node-llama-cpp 3.0.0-beta.2 → 3.0.0-beta.20
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 +14 -11
- package/dist/ChatWrapper.d.ts +2 -15
- package/dist/ChatWrapper.js +28 -33
- package/dist/ChatWrapper.js.map +1 -1
- package/dist/apiDocsOverrides.d.ts +1 -0
- package/dist/apiDocsOverrides.js +5 -0
- package/dist/apiDocsOverrides.js.map +1 -0
- package/dist/{utils/getBin.d.ts → bindings/AddonTypes.d.ts} +54 -7
- package/dist/bindings/AddonTypes.js +2 -0
- package/dist/bindings/AddonTypes.js.map +1 -0
- package/dist/bindings/Llama.d.ts +47 -0
- package/dist/bindings/Llama.js +343 -0
- package/dist/bindings/Llama.js.map +1 -0
- package/dist/bindings/consts.d.ts +2 -0
- package/dist/bindings/consts.js +11 -0
- package/dist/bindings/consts.js.map +1 -0
- package/dist/bindings/getLlama.d.ts +145 -0
- package/dist/bindings/getLlama.js +389 -0
- package/dist/bindings/getLlama.js.map +1 -0
- package/dist/bindings/types.d.ts +55 -0
- package/dist/bindings/types.js +77 -0
- package/dist/bindings/types.js.map +1 -0
- package/dist/bindings/utils/MemoryOrchestrator.d.ts +21 -0
- package/dist/bindings/utils/MemoryOrchestrator.js +49 -0
- package/dist/bindings/utils/MemoryOrchestrator.js.map +1 -0
- package/dist/bindings/utils/NoBinaryFoundError.d.ts +2 -0
- package/dist/bindings/utils/NoBinaryFoundError.js +7 -0
- package/dist/bindings/utils/NoBinaryFoundError.js.map +1 -0
- package/dist/bindings/utils/asyncEvery.d.ts +5 -0
- package/dist/bindings/utils/asyncEvery.js +15 -0
- package/dist/bindings/utils/asyncEvery.js.map +1 -0
- package/dist/bindings/utils/asyncSome.d.ts +5 -0
- package/dist/bindings/utils/asyncSome.js +27 -0
- package/dist/bindings/utils/asyncSome.js.map +1 -0
- package/dist/{utils → bindings/utils}/binariesGithubRelease.js +1 -1
- package/dist/bindings/utils/binariesGithubRelease.js.map +1 -0
- package/dist/bindings/utils/clearAllLocalBuilds.d.ts +1 -0
- package/dist/bindings/utils/clearAllLocalBuilds.js +47 -0
- package/dist/bindings/utils/clearAllLocalBuilds.js.map +1 -0
- package/dist/bindings/utils/cloneLlamaCppRepo.d.ts +11 -0
- package/dist/bindings/utils/cloneLlamaCppRepo.js +166 -0
- package/dist/bindings/utils/cloneLlamaCppRepo.js.map +1 -0
- package/dist/bindings/utils/compileLLamaCpp.d.ts +15 -0
- package/dist/bindings/utils/compileLLamaCpp.js +221 -0
- package/dist/bindings/utils/compileLLamaCpp.js.map +1 -0
- package/dist/bindings/utils/detectAvailableComputeLayers.d.ts +14 -0
- package/dist/bindings/utils/detectAvailableComputeLayers.js +304 -0
- package/dist/bindings/utils/detectAvailableComputeLayers.js.map +1 -0
- package/dist/bindings/utils/detectGlibc.d.ts +4 -0
- package/dist/bindings/utils/detectGlibc.js +46 -0
- package/dist/bindings/utils/detectGlibc.js.map +1 -0
- package/dist/bindings/utils/getBestComputeLayersAvailable.d.ts +9 -0
- package/dist/bindings/utils/getBestComputeLayersAvailable.js +29 -0
- package/dist/bindings/utils/getBestComputeLayersAvailable.js.map +1 -0
- package/dist/bindings/utils/getBuildFolderNameForBuildOptions.d.ts +5 -0
- package/dist/bindings/utils/getBuildFolderNameForBuildOptions.js +93 -0
- package/dist/bindings/utils/getBuildFolderNameForBuildOptions.js.map +1 -0
- package/dist/bindings/utils/getCanUsePrebuiltBinaries.d.ts +1 -0
- package/dist/bindings/utils/getCanUsePrebuiltBinaries.js +8 -0
- package/dist/bindings/utils/getCanUsePrebuiltBinaries.js.map +1 -0
- package/dist/bindings/utils/getExampleUsageCodeOfGetLlama.d.ts +2 -0
- package/dist/bindings/utils/getExampleUsageCodeOfGetLlama.js +21 -0
- package/dist/bindings/utils/getExampleUsageCodeOfGetLlama.js.map +1 -0
- package/dist/bindings/utils/getGpuTypesToUseForOption.d.ts +11 -0
- package/dist/bindings/utils/getGpuTypesToUseForOption.js +30 -0
- package/dist/bindings/utils/getGpuTypesToUseForOption.js.map +1 -0
- package/dist/bindings/utils/getLinuxDistroInfo.d.ts +9 -0
- package/dist/bindings/utils/getLinuxDistroInfo.js +46 -0
- package/dist/bindings/utils/getLinuxDistroInfo.js.map +1 -0
- package/dist/bindings/utils/getLlamaWithoutBackend.d.ts +5 -0
- package/dist/bindings/utils/getLlamaWithoutBackend.js +27 -0
- package/dist/bindings/utils/getLlamaWithoutBackend.js.map +1 -0
