAAC,CAAC,OAAO,CAAC,IAAI,EAAE;YAAE,KAAK,CAAC,IAAI,CAAC,CAAC,CAAC,OAAO,CAAC,IAAI,EAAE,CAAC,CAAA;QACnF,IAAI,OAAO,CAAC,CAAC,OAAO,KAAK,QAAQ,IAAI,CAAC,CAAC,OAAO,CAAC,IAAI,EAAE;YAAE,KAAK,CAAC,IAAI,CAAC,CAAC,CAAC,OAAO,CAAC,IAAI,EAAE,CAAC,CAAA;QACnF,IAAI,KAAK,CAAC,MAAM,GAAG,CAAC,EAAE,CAAC;YACrB,MAAM,IAAI,GAAG,OAAO,CAAC,CAAC,IAAI,KAAK,QAAQ,IAAI,CAAC,CAAC,IAAI,CAAC,CAAC,CAAC,KAAK,CAAC,CAAC,IAAI,GAAG,CAAC,CAAC,CAAC,EAAE,CAAA;YACvE,OAAO,GAAG,KAAK,CAAC,IAAI,CAAC,IAAI,CAAC,GAAG,IAAI,EAAE,CAAA;QACrC,CAAC;QACD,IAAI,CAAC;YACH,MAAM,IAAI,GAAG,IAAI,CAAC,SAAS,CAAC,GAAG,CAAC,CAAA;YAChC,IAAI,IAAI,IAAI,IAAI,KAAK,IAAI,IAAI,IAAI,KAAK,MAAM;gBAAE,OAAO,IAAI,CAAA;QAC3D,CAAC;QAAC,MAAM,CAAC;YACP,kBAAkB;QACpB,CAAC;IACH,CAAC;IACD,OAAO,MAAM,CAAC,GAAG,CAAC,CAAA;AACpB,CAAC;AAED,sGAAsG;AACtG,MAAM,CAAC,MAAM,4BAA4B,GAAwB,IAAI,GAAG,CAAC;IACvE,sBAAsB;IACtB,oBAAoB;IACpB,yBAAyB;IACzB,kCAAkC;IAClC,kBAAkB;IAClB,0BAA0B;IAC1B,wBAAwB;IACxB,YAAY;CACb,CAAC,CAAA;AAEF,MAAM,UAAU,sBAAsB,CAAC,GAAY;IACjD,MAAM,IAAI,GAAI,GAA6C,EAAE,IAAI,CAAA;IACjE,OAAO,OAAO,IAAI,KAAK,QAAQ,IAAI,4BAA4B,CAAC,GAAG,CAAC,IAAI,CAAC,CAAA;AAC3E,CAAC;AAED,iEAAiE;AACjE,MAAM,UAAU,cAAc,CAAC,IAAe,EAAE,OAAe,EAAE,OAAgB;IAC/E,OAAO,IAAI,eAAe,CAAC,IAAI,EAAE,OAAO,EAAE,OAAO,CAAC,CAAA;AACpD,CAAC;AAED;;;GAGG;AACH,MAAM,UAAU,YAAY,CAAC,OAAe;IAC1C,MAAM,GAAG,GAAG,OAAO,CAAC,OAAO,CAAC,CAAA;IAC5B,MAAM,KAAK,GAAG,OAAO,CAAC,GAAG,CAAC,CAAA;IAC1B,OAAO,QAAQ,CAAC,GAAG,CAAC,KAAK,KAAK,IAAI,QAAQ,CAAC,KAAK,CAAC,KAAK,OAAO,CAAC,CAAC,CAAC,OAAO,CAAC,KAAK,CAAC,CAAC,CAAC,CAAC,GAAG,CAAA;AACtF,CAAC;AAED;;;;GAIG;AACH,MAAM,UAAU,eAAe,CAC7B,QAA2E;IAE3E,OAAO,QAAQ;SACZ,MAAM,CACL,CAAC,MAAM,EAAE,EAAE,CAAC,MAAM,CAAC,KAAK,KAAK,OAAO,IAAI,MAAM,CAAC,KAAK,KAAK,UAAU,IAAI,MAAM,CAAC,KAAK,KAAK,YAAY,CACrG;SACA,OAAO,CAAC,CAAC,MAAM,EAAE,EAAE,CAAC,CAAC,MAAM,CAAC,MAAM,CAAC,CAAC,CAAC,CAAC,YAAY,CAAC,MAAM,CAAC,MAAM,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC,CAAA;AACnF,CAAC"}
|
|
@@ -1,13 +1,33 @@
|
|
|
1
1
|
/**
|
|
2
2
|
* Capability probe: does the installed `llama-server` advertise a speculative type in its `-h`
|
|
3
3
|
* output? Port of `check_spec_type_support` in the plugin's `commands.rs` (5 s budget, substring
|
|
4
|
-
* match over stdout+stderr).
|
|
4
|
+
* match over stdout+stderr). `runHelp` is the run itself, for a caller that judges the output and
|
|
5
|
+
* the exit on its own (the decision engine gate, which takes only a finished help screen as proof).
|
|
5
6
|
*/
|
|
6
7
|
import type { ManagedProcess, SpawnSpec } from '../shared/index.js';
|
|
8
|
+
import type { ExitInfo } from './errors.js';
|
|
7
9
|
export declare const DFLASH_SPEC_TYPE = "draft-dflash";
|
|
8
10
|
export declare const PROBE_TIMEOUT_MS = 5000;
|
|
11
|
+
/** How much of a probe's output a failure's details keep: the end, where an error is. */
|
|
12
|
+
export declare const PROBE_OUTPUT_TAIL_CHARS = 600;
|
|
9
13
|
export interface ProbeDeps {
|
|
10
14
|
spawn?: (spec: SpawnSpec) => ManagedProcess;
|
|
11
15
|
timeoutMs?: number;
|
|
16
|
+
now?: () => number;
|
|
12
17
|
}
|
|
18
|
+
/** What `llama-server -h` printed, how it exited and how long it took. */
|
|
19
|
+
export interface HelpRun {
|
|
20
|
+
/** stdout, then stderr. */
|
|
21
|
+
output: string;
|
|
22
|
+
exit: ExitInfo;
|
|
23
|
+
elapsedMs: number;
|
|
24
|
+
}
|
|
25
|
+
/** The last `chars` characters of `text` on one line, for a failure's details; `''` for no output. */
|
|
26
|
+
export declare function probeOutputTail(text: string, chars?: number): string;
|
|
27
|
+
/**
|
|
28
|
+
* Run `exePath -h` and collect its output. Rejects with `MODEL_LOAD_FAILED` when the process could
|
|
29
|
+
* not be started and with `MODEL_LOAD_TIMED_OUT` when it did not finish within the budget: neither
|
|
30
|
+
* says anything about what the build supports.
|
|
31
|
+
*/
|
|
32
|
+
export declare function runHelp(exePath: string, env: Record<string, string>, cwd: string | undefined, deps?: ProbeDeps): Promise<HelpRun>;
|
|
13
33
|
export declare function checkSpecTypeSupport(exePath: string, specType: string, env: Record<string, string>, cwd: string | undefined, deps?: ProbeDeps): Promise<boolean>;
|
|
@@ -1,15 +1,30 @@
|
|
|
1
1
|
/**
|
|
2
2
|
* Capability probe: does the installed `llama-server` advertise a speculative type in its `-h`
|
|
3
3
|
* output? Port of `check_spec_type_support` in the plugin's `commands.rs` (5 s budget, substring
|
|
4
|
-
* match over stdout+stderr).
|
|
4
|
+
* match over stdout+stderr). `runHelp` is the run itself, for a caller that judges the output and
|
|
5
|
+
* the exit on its own (the decision engine gate, which takes only a finished help screen as proof).
|
|
5
6
|
*/
|
|
6
7
|
import { AtomicCoreError } from '../../contracts/index.js';
|
|
7
8
|
import { spawnManaged } from '../shared/index.js';
|
|
8
9
|
export const DFLASH_SPEC_TYPE = 'draft-dflash';
|
|
9
10
|
export const PROBE_TIMEOUT_MS = 5000;
|
|
10
|
-
|
|
11
|
+
/** How much of a probe's output a failure's details keep: the end, where an error is. */
|
|
12
|
+
export const PROBE_OUTPUT_TAIL_CHARS = 600;
|
|
13
|
+
/** The last `chars` characters of `text` on one line, for a failure's details; `''` for no output. */
|
|
14
|
+
export function probeOutputTail(text, chars = PROBE_OUTPUT_TAIL_CHARS) {
|
|
15
|
+
const flat = text.replace(/\s+/g, ' ').trim();
|
|
16
|
+
return flat.length > chars ? `…${flat.slice(-chars)}` : flat;
|
|
17
|
+
}
|
|
18
|
+
/**
|
|
19
|
+
* Run `exePath -h` and collect its output. Rejects with `MODEL_LOAD_FAILED` when the process could
|
|
20
|
+
* not be started and with `MODEL_LOAD_TIMED_OUT` when it did not finish within the budget: neither
|
|
21
|
+
* says anything about what the build supports.
|
|
22
|
+
*/
|
|
23
|
+
export async function runHelp(exePath, env, cwd, deps = {}) {
|
|
11
24
|
const spawn = deps.spawn ?? spawnManaged;
|
|
12
25
|
const timeoutMs = deps.timeoutMs ?? PROBE_TIMEOUT_MS;
|
|
26
|
+
const now = deps.now ?? Date.now;
|
|
27
|
+
const started = now();
|
|
13
28
|
let proc;
|
|
14
29
|
try {
|
|
15
30
|
proc = spawn({ exe: exePath, args: ['-h'], env, cwd });
|
|
@@ -17,11 +32,19 @@ export async function checkSpecTypeSupport(exePath, specType, env, cwd, deps = {
|
|
|
17
32
|
catch (e) {
|
|
18
33
|
throw new AtomicCoreError('MODEL_LOAD_FAILED', 'Could not probe llama.cpp backend capabilities.', e.message);
|
|
19
34
|
}
|
|
20
|
-
|
|
21
|
-
const
|
|
35
|
+
let timer;
|
|
36
|
+
const timeout = new Promise((r) => {
|
|
37
|
+
timer = setTimeout(() => r('timeout'), timeoutMs);
|
|
38
|
+
timer.unref();
|
|
39
|
+
});
|
|
40
|
+
const outcome = await Promise.race([proc.exited, timeout]);
|
|
41
|
+
clearTimeout(timer);
|
|
22
42
|
if (outcome === 'timeout') {
|
|
43
|
+
const { stdout, stderr } = proc.output();
|
|
44
|
+
const tail = probeOutputTail(`${stdout}\n${stderr}`);
|
|
23
45
|
await proc.terminate(500);
|
|
24
|
-
throw new AtomicCoreError('MODEL_LOAD_TIMED_OUT', 'Timed out while probing llama.cpp backend capabilities.', `llama-server -h did not finish within ${Math.round(timeoutMs / 1000)}s for ${exePath}`
|
|
46
|
+
throw new AtomicCoreError('MODEL_LOAD_TIMED_OUT', 'Timed out while probing llama.cpp backend capabilities.', `llama-server -h did not finish within ${Math.round(timeoutMs / 1000)}s for ${exePath}` +
|
|
47
|
+
(tail ? `; output so far: ${tail}` : '; no output'));
|
|
25
48
|
}
|
|
26
49
|
const failure = proc.spawnFailure();
|
|
27
50
|
if (failure) {
|
|