- package/dist/bindings/utils/getPlatform.d.ts +2 -0
- package/dist/bindings/utils/getPlatform.js +15 -0
- package/dist/bindings/utils/getPlatform.js.map +1 -0
- package/dist/bindings/utils/getPlatformInfo.d.ts +5 -0
- package/dist/bindings/utils/getPlatformInfo.js +28 -0
- package/dist/bindings/utils/getPlatformInfo.js.map +1 -0
- package/dist/bindings/utils/hasBuildingFromSourceDependenciesInstalled.d.ts +3 -0
- package/dist/bindings/utils/hasBuildingFromSourceDependenciesInstalled.js +27 -0
- package/dist/bindings/utils/hasBuildingFromSourceDependenciesInstalled.js.map +1 -0
- package/dist/bindings/utils/hasFileInPath.d.ts +2 -0
- package/dist/bindings/utils/hasFileInPath.js +34 -0
- package/dist/bindings/utils/hasFileInPath.js.map +1 -0
- package/dist/bindings/utils/lastBuildInfo.d.ts +6 -0
- package/dist/bindings/utils/lastBuildInfo.js +17 -0
- package/dist/bindings/utils/lastBuildInfo.js.map +1 -0
- package/dist/bindings/utils/logBinaryUsageExampleToConsole.d.ts +2 -0
- package/dist/bindings/utils/logBinaryUsageExampleToConsole.js +22 -0
- package/dist/bindings/utils/logBinaryUsageExampleToConsole.js.map +1 -0
- package/dist/bindings/utils/logDistroInstallInstruction.d.ts +14 -0
- package/dist/bindings/utils/logDistroInstallInstruction.js +48 -0
- package/dist/bindings/utils/logDistroInstallInstruction.js.map +1 -0
- package/dist/bindings/utils/resolveCustomCmakeOptions.d.ts +1 -0
- package/dist/bindings/utils/resolveCustomCmakeOptions.js +45 -0
- package/dist/bindings/utils/resolveCustomCmakeOptions.js.map +1 -0
- package/dist/bindings/utils/testBindingBinary.d.ts +1 -0
- package/dist/bindings/utils/testBindingBinary.js +98 -0
- package/dist/bindings/utils/testBindingBinary.js.map +1 -0
- package/dist/bindings/utils/testCmakeBinary.d.ts +5 -0
- package/dist/bindings/utils/testCmakeBinary.js +32 -0
- package/dist/bindings/utils/testCmakeBinary.js.map +1 -0
- package/dist/chatWrappers/AlpacaChatWrapper.d.ts +2 -1
- package/dist/chatWrappers/AlpacaChatWrapper.js +9 -2
- package/dist/chatWrappers/AlpacaChatWrapper.js.map +1 -1
- package/dist/chatWrappers/ChatMLChatWrapper.d.ts +5 -0
- package/dist/chatWrappers/ChatMLChatWrapper.js +13 -11
- package/dist/chatWrappers/ChatMLChatWrapper.js.map +1 -1
- package/dist/chatWrappers/FalconChatWrapper.d.ts +2 -1
- package/dist/chatWrappers/FalconChatWrapper.js +28 -11
- package/dist/chatWrappers/FalconChatWrapper.js.map +1 -1
- package/dist/chatWrappers/FunctionaryChatWrapper.js +86 -73
- package/dist/chatWrappers/FunctionaryChatWrapper.js.map +1 -1
- package/dist/chatWrappers/{LlamaChatWrapper.d.ts → GemmaChatWrapper.d.ts} +6 -1
- package/dist/chatWrappers/GemmaChatWrapper.js +88 -0
- package/dist/chatWrappers/GemmaChatWrapper.js.map +1 -0
- package/dist/chatWrappers/GeneralChatWrapper.d.ts +2 -1
- package/dist/chatWrappers/GeneralChatWrapper.js +35 -12
- package/dist/chatWrappers/GeneralChatWrapper.js.map +1 -1
- package/dist/chatWrappers/Llama2ChatWrapper.d.ts +20 -0
- package/dist/chatWrappers/{LlamaChatWrapper.js → Llama2ChatWrapper.js} +29 -11
- package/dist/chatWrappers/Llama2ChatWrapper.js.map +1 -0
- package/dist/chatWrappers/Llama3ChatWrapper.d.ts +31 -0
- package/dist/chatWrappers/Llama3ChatWrapper.js +129 -0
- package/dist/chatWrappers/Llama3ChatWrapper.js.map +1 -0
- package/dist/chatWrappers/generic/JinjaTemplateChatWrapper.d.ts +73 -0
- package/dist/chatWrappers/generic/JinjaTemplateChatWrapper.js +359 -0
- package/dist/chatWrappers/generic/JinjaTemplateChatWrapper.js.map +1 -0
- package/dist/chatWrappers/generic/TemplateChatWrapper.d.ts +64 -0
- package/dist/chatWrappers/generic/TemplateChatWrapper.js +200 -0
- package/dist/chatWrappers/generic/TemplateChatWrapper.js.map +1 -0
- package/dist/chatWrappers/generic/utils/chatHistoryFunctionCallMessageTemplate.d.ts +33 -0
- package/dist/chatWrappers/generic/utils/chatHistoryFunctionCallMessageTemplate.js +45 -0
- package/dist/chatWrappers/generic/utils/chatHistoryFunctionCallMessageTemplate.js.map +1 -0
- package/dist/chatWrappers/utils/ChatModelFunctionsDocumentationGenerator.d.ts +42 -0
- package/dist/chatWrappers/utils/ChatModelFunctionsDocumentationGenerator.js +82 -0