@@ -30,6 +53,9 @@ export async function checkSpecTypeSupport(exePath, specType, env, cwd, deps = {
|
|
|
30
53
|
// let the line readers drain
|
|
31
54
|
await new Promise((r) => setTimeout(r, 10));
|
|
32
55
|
const { stdout, stderr } = proc.output();
|
|
33
|
-
return `${stdout}\n${stderr}
|
|
56
|
+
return { output: `${stdout}\n${stderr}`, exit: outcome, elapsedMs: now() - started };
|
|
57
|
+
}
|
|
58
|
+
export async function checkSpecTypeSupport(exePath, specType, env, cwd, deps = {}) {
|
|
59
|
+
return (await runHelp(exePath, env, cwd, deps)).output.includes(specType);
|
|
34
60
|
}
|
|
35
61
|
//# sourceMappingURL=probe.js.map
|
|
@@ -1 +1 @@
|
|
|
1
|
-
{"version":3,"file":"probe.js","sourceRoot":"","sources":["../../../src/runtime/llamacpp/probe.ts"],"names":[],"mappings":"AAAA
|
|
1
|
+
{"version":3,"file":"probe.js","sourceRoot":"","sources":["../../../src/runtime/llamacpp/probe.ts"],"names":[],"mappings":"AAAA;;;;;GAKG;AAEH,OAAO,EAAE,eAAe,EAAE,MAAM,0BAA0B,CAAA;AAC1D,OAAO,EAAE,YAAY,EAAE,MAAM,oBAAoB,CAAA;AAIjD,MAAM,CAAC,MAAM,gBAAgB,GAAG,cAAc,CAAA;AAC9C,MAAM,CAAC,MAAM,gBAAgB,GAAG,IAAI,CAAA;AACpC,yFAAyF;AACzF,MAAM,CAAC,MAAM,uBAAuB,GAAG,GAAG,CAAA;AAgB1C,sGAAsG;AACtG,MAAM,UAAU,eAAe,CAAC,IAAY,EAAE,KAAK,GAAG,uBAAuB;IAC3E,MAAM,IAAI,GAAG,IAAI,CAAC,OAAO,CAAC,MAAM,EAAE,GAAG,CAAC,CAAC,IAAI,EAAE,CAAA;IAC7C,OAAO,IAAI,CAAC,MAAM,GAAG,KAAK,CAAC,CAAC,CAAC,IAAI,IAAI,CAAC,KAAK,CAAC,CAAC,KAAK,CAAC,EAAE,CAAC,CAAC,CAAC,IAAI,CAAA;AAC9D,CAAC;AAED;;;;GAIG;AACH,MAAM,CAAC,KAAK,UAAU,OAAO,CAC3B,OAAe,EACf,GAA2B,EAC3B,GAAuB,EACvB,OAAkB,EAAE;IAEpB,MAAM,KAAK,GAAG,IAAI,CAAC,KAAK,IAAI,YAAY,CAAA;IACxC,MAAM,SAAS,GAAG,IAAI,CAAC,SAAS,IAAI,gBAAgB,CAAA;IACpD,MAAM,GAAG,GAAG,IAAI,CAAC,GAAG,IAAI,IAAI,CAAC,GAAG,CAAA;IAChC,MAAM,OAAO,GAAG,GAAG,EAAE,CAAA;IACrB,IAAI,IAAoB,CAAA;IACxB,IAAI,CAAC;QACH,IAAI,GAAG,KAAK,CAAC,EAAE,GAAG,EAAE,OAAO,EAAE,IAAI,EAAE,CAAC,IAAI,CAAC,EAAE,GAAG,EAAE,GAAG,EAAE,CAAC,CAAA;IACxD,CAAC;IAAC,OAAO,CAAC,EAAE,CAAC;QACX,MAAM,IAAI,eAAe,CACvB,mBAAmB,EACnB,iDAAiD,EAChD,CAAW,CAAC,OAAO,CACrB,CAAA;IACH,CAAC;IACD,IAAI,KAAiC,CAAA;IACrC,MAAM,OAAO,GAAG,IAAI,OAAO,CAAY,CAAC,CAAC,EAAE,EAAE;QAC3C,KAAK,GAAG,UAAU,CAAC,GAAG,EAAE,CAAC,CAAC,CAAC,SAAS,CAAC,EAAE,SAAS,CAAC,CAAA;QACjD,KAAK,CAAC,KAAK,EAAE,CAAA;IACf,CAAC,CAAC,CAAA;IACF,MAAM,OAAO,GAAG,MAAM,OAAO,CAAC,IAAI,CAAC,CAAC,IAAI,CAAC,MAAM,EAAE,OAAO,CAAC,CAAC,CAAA;IAC1D,YAAY,CAAC,KAAK,CAAC,CAAA;IACnB,IAAI,OAAO,KAAK,SAAS,EAAE,CAAC;QAC1B,MAAM,EAAE,MAAM,EAAE,MAAM,EAAE,GAAG,IAAI,CAAC,MAAM,EAAE,CAAA;QACxC,MAAM,IAAI,GAAG,eAAe,CAAC,GAAG,MAAM,KAAK,MAAM,EAAE,CAAC,CAAA;QACpD,MAAM,IAAI,CAAC,SAAS,CAAC,GAAG,CAAC,CAAA;QACzB,MAAM,IAAI,eAAe,CACvB,sBAAsB,EACtB,yDAAyD,EACzD,yCAAyC,IAAI,CAAC,KAAK,CAAC,SAAS,GAAG,IAAI,CAAC,SAAS,OAAO,EAAE;YACrF,CAAC,IAAI,CAAC,CAAC,CAAC,oBAAoB,IAAI,EAAE,CAAC,CAAC,CAAC,aAAa,CAAC,CACtD,CAAA;IACH,CAAC;IACD,MAAM,OAAO,GAAG,IAAI,CAAC,YAAY,EAAE,CAAA;IACnC,IAAI,OAAO,EAAE,CAAC;QACZ,MAAM,IAAI,eAAe,CACvB,mBAAmB,EACnB,iDAAiD,EACjD,OAAO,CAAC,OAAO,CAChB,CAAA;IACH,CAAC;IACD,6BAA6B;IAC7B,MAAM,IAAI,OAAO,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,UAAU,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,CAAA;IAC3C,MAAM,EAAE,MAAM,EAAE,MAAM,EAAE,GAAG,IAAI,CAAC,MAAM,EAAE,CAAA;IACxC,OAAO,EAAE,MAAM,EAAE,GAAG,MAAM,KAAK,MAAM,EAAE,EAAE,IAAI,EAAE,OAAO,EAAE,SAAS,EAAE,GAAG,EAAE,GAAG,OAAO,EAAE,CAAA;AACtF,CAAC;AAED,MAAM,CAAC,KAAK,UAAU,oBAAoB,CACxC,OAAe,EACf,QAAgB,EAChB,GAA2B,EAC3B,GAAuB,EACvB,OAAkB,EAAE;IAEpB,OAAO,CAAC,MAAM,OAAO,CAAC,OAAO,EAAE,GAAG,EAAE,GAAG,EAAE,IAAI,CAAC,CAAC,CAAC,MAAM,CAAC,QAAQ,CAAC,QAAQ,CAAC,CAAA;AAC3E,CAAC"}
|
|
@@ -19,7 +19,7 @@ import type { DataLayout } from '../../config/index.js';
|
|
|
19
19
|
import type { ProcessJournal } from '../../lock/index.js';
|
|
20
20
|
import type { ModelRegistry } from '../../models/index.js';
|
|
21
21
|
import { spawnAndAwaitReady } from '../shared/index.js';
|
|
22
|
-
import type { ManagedProcess, SpawnSpec, CtxIncreaseResult, LocalRuntime, RecreateResult } from '../shared/index.js';
|
|
22
|
+
import type { BackendOutputSink, ManagedProcess, SpawnSpec, CtxIncreaseResult, GpuClaimHook, GpuOccupancy, LocalRuntime, RecreateResult } from '../shared/index.js';
|
|
23
23
|
import type { LlamacppConfigInput } from './args.js';
|
|
24
24
|
import type { LlamacppEngineSettings, LoadPlanDeps } from './load-plan.js';
|
|
25
25
|
export type EmitFn = <K extends keyof CoreEvents>(name: K, payload: CoreEvents[K]) => void;
|
|
@@ -63,16 +63,38 @@ export interface LlamacppRuntimeOptions {
|
|
|
63
63
|
platform?: NodeJS.Platform;
|
|
64
64
|
baseEnv?: NodeJS.ProcessEnv;
|
|
65
65
|
fetch?: typeof fetch;
|
|
66
|
+
/**
|
|
67
|
+
* Every stdout/stderr line the backend prints, for the life of the session. A throwing sink is
|
|
68
|
+
* swallowed, with one `warn` through `log` per session.
|
|
69
|
+
*/
|
|
70
|
+
backendOutput?: BackendOutputSink;
|
|
71
|
+
/** The core's own logger: the one warning about a throwing `backendOutput` sink goes here. */
|
|
72
|
+
log?: (level: 'debug' | 'info' | 'warn' | 'error', message: string) => void;
|
|
66
73
|
/** Provider settings for a load; re-read per load so a settings change lands on the next one. */
|
|
67
74
|
readSettings: () => Promise<RuntimeSettings>;
|
|
68
75
|
/** Resolve `<version>/<backend>` to an installed executable (the backend service in a full core). */
|
|
69
76
|
ensureBackendReady?: LoadPlanDeps['ensureBackendReady'] | undefined;
|
|
70
77
|
resolveLatestBackend?: LoadPlanDeps['resolveLatestBackend'] | undefined;
|
|
71
78
|
cpuInfo?: LoadPlanDeps['cpuInfo'] | undefined;
|
|
79
|
+
unifiedMemory?: LoadPlanDeps['unifiedMemory'] | undefined;
|
|
72
80
|
ensureGemmaMtpDraft?: LoadPlanDeps['ensureGemmaMtpDraft'] | undefined;
|
|
73
81
|
ensureDflashDraft?: LoadPlanDeps['ensureDflashDraft'] | undefined;
|
|
74
82
|
/** Model that must never be auto-unloaded. */
|
|
75
83
|
transcriptionModelId?: string;
|
|
84
|
+
/**
|
|
85
|
+
* Core's GPU residency (spec `gpu-residency`): asked once a load knows its backend, before anything
|
|
86
|
+
* is spawned, to free the cards it will take from every other engine. Absent, nothing is asked.
|
|
87
|
+
*/
|
|
88
|
+
claimGpu?: GpuClaimHook | undefined;
|
|
89
|
+
/**
|
|
90
|
+
* Refuses a file this engine cannot run (a PrismML-only Bonsai file on stock llama.cpp), by
|
|
91
|
+
* throwing. Asked before the auto-unload, so a refused load never costs the user the chat model.
|
|
92
|
+
*/
|
|
93
|
+
checkCompatibility?: ((target: {
|
|
94
|
+
modelId: string;
|
|
95
|
+
modelPath: string;
|
|
96
|
+
sha256?: string;
|
|
97
|
+
}) => Promise<void>) | undefined;
|
|
76
98
|
/**
|
|
77
99
|
* Read a model's GGUF metadata. A seam because the model's trained context comes from here, and
|
|
78
100
|
* it is what decides when the context ladder has nowhere left to climb — untestable otherwise
|
|
@@ -88,6 +110,13 @@ export declare class LlamacppRuntime implements LocalRuntime {
|
|
|
88
110
|
private readonly options;
|
|
89
111
|
private readonly sessions;
|
|
90
112
|
private readonly loading;
|
|
113
|
+
/** Loads past their GPU claim and not yet a session: they hold their cards as `loading`. */
|
|
114
|
+
private readonly claimed;
|
|
115
|
+
/**
|
|
116
|
+
* Sessions being stopped: out of the session table, but their process has not exited yet, so they
|
|
117
|
+
* still hold their cards (`stopping`) and a second unload waits for the same exit.