- package/dist/chatWrappers/utils/ChatModelFunctionsDocumentationGenerator.js.map +1 -0
- package/dist/chatWrappers/utils/isJinjaTemplateEquivalentToSpecializedChatWrapper.d.ts +4 -0
- package/dist/chatWrappers/utils/isJinjaTemplateEquivalentToSpecializedChatWrapper.js +206 -0
- package/dist/chatWrappers/utils/isJinjaTemplateEquivalentToSpecializedChatWrapper.js.map +1 -0
- package/dist/chatWrappers/utils/resolveChatWrapper.d.ts +69 -0
- package/dist/chatWrappers/utils/resolveChatWrapper.js +214 -0
- package/dist/chatWrappers/utils/resolveChatWrapper.js.map +1 -0
- package/dist/cli/cli.js +21 -7
- package/dist/cli/cli.js.map +1 -1
- package/dist/cli/commands/BuildCommand.d.ts +6 -4
- package/dist/cli/commands/BuildCommand.js +103 -41
- package/dist/cli/commands/BuildCommand.js.map +1 -1
- package/dist/cli/commands/ChatCommand.d.ts +18 -6
- package/dist/cli/commands/ChatCommand.js +298 -142
- package/dist/cli/commands/ChatCommand.js.map +1 -1
- package/dist/cli/commands/ClearCommand.d.ts +1 -1
- package/dist/cli/commands/ClearCommand.js +11 -12
- package/dist/cli/commands/ClearCommand.js.map +1 -1
- package/dist/cli/commands/CompleteCommand.d.ts +29 -0
- package/dist/cli/commands/CompleteCommand.js +365 -0
- package/dist/cli/commands/CompleteCommand.js.map +1 -0
- package/dist/cli/commands/DebugCommand.d.ts +7 -0
- package/dist/cli/commands/DebugCommand.js +54 -0
- package/dist/cli/commands/DebugCommand.js.map +1 -0
- package/dist/cli/commands/DownloadCommand.d.ts +6 -4
- package/dist/cli/commands/DownloadCommand.js +120 -69
- package/dist/cli/commands/DownloadCommand.js.map +1 -1
- package/dist/cli/commands/InfillCommand.d.ts +31 -0
- package/dist/cli/commands/InfillCommand.js +401 -0
- package/dist/cli/commands/InfillCommand.js.map +1 -0
- package/dist/cli/commands/InitCommand.d.ts +11 -0
- package/dist/cli/commands/InitCommand.js +195 -0
- package/dist/cli/commands/InitCommand.js.map +1 -0
- package/dist/cli/commands/OnPostInstallCommand.js +9 -10
- package/dist/cli/commands/OnPostInstallCommand.js.map +1 -1
- package/dist/cli/commands/PullCommand.d.ts +12 -0
- package/dist/cli/commands/PullCommand.js +117 -0
- package/dist/cli/commands/PullCommand.js.map +1 -0
- package/dist/cli/commands/inspect/InspectCommand.d.ts +4 -0
- package/dist/cli/commands/inspect/InspectCommand.js +19 -0
- package/dist/cli/commands/inspect/InspectCommand.js.map +1 -0
- package/dist/cli/commands/inspect/commands/InspectGgufCommand.d.ts +12 -0
- package/dist/cli/commands/inspect/commands/InspectGgufCommand.js +136 -0
- package/dist/cli/commands/inspect/commands/InspectGgufCommand.js.map +1 -0
- package/dist/cli/commands/inspect/commands/InspectGpuCommand.d.ts +4 -0
- package/dist/cli/commands/inspect/commands/InspectGpuCommand.js +138 -0
- package/dist/cli/commands/inspect/commands/InspectGpuCommand.js.map +1 -0
- package/dist/cli/commands/inspect/commands/InspectMeasureCommand.d.ts +17 -0
- package/dist/cli/commands/inspect/commands/InspectMeasureCommand.js +613 -0
- package/dist/cli/commands/inspect/commands/InspectMeasureCommand.js.map +1 -0
- package/dist/cli/projectTemplates.d.ts +7 -0
- package/dist/cli/projectTemplates.js +10 -0
- package/dist/cli/projectTemplates.js.map +1 -0
- package/dist/cli/recommendedModels.d.ts +2 -0
- package/dist/cli/recommendedModels.js +342 -0
- package/dist/cli/recommendedModels.js.map +1 -0
- package/dist/cli/startCreateCli.d.ts +2 -0
- package/dist/cli/startCreateCli.js +26 -0
- package/dist/cli/startCreateCli.js.map +1 -0
- package/dist/cli/utils/ConsoleInteraction.d.ts +23 -0
- package/dist/cli/utils/ConsoleInteraction.js +122 -0
- package/dist/cli/utils/ConsoleInteraction.js.map +1 -0
- package/dist/cli/utils/ConsoleTable.d.ts +23 -0
- package/dist/cli/utils/ConsoleTable.js +86 -0
- package/dist/cli/utils/ConsoleTable.js.map +1 -0
- package/dist/cli/utils/basicChooseFromListConsoleInteraction.d.ts +13 -0
- package/dist/cli/utils/basicChooseFromListConsoleInteraction.js +111 -0
- package/dist/cli/utils/basicChooseFromListConsoleInteraction.js.map +1 -0
- package/dist/cli/utils/consolePromptQuestion.d.ts +6 -0
- package/dist/cli/utils/consolePromptQuestion.js +82 -0