|
|
118
|
+
*/
|
|
119
|
+
private readonly stopping;
|
|
91
120
|
private loadTail;
|
|
92
121
|
private readonly shutdownController;
|
|
93
122
|
private closing;
|
|
@@ -100,12 +129,37 @@ export declare class LlamacppRuntime implements LocalRuntime {
|
|
|
100
129
|
list(): SessionInfo[];
|
|
101
130
|
findSession(modelId: string): SessionInfo | undefined;
|
|
102
131
|
getLoadedModels(): string[];
|
|
132
|
+
/** The GPUs every session holds, and every load past its claim, for core's residency rule. */
|
|
133
|
+
gpuOccupancy(): GpuOccupancy[];
|
|
134
|
+
private footprint;
|
|
103
135
|
getRuntimeDeviceInfo(modelId: string): RuntimeDeviceInfo | undefined;
|
|
104
136
|
getMaxCtxTrain(modelId: string): number | undefined;
|
|
105
137
|
isLoading(modelId: string): boolean;
|
|
138
|
+
/**
|
|
139
|
+
* The pack directories the sessions run from, the ones still stopping included: what an engine
|
|
140
|
+
* removal must leave alone (change `unify-engine-lifecycle`, task 2.4). `<pack>/build/bin/<exe>`,
|
|
141
|
+
* or the legacy flat `<pack>/<exe>`.
|
|
142
|
+
*/
|
|
143
|
+
buildDirsInUse(): string[];
|
|
144
|
+
/**
|
|
145
|
+
* Run `fn` in the load queue's turn: it starts once the load in flight is done, and every load asked
|
|
146
|
+
* for meanwhile waits for it. An engine update or removal switches or deletes a build in here, so no
|
|
147
|
+
* load ever starts from a build half-way through either.
|
|
148
|
+
*/
|
|
149
|
+
exclusive<T>(fn: () => Promise<T>): Promise<T>;
|
|
106
150
|
/** Load a model, or join the load already in flight for it. */
|
|
107
151
|
load(modelId: string, opts?: LoadOptions): Promise<SessionInfo>;
|
|
108
152
|
private loadOnce;
|
|
153
|
+
/** The spawn, and the two post-failure retries of `performLoad` (drop mmproj, drop MTP), once each. */
|
|
154
|
+
private spawnWithRetries;
|
|
155
|
+
/**
|
|
156
|
+
* A decision model (a laya GGUF, a GGUF stamped with `decision.spec`, or a laya checkpoint folder)
|
|
157
|
+
* runs only in the decision module: as a chat or embedding server it answers nothing useful, or
|
|
158
|
+
* crashes. Refused before the auto-unload, so asking for one never costs the user the chat model.
|
|
159
|
+
* A model whose header cannot be read is left to the load plan, which says why.
|
|
160
|
+
*/
|
|
161
|
+
private assertNotDecisionModel;
|
|
162
|
+
private assertCompatible;
|
|
109
163
|
private enqueueLoad;
|
|
110
164
|
private planDeps;
|
|
111
165
|
/** Ports this runtime already handed out, so a second load never lands on the same one. */
|
|
@@ -142,6 +196,7 @@ export declare class LlamacppRuntime implements LocalRuntime {
|
|
|
142
196
|
*/
|
|
143
197
|
recreateSession(modelId: string): Promise<RecreateResult>;
|
|
144
198
|
private unloadSession;
|
|
199
|
+
private terminateSession;
|
|
145
200
|
unloadAll(): Promise<void>;
|
|
146
201
|
/** `llama-server --list-devices` against an installed backend. */
|
|
147
202
|
getDevices(exePath: string): Promise<DeviceInfo[]>;
|
|
@@ -14,20 +14,20 @@
|
|
|
14
14
|
* - a load the user cancels stops wherever it is — queued, planning, spawning or just ready — kills
|
|
15
15
|
* what it started at once, and never falls into a retry.
|
|
16
16
|
*/
|
|
17
|
-
import { dirname } from 'node:path';
|
|
17
|
+
import { dirname, win32 } from 'node:path';
|
|
18
18
|
import { AtomicCoreError } from '../../contracts/index.js';
|
|
19
19
|
import { DEFAULT_CTX_LEN, computeNextCtxLen } from './ctx-ladder.js';
|
|
20
20
|
import { processStartId } from '../../lock/index.js';
|
|
21
|
-
import { readGgufMetadataFromFile } from '../../models/index.js';
|
|
21
|
+
import { isDecisionCheckpointDir, isDecisionGguf, readGgufMetadataFromFile } from '../../models/index.js';
|
|
22
22
|
import { resolveLlama3TemplateOverride, STRICT_SYSTEM_GUARD_SIGNATURE } from '../../speculative/index.js';
|
|
23
23
|
import { checkDflashSupport } from '../../speculative/dflash-registry.js';
|
|
24
24
|
import { checkGemmaMtpSupport } from '../../speculative/gemma-mtp-registry.js';
|
|
25
|
-
import { buildProcessEnv, discoverCudaPaths, nodeCudaProbeEnv, textMentionsCudaRuntime, randomFreePort, spawnAndAwaitReady, spawnManaged, LLAMA_READY_MARKERS, closeLogStream, openLogStream, isLoadCancelled, raceLoadCancel, throwIfLoadCancelled, } from '../shared/index.js';
|
|
25
|
+
import { buildProcessEnv, discoverCudaPaths, nodeCudaProbeEnv, textMentionsCudaRuntime, randomFreePort, backendOutputReporter, spawnAndAwaitReady, spawnManaged, LLAMA_READY_MARKERS, closeLogStream, openLogStream, isLoadCancelled, raceLoadCancel, redactArgs, throwIfLoadCancelled, hostPid, } from '../shared/index.js';
|
|
26
26
|
import { planLlamaArgs } from './args.js';
|
|
27
27
|
import { parseDeviceOutput } from './devices.js';
|
|
28
28
|
import { classifyProcessOutput } from './errors.js';
|
|
29
|
-
import { autoUnloadTargets, nextRetry, planLlamaLoad } from './load-plan.js';
|
|
30
|
-
import { classifyBackendMismatch, formatLoadError, isConcreteVersionBackend } from './policy.js';
|
|
29
|
+
import { autoUnloadTargets, llamaGpuFootprint, nextRetry, planLlamaLoad } from './load-plan.js';
|
|
30
|
+
import { classifyBackendMismatch, formatLoadError, isConcreteVersionBackend, packDirOfExe } from './policy.js';
|
|
31
31
|
import { checkSpecTypeSupport, DFLASH_SPEC_TYPE } from './probe.js';
|
|
32
32
|
import { RuntimeDeviceAccumulator } from './runtime-device.js';
|
|
33
33
|
export const DEVICE_PROBE_TIMEOUT_MS = 10_000;
|
|
@@ -35,6 +35,13 @@ export class LlamacppRuntime {
|
|
|
35
35
|
options;
|
|
36
36
|
sessions = new Map();
|
|
37
37
|
loading = new Map();
|
|
38
|
+
/** Loads past their GPU claim and not yet a session: they hold their cards as `loading`. */
|
|
39
|
+
claimed = new Map();
|
|
40
|
+
/**
|
|
41
|
+
* Sessions being stopped: out of the session table, but their process has not exited yet, so they
|
|
42
|
+
* still hold their cards (`stopping`) and a second unload waits for the same exit.