- package/dist/cli/utils/consolePromptQuestion.js.map +1 -0
- package/dist/cli/utils/getReadablePath.d.ts +1 -0
- package/dist/cli/utils/getReadablePath.js +14 -0
- package/dist/cli/utils/getReadablePath.js.map +1 -0
- package/dist/cli/utils/interactivelyAskForModel.d.ts +7 -0
- package/dist/cli/utils/interactivelyAskForModel.js +451 -0
- package/dist/cli/utils/interactivelyAskForModel.js.map +1 -0
- package/dist/cli/utils/logUsedGpuTypeOption.d.ts +2 -0
- package/dist/cli/utils/logUsedGpuTypeOption.js +9 -0
- package/dist/cli/utils/logUsedGpuTypeOption.js.map +1 -0
- package/dist/cli/utils/printCommonInfoLines.d.ts +9 -0
- package/dist/cli/utils/printCommonInfoLines.js +71 -0
- package/dist/cli/utils/printCommonInfoLines.js.map +1 -0
- package/dist/cli/utils/printInfoLine.d.ts +12 -0
- package/dist/cli/utils/printInfoLine.js +54 -0
- package/dist/cli/utils/printInfoLine.js.map +1 -0
- package/dist/cli/utils/projectTemplates.d.ts +19 -0
- package/dist/cli/utils/projectTemplates.js +47 -0
- package/dist/cli/utils/projectTemplates.js.map +1 -0
- package/dist/cli/utils/resolveCommandGgufPath.d.ts +4 -0
- package/dist/cli/utils/resolveCommandGgufPath.js +71 -0
- package/dist/cli/utils/resolveCommandGgufPath.js.map +1 -0
- package/dist/cli/utils/resolveHeaderFlag.d.ts +1 -0
- package/dist/cli/utils/resolveHeaderFlag.js +21 -0
- package/dist/cli/utils/resolveHeaderFlag.js.map +1 -0
- package/dist/cli/utils/resolveModelRecommendationFileOptions.d.ts +19 -0
- package/dist/cli/utils/resolveModelRecommendationFileOptions.js +7 -0
- package/dist/cli/utils/resolveModelRecommendationFileOptions.js.map +1 -0
- package/dist/cli/utils/splitAnsiToLines.d.ts +1 -0
- package/dist/cli/utils/splitAnsiToLines.js +32 -0
- package/dist/cli/utils/splitAnsiToLines.js.map +1 -0
- package/dist/cli/utils/withCliCommandDescriptionDocsUrl.d.ts +2 -0
- package/dist/cli/utils/withCliCommandDescriptionDocsUrl.js +23 -0
- package/dist/cli/utils/withCliCommandDescriptionDocsUrl.js.map +1 -0
- package/dist/commands.d.ts +1 -0
- package/dist/commands.js +3 -0
- package/dist/commands.js.map +1 -1
- package/dist/config.d.ts +38 -5
- package/dist/config.js +61 -16
- package/dist/config.js.map +1 -1
- package/dist/consts.d.ts +3 -0
- package/dist/consts.js +10 -0
- package/dist/consts.js.map +1 -0
- package/dist/{llamaEvaluator → evaluator}/LlamaChat/LlamaChat.d.ts +37 -35
- package/dist/{llamaEvaluator → evaluator}/LlamaChat/LlamaChat.js +298 -221
- package/dist/evaluator/LlamaChat/LlamaChat.js.map +1 -0
- package/dist/{llamaEvaluator → evaluator}/LlamaChat/utils/FunctionCallGrammar.d.ts +2 -1
- package/dist/{llamaEvaluator → evaluator}/LlamaChat/utils/FunctionCallGrammar.js +5 -3
- package/dist/evaluator/LlamaChat/utils/FunctionCallGrammar.js.map +1 -0
- package/dist/{llamaEvaluator → evaluator}/LlamaChat/utils/contextShiftStrategies/eraseFirstResponseAndKeepFirstSystemChatContextShiftStrategy.js +18 -0
- package/dist/evaluator/LlamaChat/utils/contextShiftStrategies/eraseFirstResponseAndKeepFirstSystemChatContextShiftStrategy.js.map +1 -0
- package/dist/{llamaEvaluator → evaluator}/LlamaChatSession/LlamaChatSession.d.ts +40 -3
- package/dist/{llamaEvaluator → evaluator}/LlamaChatSession/LlamaChatSession.js +28 -7
- package/dist/evaluator/LlamaChatSession/LlamaChatSession.js.map +1 -0
- package/dist/{llamaEvaluator → evaluator}/LlamaChatSession/utils/defineChatSessionFunction.d.ts +3 -0
- package/dist/{llamaEvaluator → evaluator}/LlamaChatSession/utils/defineChatSessionFunction.js +3 -0
- package/dist/evaluator/LlamaChatSession/utils/defineChatSessionFunction.js.map +1 -0
- package/dist/evaluator/LlamaCompletion.d.ts +155 -0
- package/dist/evaluator/LlamaCompletion.js +405 -0
- package/dist/evaluator/LlamaCompletion.js.map +1 -0
- package/dist/{llamaEvaluator → evaluator}/LlamaContext/LlamaContext.d.ts +41 -20
- package/dist/{llamaEvaluator → evaluator}/LlamaContext/LlamaContext.js +271 -81
- package/dist/evaluator/LlamaContext/LlamaContext.js.map +1 -0
- package/dist/evaluator/LlamaContext/types.d.ts +140 -0
- package/dist/evaluator/LlamaContext/types.js.map +1 -0
- package/dist/evaluator/LlamaContext/utils/batchItemsPrioritizationStrategies/firstInFirstOutStrategy.js.map +1 -0
- package/dist/evaluator/LlamaContext/utils/batchItemsPrioritizationStrategies/maximumParallelismStrategy.js.map +1 -0
- package/dist/evaluator/LlamaContext/utils/resolveBatchItemsPrioritizationStrategy.d.ts +2 -0
- package/dist/{llamaEvaluator/LlamaContext/utils/resolveBatchItemsPrioritizingStrategy.js → evaluator/LlamaContext/utils/resolveBatchItemsPrioritizationStrategy.js} +4 -4
- package/dist/evaluator/LlamaContext/utils/resolveBatchItemsPrioritizationStrategy.js.map +1 -0
- package/dist/evaluator/LlamaEmbeddingContext.d.ts +51 -0
- package/dist/evaluator/LlamaEmbeddingContext.js +73 -0
- package/dist/evaluator/LlamaEmbeddingContext.js.map +1 -0
- package/dist/{llamaEvaluator → evaluator}/LlamaGrammar.d.ts +8 -5
- package/dist/{llamaEvaluator → evaluator}/LlamaGrammar.js +13 -10
- package/dist/evaluator/LlamaGrammar.js.map +1 -0
- package/dist/{llamaEvaluator → evaluator}/LlamaGrammarEvaluationState.js +4 -4
- package/dist/evaluator/LlamaGrammarEvaluationState.js.map +1 -0
- package/dist/{llamaEvaluator → evaluator}/LlamaJsonSchemaGrammar.d.ts +2 -1
- package/dist/{llamaEvaluator → evaluator}/LlamaJsonSchemaGrammar.js +3 -3
- package/dist/evaluator/LlamaJsonSchemaGrammar.js.map +1 -0
- package/dist/evaluator/LlamaModel.d.ts +230 -0
- package/dist/evaluator/LlamaModel.js +597 -0
- package/dist/evaluator/LlamaModel.js.map +1 -0
- package/dist/evaluator/TokenBias.d.ts +22 -0
- package/dist/evaluator/TokenBias.js +33 -0
- package/dist/evaluator/TokenBias.js.map +1 -0
- package/dist/evaluator/TokenMeter.d.ts +54 -0
- package/dist/evaluator/TokenMeter.js +86 -0
- package/dist/evaluator/TokenMeter.js.map +1 -0
- package/dist/gguf/consts.d.ts +3 -0
- package/dist/gguf/consts.js +8 -0
- package/dist/gguf/consts.js.map +1 -0
- package/dist/gguf/errors/InvalidGgufMagicError.d.ts +3 -0
- package/dist/gguf/errors/InvalidGgufMagicError.js +6 -0
- package/dist/gguf/errors/InvalidGgufMagicError.js.map +1 -0
- package/dist/gguf/errors/UnsupportedGgufValueTypeError.d.ts +4 -0
- package/dist/gguf/errors/UnsupportedGgufValueTypeError.js +9 -0
- package/dist/gguf/errors/UnsupportedGgufValueTypeError.js.map +1 -0
- package/dist/gguf/fileReaders/GgufFileReader.d.ts +33 -0
- package/dist/gguf/fileReaders/GgufFileReader.js +76 -0
- package/dist/gguf/fileReaders/GgufFileReader.js.map +1 -0
- package/dist/gguf/fileReaders/GgufFsFileReader.d.ts +17 -0
- package/dist/gguf/fileReaders/GgufFsFileReader.js +45 -0
- package/dist/gguf/fileReaders/GgufFsFileReader.js.map +1 -0
- package/dist/gguf/fileReaders/GgufNetworkFetchFileReader.d.ts +22 -0
- package/dist/gguf/fileReaders/GgufNetworkFetchFileReader.js +63 -0
- package/dist/gguf/fileReaders/GgufNetworkFetchFileReader.js.map +1 -0
- package/dist/gguf/insights/GgufInsights.d.ts +42 -0
- package/dist/gguf/insights/GgufInsights.js +361 -0
- package/dist/gguf/insights/GgufInsights.js.map +1 -0
- package/dist/gguf/insights/GgufInsightsConfigurationResolver.d.ts +87 -0
- package/dist/gguf/insights/GgufInsightsConfigurationResolver.js +136 -0
- package/dist/gguf/insights/GgufInsightsConfigurationResolver.js.map +1 -0
- package/dist/gguf/insights/utils/resolveContextContextSizeOption.d.ts +18 -0
- package/dist/gguf/insights/utils/resolveContextContextSizeOption.js +76 -0
- package/dist/gguf/insights/utils/resolveContextContextSizeOption.js.map +1 -0
- package/dist/gguf/insights/utils/resolveModelGpuLayersOption.d.ts +14 -0
- package/dist/gguf/insights/utils/resolveModelGpuLayersOption.js +177 -0
- package/dist/gguf/insights/utils/resolveModelGpuLayersOption.js.map +1 -0
- package/dist/gguf/insights/utils/scoreLevels.d.ts +5 -0
- package/dist/gguf/insights/utils/scoreLevels.js +16 -0
- package/dist/gguf/insights/utils/scoreLevels.js.map +1 -0
- package/dist/gguf/parser/GgufV2Parser.d.ts +19 -0
- package/dist/gguf/parser/GgufV2Parser.js +115 -0
- package/dist/gguf/parser/GgufV2Parser.js.map +1 -0
- package/dist/gguf/parser/GgufV3Parser.d.ts +3 -0
- package/dist/gguf/parser/GgufV3Parser.js +4 -0
- package/dist/gguf/parser/GgufV3Parser.js.map +1 -0
- package/dist/gguf/parser/parseGguf.d.ts +8 -0
- package/dist/gguf/parser/parseGguf.js +63 -0