|
|
43
|
+
*/
|
|
44
|
+
stopping = new Map();
|
|
38
45
|
loadTail = Promise.resolve();
|
|
39
46
|
shutdownController = new AbortController();
|
|
40
47
|
closing = false;
|
|
@@ -60,6 +67,22 @@ export class LlamacppRuntime {
|
|
|
60
67
|
getLoadedModels() {
|
|
61
68
|
return [...this.sessions.keys()];
|
|
62
69
|
}
|
|
70
|
+
/** The GPUs every session holds, and every load past its claim, for core's residency rule. */
|
|
71
|
+
gpuOccupancy() {
|
|
72
|
+
const ready = [...this.sessions.values()].map((session) => {
|
|
73
|
+
return { model_id: session.plan.modelId, ...this.footprint(session.plan), state: 'ready' };
|
|
74
|
+
});
|
|
75
|
+
const stopping = [...this.stopping.values()].map(({ session }) => ({
|
|
76
|
+
model_id: session.plan.modelId,
|
|
77
|
+
...this.footprint(session.plan),
|
|
78
|
+
state: 'stopping',
|
|
79
|
+
}));
|
|
80
|
+
const loading = [...this.claimed.values()].filter((claim) => !this.sessions.has(claim.model_id));
|
|
81
|
+
return [...ready, ...stopping, ...loading];
|
|
82
|
+
}
|
|
83
|
+
footprint(plan) {
|
|
84
|
+
return llamaGpuFootprint(plan, this.options.transcriptionModelId);
|
|
85
|
+
}
|
|
63
86
|
getRuntimeDeviceInfo(modelId) {
|
|
64
87
|
return this.sessions.get(modelId)?.devices.snapshot();
|
|
65
88
|
}
|
|
@@ -69,6 +92,23 @@ export class LlamacppRuntime {
|
|
|
69
92
|
isLoading(modelId) {
|
|
70
93
|
return this.loading.has(modelId);
|
|
71
94
|
}
|
|
95
|
+
/**
|
|
96
|
+
* The pack directories the sessions run from, the ones still stopping included: what an engine
|
|
97
|
+
* removal must leave alone (change `unify-engine-lifecycle`, task 2.4). `<pack>/build/bin/<exe>`,
|
|
98
|
+
* or the legacy flat `<pack>/<exe>`.
|
|
99
|
+
*/
|
|
100
|
+
buildDirsInUse() {
|
|
101
|
+
const sessions = [...this.sessions.values(), ...[...this.stopping.values()].map(({ session }) => session)];
|
|
102
|
+
return [...new Set(sessions.map((session) => packDirOfExe(session.plan.exePath)))];
|
|
103
|
+
}
|
|
104
|
+
/**
|
|
105
|
+
* Run `fn` in the load queue's turn: it starts once the load in flight is done, and every load asked
|
|
106
|
+
* for meanwhile waits for it. An engine update or removal switches or deletes a build in here, so no
|
|
107
|
+
* load ever starts from a build half-way through either.
|
|
108
|
+
*/
|
|
109
|
+
exclusive(fn) {
|
|
110
|
+
return this.enqueueLoad(fn);
|
|
111
|
+
}
|
|
72
112
|
/** Load a model, or join the load already in flight for it. */
|
|
73
113
|
async load(modelId, opts = {}) {
|
|
74
114
|
this.assertRunning();
|
|
@@ -87,6 +127,10 @@ export class LlamacppRuntime {
|
|
|
87
127
|
throwIfLoadCancelled(opts.signal);
|
|
88
128
|
const settings = await this.options.readSettings();
|
|
89
129
|
throwIfLoadCancelled(opts.signal);
|
|
130
|
+
await this.assertNotDecisionModel(modelId, opts);
|
|
131
|
+
throwIfLoadCancelled(opts.signal);
|
|
132
|
+
await this.assertCompatible(modelId, opts);
|
|
133
|
+
throwIfLoadCancelled(opts.signal);
|
|
90
134
|
const config = { ...settings.config };
|
|
91
135
|
if (opts.exePath)
|
|
92
136
|
config.version_backend = opts.versionBackend ?? 'cli/llama-server';
|
|
@@ -139,6 +183,34 @@ export class LlamacppRuntime {
|
|
|
139
183
|
}, this.planDeps(opts)), opts.signal);
|
|
140
184
|
if (opts.port !== undefined)
|
|
141
185
|
plan = { ...plan, port: opts.port };
|
|
186
|
+
// GPU residency: the backend is known now, and nothing is started yet. Every other engine on the
|
|
187
|
+
// cards this build takes is stopped, with its exit confirmed, before the spawn below.
|
|
188
|
+
// It holds its cards as `loading` from the moment core grants the claim, inside core's turn.