- package/dist/gguf/parser/parseGguf.js.map +1 -0
- package/dist/gguf/readGgufFileInfo.d.ts +33 -0
- package/dist/gguf/readGgufFileInfo.js +66 -0
- package/dist/gguf/readGgufFileInfo.js.map +1 -0
- package/dist/gguf/types/GgufFileInfoTypes.d.ts +84 -0
- package/dist/gguf/types/GgufFileInfoTypes.js +18 -0
- package/dist/gguf/types/GgufFileInfoTypes.js.map +1 -0
- package/dist/gguf/types/GgufMetadataTypes.d.ts +334 -0
- package/dist/gguf/types/GgufMetadataTypes.js +86 -0
- package/dist/gguf/types/GgufMetadataTypes.js.map +1 -0
- package/dist/gguf/types/GgufTensorInfoTypes.d.ts +37 -0
- package/dist/gguf/types/GgufTensorInfoTypes.js +33 -0
- package/dist/gguf/types/GgufTensorInfoTypes.js.map +1 -0
- package/dist/gguf/utils/GgufReadOffset.d.ts +6 -0
- package/dist/gguf/utils/GgufReadOffset.js +18 -0
- package/dist/gguf/utils/GgufReadOffset.js.map +1 -0
- package/dist/gguf/utils/convertMetadataKeyValueRecordToNestedObject.d.ts +5 -0
- package/dist/gguf/utils/convertMetadataKeyValueRecordToNestedObject.js +38 -0
- package/dist/gguf/utils/convertMetadataKeyValueRecordToNestedObject.js.map +1 -0
- package/dist/gguf/utils/getGgufFileTypeName.d.ts +4 -0
- package/dist/gguf/utils/getGgufFileTypeName.js +13 -0
- package/dist/gguf/utils/getGgufFileTypeName.js.map +1 -0
- package/dist/gguf/utils/getGgufMetadataArchitectureData.d.ts +3 -0
- package/dist/gguf/utils/getGgufMetadataArchitectureData.js +4 -0
- package/dist/gguf/utils/getGgufMetadataArchitectureData.js.map +1 -0
- package/dist/gguf/utils/normalizeGgufDownloadUrl.d.ts +1 -0
- package/dist/gguf/utils/normalizeGgufDownloadUrl.js +16 -0
- package/dist/gguf/utils/normalizeGgufDownloadUrl.js.map +1 -0
- package/dist/gguf/utils/resolveBinarySplitGgufPartUrls.d.ts +2 -0
- package/dist/gguf/utils/resolveBinarySplitGgufPartUrls.js +39 -0
- package/dist/gguf/utils/resolveBinarySplitGgufPartUrls.js.map +1 -0
- package/dist/gguf/utils/resolveSplitGgufParts.d.ts +7 -0
- package/dist/gguf/utils/resolveSplitGgufParts.js +55 -0
- package/dist/gguf/utils/resolveSplitGgufParts.js.map +1 -0
- package/dist/index.d.ts +37 -17
- package/dist/index.js +33 -14
- package/dist/index.js.map +1 -1
- package/dist/state.d.ts +4 -0
- package/dist/state.js +14 -0
- package/dist/state.js.map +1 -1
- package/dist/types.d.ts +53 -2
- package/dist/types.js.map +1 -1
- package/dist/utils/DisposeGuard.d.ts +13 -0
- package/dist/utils/DisposeGuard.js +120 -0
- package/dist/utils/DisposeGuard.js.map +1 -0
- package/dist/utils/InsufficientMemoryError.d.ts +3 -0
- package/dist/utils/InsufficientMemoryError.js +6 -0
- package/dist/utils/InsufficientMemoryError.js.map +1 -0
- package/dist/utils/LlamaText.d.ts +50 -25
- package/dist/utils/LlamaText.js +367 -155
- package/dist/utils/LlamaText.js.map +1 -1
- package/dist/utils/StopGenerationDetector.d.ts +1 -1
- package/dist/utils/StopGenerationDetector.js +23 -18
- package/dist/utils/StopGenerationDetector.js.map +1 -1
- package/dist/utils/TokenStreamRegulator.d.ts +8 -4
- package/dist/utils/TokenStreamRegulator.js +78 -8
- package/dist/utils/TokenStreamRegulator.js.map +1 -1
- package/dist/utils/UnsupportedError.d.ts +2 -0
- package/dist/utils/UnsupportedError.js +7 -0
- package/dist/utils/UnsupportedError.js.map +1 -0
- package/dist/utils/cmake.js +38 -20
- package/dist/utils/cmake.js.map +1 -1
- package/dist/utils/createModelDownloader.d.ts +102 -0
- package/dist/utils/createModelDownloader.js +226 -0
- package/dist/utils/createModelDownloader.js.map +1 -0
- package/dist/utils/findBestOption.d.ts +4 -0
- package/dist/utils/findBestOption.js +15 -0
- package/dist/utils/findBestOption.js.map +1 -0
- package/dist/utils/findCharacterRemovalCountToFitChatHistoryInContext.js +18 -8
- package/dist/utils/findCharacterRemovalCountToFitChatHistoryInContext.js.map +1 -1
- package/dist/utils/gbnfJson/getGbnfGrammarForGbnfJsonSchema.d.ts +5 -0
- package/dist/utils/gbnfJson/getGbnfGrammarForGbnfJsonSchema.js +11 -0
- package/dist/utils/gbnfJson/getGbnfGrammarForGbnfJsonSchema.js.map +1 -0
- package/dist/utils/gbnfJson/terminals/GbnfArray.d.ts +3 -1
- package/dist/utils/gbnfJson/terminals/GbnfArray.js +10 -5
- package/dist/utils/gbnfJson/terminals/GbnfArray.js.map +1 -1
- package/dist/utils/gbnfJson/terminals/GbnfBoolean.d.ts +1 -1
- package/dist/utils/gbnfJson/terminals/GbnfBoolean.js.map +1 -1
- package/dist/utils/gbnfJson/terminals/GbnfBooleanValue.js.map +1 -1
- package/dist/utils/gbnfJson/terminals/GbnfGrammar.js.map +1 -1
- package/dist/utils/gbnfJson/terminals/GbnfNull.d.ts +1 -1
- package/dist/utils/gbnfJson/terminals/GbnfNull.js.map +1 -1
- package/dist/utils/gbnfJson/terminals/GbnfNumber.d.ts +1 -1
- package/dist/utils/gbnfJson/terminals/GbnfNumber.js.map +1 -1
- package/dist/utils/gbnfJson/terminals/GbnfNumberValue.js.map +1 -1
- package/dist/utils/gbnfJson/terminals/GbnfObjectMap.d.ts +3 -1
- package/dist/utils/gbnfJson/terminals/GbnfObjectMap.js +9 -4
- package/dist/utils/gbnfJson/terminals/GbnfObjectMap.js.map +1 -1
- package/dist/utils/gbnfJson/terminals/GbnfOr.js.map +1 -1
- package/dist/utils/gbnfJson/terminals/GbnfRepetition.d.ts +9 -0
- package/dist/utils/gbnfJson/terminals/GbnfRepetition.js +37 -0
- package/dist/utils/gbnfJson/terminals/GbnfRepetition.js.map +1 -0
- package/dist/utils/gbnfJson/terminals/GbnfString.d.ts +1 -1
- package/dist/utils/gbnfJson/terminals/GbnfString.js +23 -5
- package/dist/utils/gbnfJson/terminals/GbnfString.js.map +1 -1
- package/dist/utils/gbnfJson/terminals/GbnfStringValue.js.map +1 -1
- package/dist/utils/gbnfJson/terminals/GbnfVerbatimText.js.map +1 -1
- package/dist/utils/gbnfJson/terminals/GbnfWhitespace.d.ts +7 -4
- package/dist/utils/gbnfJson/terminals/GbnfWhitespace.js +37 -9
- package/dist/utils/gbnfJson/terminals/GbnfWhitespace.js.map +1 -1
- package/dist/utils/gbnfJson/terminals/gbnfConsts.d.ts +5 -4
- package/dist/utils/gbnfJson/terminals/gbnfConsts.js +14 -3
- package/dist/utils/gbnfJson/terminals/gbnfConsts.js.map +1 -1
- package/dist/utils/gbnfJson/utils/GbnfJsonScopeState.d.ts +10 -0
- package/dist/utils/gbnfJson/utils/GbnfJsonScopeState.js +15 -0
- package/dist/utils/gbnfJson/utils/GbnfJsonScopeState.js.map +1 -0
- package/dist/utils/gbnfJson/utils/getGbnfJsonTerminalForGbnfJsonSchema.d.ts +2 -1
- package/dist/utils/gbnfJson/utils/getGbnfJsonTerminalForGbnfJsonSchema.js +6 -5
- package/dist/utils/gbnfJson/utils/getGbnfJsonTerminalForGbnfJsonSchema.js.map +1 -1
- package/dist/utils/getBuildDefaults.d.ts +1 -2
- package/dist/utils/getBuildDefaults.js +2 -3
- package/dist/utils/getBuildDefaults.js.map +1 -1
- package/dist/utils/getConsoleLogPrefix.d.ts +1 -0
- package/dist/utils/getConsoleLogPrefix.js +10 -0
- package/dist/utils/getConsoleLogPrefix.js.map +1 -0
- package/dist/utils/getGrammarsFolder.d.ts +2 -1
- package/dist/utils/getGrammarsFolder.js +8 -7
- package/dist/utils/getGrammarsFolder.js.map +1 -1
- package/dist/utils/getModuleVersion.d.ts +1 -0
- package/dist/utils/getModuleVersion.js +13 -0
- package/dist/utils/getModuleVersion.js.map +1 -0
- package/dist/utils/getQueuedTokensBeforeStopTrigger.d.ts +6 -0
- package/dist/utils/getQueuedTokensBeforeStopTrigger.js +22 -0
- package/dist/utils/getQueuedTokensBeforeStopTrigger.js.map +1 -0
- package/dist/utils/getReadableContextSize.d.ts +1 -0
- package/dist/utils/getReadableContextSize.js +7 -0
- package/dist/utils/getReadableContextSize.js.map +1 -0
- package/dist/utils/getTypeScriptTypeStringForGbnfJsonSchema.js +15 -11
- package/dist/utils/getTypeScriptTypeStringForGbnfJsonSchema.js.map +1 -1
- package/dist/utils/gitReleaseBundles.js +73 -5
- package/dist/utils/gitReleaseBundles.js.map +1 -1
- package/dist/utils/hashString.d.ts +1 -0
- package/dist/utils/hashString.js +8 -0
- package/dist/utils/hashString.js.map +1 -0
- package/dist/utils/isLockfileActive.d.ts +4 -0
- package/dist/utils/isLockfileActive.js +12 -0
- package/dist/utils/isLockfileActive.js.map +1 -0
- package/dist/utils/isToken.d.ts +2 -0
- package/dist/utils/isToken.js +4 -0
- package/dist/utils/isToken.js.map +1 -0
- package/dist/utils/isUrl.d.ts +1 -0
- package/dist/utils/isUrl.js +15 -0
- package/dist/utils/isUrl.js.map +1 -0
- package/dist/utils/mergeUnionTypes.d.ts +10 -0
- package/dist/utils/mergeUnionTypes.js +2 -0
- package/dist/utils/mergeUnionTypes.js.map +1 -0
- package/dist/utils/parseModelFileName.d.ts +1 -0