|
|
189
|
+
const footprint = this.footprint(plan);
|
|
190
|
+
const occupancy = { model_id: modelId, ...footprint, state: 'loading' };
|
|
191
|
+
const hold = () => {
|
|
192
|
+
this.claimed.set(modelId, occupancy);
|
|
193
|
+
};
|
|
194
|
+
try {
|
|
195
|
+
if (this.options.claimGpu) {
|
|
196
|
+
await this.options.claimGpu({ model_id: modelId, ...footprint }, opts.signal, hold);
|
|
197
|
+
hold();
|
|
198
|
+
this.assertRunning();
|
|
199
|
+
throwIfLoadCancelled(opts.signal);
|
|
200
|
+
}
|
|
201
|
+
else
|
|
202
|
+
hold();
|
|
203
|
+
return await this.spawnWithRetries(plan, opts);
|
|
204
|
+
}
|
|
205
|
+
finally {
|
|
206
|
+
if (this.claimed.get(modelId) === occupancy)
|
|
207
|
+
this.claimed.delete(modelId);
|
|
208
|
+
}
|
|
209
|
+
}
|
|
210
|
+
/** The spawn, and the two post-failure retries of `performLoad` (drop mmproj, drop MTP), once each. */
|
|
211
|
+
async spawnWithRetries(initial, opts) {
|
|
212
|
+
let plan = initial;
|
|
213
|
+
const { modelId } = plan;
|
|
142
214
|
for (let attempt = 0; attempt < 3; attempt++) {
|
|
143
215
|
this.assertRunning();
|
|
144
216
|
throwIfLoadCancelled(opts.signal);
|
|
@@ -164,6 +236,38 @@ export class LlamacppRuntime {
|
|
|
164
236
|
}
|
|
165
237
|
throw new AtomicCoreError('MODEL_LOAD_FAILED', `Could not load model "${modelId}" after fallback attempts.`);
|
|
166
238
|
}
|
|
239
|
+
/**
|
|
240
|
+
* A decision model (a laya GGUF, a GGUF stamped with `decision.spec`, or a laya checkpoint folder)
|
|
241
|
+
* runs only in the decision module: as a chat or embedding server it answers nothing useful, or
|
|
242
|
+
* crashes. Refused before the auto-unload, so asking for one never costs the user the chat model.
|
|
243
|
+
* A model whose header cannot be read is left to the load plan, which says why.
|
|
244
|
+
*/
|
|
245
|
+
async assertNotDecisionModel(modelId, opts) {
|
|
246
|
+
const { registry } = this.options;
|
|
247
|
+
const configured = opts.modelPath ?? (await registry.read(modelId).catch(() => undefined))?.model_path;
|
|
248
|
+
if (!configured)
|
|
249
|
+
return;
|
|
250
|
+
const path = registry.resolvePaths({ model_path: configured }).modelPath;
|
|
251
|
+
const read = this.options.readGgufMetadata ?? (async (p) => (await readGgufMetadataFromFile(p)).metadata);
|
|
252
|
+
const decision = (await isDecisionCheckpointDir(path)) || isDecisionGguf(await read(path).catch(() => undefined));
|
|
253
|
+
if (decision)
|
|
254
|
+
throw new AtomicCoreError('DECISION_MODEL_NOT_CHAT', `"${modelId}" is a decision model and cannot be loaded for chat. Decision models run in the decision module (settings: decision.model_path).`, path);
|
|
255
|
+
}
|
|
256
|
+
async assertCompatible(modelId, opts) {
|
|
257
|
+
const { registry, checkCompatibility } = this.options;
|
|
258
|
+
if (!checkCompatibility)
|
|
259
|
+
return;
|
|
260
|
+
const yml = opts.modelPath === undefined ? await registry.read(modelId).catch(() => undefined) : undefined;
|
|
261
|
+
const configured = opts.modelPath ?? yml?.model_path;
|
|
262
|
+
if (!configured)
|
|
263
|
+
return;
|
|
264
|
+
const modelPath = registry.resolvePaths({ model_path: configured }).modelPath;
|
|
265
|
+
await checkCompatibility({
|
|
266
|
+
modelId,
|
|
267
|
+
modelPath,
|
|
268
|
+
...(yml?.model_sha256 ? { sha256: yml.model_sha256 } : {}),
|
|
269
|
+
});
|
|
270
|
+
}
|
|
167
271
|
enqueueLoad(operation, signal) {
|
|
168
272
|
const turn = this.loadTail;
|
|
169
273
|
const result = raceLoadCancel(turn, signal).then(operation);
|
|
@@ -190,6 +294,8 @@ export class LlamacppRuntime {
|
|
|
190
294
|
throw new AtomicCoreError('BINARY_NOT_FOUND', 'No llama.cpp backend is available for this data folder.', `${version}/${backend} in ${layout.provider(this.provider).backendsDir}`);
|
|
191
295
|
},
|
|
192
296
|
cpuInfo: this.options.cpuInfo ?? (async () => undefined),
|
|
297
|
+
...(this.options.unifiedMemory ? { unifiedMemory: this.options.unifiedMemory } : {}),
|
|
298
|
+
listDevices: (exePath) => this.getDevices(exePath),
|
|
193
299
|
exists: async (path) => {
|
|
194
300
|
const { stat } = await import('node:fs/promises');
|
|
195
301
|
return stat(path).then(() => true, () => false);
|
|
@@ -235,6 +341,8 @@ export class LlamacppRuntime {
|
|
|
235
341
|
const apiKey = plan.apiKey;
|
|
236
342
|
throwIfLoadCancelled(opts.signal);
|
|
237
343
|
const logStream = opts.logPath ? await openLogStream(opts.logPath) : undefined;
|
|
344
|
+
const reportOutput = backendOutputReporter(this.options.backendOutput, this.options.log);
|
|
345
|
+
this.options.log?.('info', `starting ${win32.basename(plan.exePath)} for ${plan.provider}/${plan.modelId}: ${redactArgs(args.argv).join(' ')}`);
|
|
238
346
|
let proc;
|
|
239
347
|
try {
|
|
240
348
|
;
|
|
@@ -255,6 +363,7 @@ export class LlamacppRuntime {
|
|
|
255
363
|
level: 'debug',
|
|
256
364
|
msg: `[${plan.provider}/${plan.modelId}][${stream}] ${line}`,
|
|
257
365
|
});
|
|
366
|
+
reportOutput({ provider: plan.provider, model: plan.modelId, stream, line });
|
|
258
367
|
},
|
|
259
368
|
classifyExit: (exit, stderr, stdout) => classifyProcessOutput(exit, stderr, stdout, this.platform, plan.provider),
|
|
260
369
|
timeoutMessage: `Timeout: ${plan.timeoutSecs}s`,
|
|
@@ -294,7 +403,7 @@ export class LlamacppRuntime {
|
|
|
294
403
|
catch (error) {
|
|
295
404
|
await proc.terminate(isLoadCancelled(opts.signal) ? 0 : undefined).catch(() => { });
|
|
296
405
|
if (session.journalled)
|
|
297
|
-
await this.options.journal?.remove(session.info
|
|
406
|
+
await this.options.journal?.remove(hostPid(session.info)).catch(() => { });
|
|
298
407
|
await closeLogStream(session.logStream);
|
|
299
408
|
throw error;
|
|
300
409
|
}
|
|
@@ -308,8 +417,8 @@ export class LlamacppRuntime {
|
|
|
308
417
|
return;
|
|
309
418
|
const record = {
|
|
310
419
|
instance_id: this.options.instanceId,
|
|
311
|
-
pid: session.info
|
|
312
|
-
process_start_id: (await processStartId(session.info
|
|
420
|
+
pid: hostPid(session.info),
|
|
421
|
+
process_start_id: (await processStartId(hostPid(session.info))) ?? null,
|
|
313
422
|
exe: session.plan.exePath,
|
|
314
423
|
provider: session.plan.provider,
|
|
315
424
|
model_id: session.plan.modelId,
|
|
@@ -326,13 +435,13 @@ export class LlamacppRuntime {
|
|
|
326
435
|
return; // already replaced or unloaded
|
|
327
436
|
this.sessions.delete(session.plan.modelId);
|
|
328
437
|
if (session.journalled)
|
|
329
|
-
await this.options.journal?.remove(session.info
|
|
438
|
+
await this.options.journal?.remove(hostPid(session.info)).catch(() => { });
|
|
330
439
|
await closeLogStream(session.logStream);
|
|
331
440
|
const { stderr, stdout } = session.process.output();
|
|
332
441
|
const error = classifyProcessOutput(exit, stderr, stdout, this.platform, session.plan.provider);
|
|
333
442
|
this.emit('session:died', {
|
|
334
443
|
provider: session.plan.provider,
|
|
335
|
-
pid: session.info
|
|
444
|
+
pid: hostPid(session.info),
|
|
336
445
|
model_id: session.plan.modelId,
|
|
337
446
|
exit_code: exit.code,
|
|
338
447
|
signal: exit.signal === null ? null : String(exit.signal),
|
|
@@ -451,20 +560,38 @@ export class LlamacppRuntime {
|
|
|
451
560
|
});
|
|
452
561
|
return { ok: true, session: info };
|
|
453
562
|
}
|
|
454
|
-
|
|
563
|
+
unloadSession(modelId) {
|
|
564
|
+
// A stop already in flight is joined: its answer comes only once the process has exited.
|
|
565
|
+
const inFlight = this.stopping.get(modelId);
|
|
566
|
+
if (inFlight)
|
|
567
|
+
return inFlight.done;
|
|
455
568
|
const session = this.sessions.get(modelId);
|
|
456
569
|
if (!session)
|
|
457
|
-
return { success: true };
|
|
570
|
+
return Promise.resolve({ success: true });
|
|
458
571
|
this.sessions.delete(modelId);
|
|
572
|
+
const done = this.terminateSession(modelId, session);
|
|
573
|
+
const entry = { session, done };
|
|
574
|
+
this.stopping.set(modelId, entry);
|
|
575
|
+
// `.catch`: `done` goes back to the caller, who handles its rejection; this derived promise would
|
|
576
|
+
// otherwise be a second, unhandled one (final review M-8).
|
|
577
|
+
void done
|
|
578
|
+
.finally(() => {
|
|
579
|
+
if (this.stopping.get(modelId) === entry)
|
|
580
|
+
this.stopping.delete(modelId);
|
|
581
|
+
})
|
|
582
|
+
.catch(() => { });
|
|
583
|
+
return done;
|
|
584
|
+
}
|
|
585
|
+
async terminateSession(modelId, session) {
|
|
459
586
|
try {
|
|
460
587
|
await session.process.terminate();
|
|
461
588
|
if (session.journalled)
|
|
462
|
-
await this.options.journal?.remove(session.info
|
|
589
|
+
await this.options.journal?.remove(hostPid(session.info)).catch(() => { });
|
|
463
590
|
await closeLogStream(session.logStream);
|
|
464
591
|
this.emit('session:unloaded', {
|
|
465
592
|
provider: session.plan.provider,
|
|
466
593
|
model_id: modelId,
|
|
467
|
-
pid: session.info
|
|
594
|
+
pid: hostPid(session.info),
|
|
468
595
|
});
|
|
469
596
|
return { success: true };
|
|
470
597
|
}
|
|
@@ -472,7 +599,7 @@ export class LlamacppRuntime {
|
|
|
472
599
|
if (session.process.child.exitCode === null && session.process.child.signalCode === null)
|
|
473
600
|
this.sessions.set(modelId, session);
|
|
474
601
|
else if (session.journalled)
|
|
475
|
-
await this.options.journal?.remove(session.info
|
|
602
|
+
await this.options.journal?.remove(hostPid(session.info)).catch(() => { });
|
|
476
603
|
return { success: false, error: e.message };
|
|
477
604
|
}
|
|
478
605
|
}
|