- package/dist/utils/parseModelFileName.js +6 -1
- package/dist/utils/parseModelFileName.js.map +1 -1
- package/dist/utils/parseTextTemplate.d.ts +66 -0
- package/dist/utils/parseTextTemplate.js +116 -0
- package/dist/utils/parseTextTemplate.js.map +1 -0
- package/dist/utils/prettyPrintObject.d.ts +10 -0
- package/dist/utils/prettyPrintObject.js +84 -0
- package/dist/utils/prettyPrintObject.js.map +1 -0
- package/dist/utils/removeNullFields.d.ts +2 -1
- package/dist/utils/removeNullFields.js +8 -0
- package/dist/utils/removeNullFields.js.map +1 -1
- package/dist/utils/resolveGithubRelease.d.ts +2 -0
- package/dist/utils/resolveGithubRelease.js +36 -0
- package/dist/utils/resolveGithubRelease.js.map +1 -0
- package/dist/utils/runtime.d.ts +4 -0
- package/dist/utils/runtime.js +8 -0
- package/dist/utils/runtime.js.map +1 -0
- package/dist/utils/spawnCommand.d.ts +11 -1
- package/dist/utils/spawnCommand.js +56 -6
- package/dist/utils/spawnCommand.js.map +1 -1
- package/dist/utils/tokenizeInput.d.ts +3 -0
- package/dist/utils/tokenizeInput.js +12 -0
- package/dist/utils/tokenizeInput.js.map +1 -0
- package/dist/utils/utilTypes.d.ts +3 -0
- package/dist/utils/utilTypes.js +2 -0
- package/dist/utils/utilTypes.js.map +1 -0
- package/dist/utils/waitForLockfileRelease.d.ts +5 -0
- package/dist/utils/waitForLockfileRelease.js +20 -0
- package/dist/utils/waitForLockfileRelease.js.map +1 -0
- package/dist/utils/withLockfile.d.ts +7 -0
- package/dist/utils/withLockfile.js +44 -0
- package/dist/utils/withLockfile.js.map +1 -0
- package/dist/utils/withOra.d.ts +2 -0
- package/dist/utils/withOra.js +22 -6
- package/dist/utils/withOra.js.map +1 -1
- package/dist/utils/withProgressLog.d.ts +23 -0
- package/dist/utils/withProgressLog.js +211 -0
- package/dist/utils/withProgressLog.js.map +1 -0
- package/dist/utils/withStatusLogs.d.ts +2 -1
- package/dist/utils/withStatusLogs.js +12 -9
- package/dist/utils/withStatusLogs.js.map +1 -1
- package/llama/.clang-format +1 -2
- package/llama/CMakeLists.txt +115 -4
- package/llama/addon.cpp +1318 -99
- package/llama/binariesGithubRelease.json +1 -1
- package/llama/gitRelease.bundle +0 -0
- package/llama/gpuInfo/cuda-gpu-info.cu +120 -0
- package/llama/gpuInfo/cuda-gpu-info.h +10 -0
- package/llama/gpuInfo/metal-gpu-info.h +8 -0
- package/llama/gpuInfo/metal-gpu-info.mm +30 -0
- package/llama/gpuInfo/vulkan-gpu-info.cpp +83 -0
- package/llama/gpuInfo/vulkan-gpu-info.h +9 -0
- package/llama/grammars/README.md +11 -1
- package/llama/grammars/json.gbnf +1 -1
- package/llama/grammars/json_arr.gbnf +1 -1
- package/llama/llama.cpp.info.json +4 -0
- package/llama/toolchains/win32.host-x64.target-arm64.cmake +41 -0
- package/llamaBins/linux-arm64/_nlcBuildMetadata.json +1 -0
- package/llamaBins/linux-arm64/llama-addon.node +0 -0
- package/llamaBins/linux-armv7l/_nlcBuildMetadata.json +1 -0
- package/llamaBins/linux-armv7l/llama-addon.node +0 -0
- package/llamaBins/linux-x64/_nlcBuildMetadata.json +1 -0
- package/llamaBins/linux-x64/llama-addon.node +0 -0
- package/llamaBins/linux-x64-cuda/_nlcBuildMetadata.json +1 -0
- package/llamaBins/linux-x64-cuda/llama-addon.node +0 -0
- package/llamaBins/linux-x64-vulkan/_nlcBuildMetadata.json +1 -0
- package/llamaBins/linux-x64-vulkan/llama-addon.node +0 -0
- package/llamaBins/mac-arm64-metal/_nlcBuildMetadata.json +1 -0
- package/llamaBins/mac-arm64-metal/default.metallib +0 -0
- package/llamaBins/mac-arm64-metal/llama-addon.node +0 -0
- package/llamaBins/mac-x64/_nlcBuildMetadata.json +1 -0
- package/llamaBins/mac-x64/llama-addon.node +0 -0
- package/llamaBins/win-arm64/_nlcBuildMetadata.json +1 -0
- package/llamaBins/win-arm64/llama-addon.exp +0 -0
- package/llamaBins/win-arm64/llama-addon.lib +0 -0
- package/llamaBins/win-arm64/llama-addon.node +0 -0
- package/llamaBins/win-x64/_nlcBuildMetadata.json +1 -0
- package/llamaBins/win-x64/llama-addon.exp +0 -0
- package/llamaBins/win-x64/llama-addon.lib +0 -0
- package/llamaBins/win-x64/llama-addon.node +0 -0
- package/llamaBins/win-x64-cuda/_nlcBuildMetadata.json +1 -0
- package/llamaBins/win-x64-cuda/llama-addon.exp +0 -0
- package/llamaBins/win-x64-cuda/llama-addon.lib +0 -0
- package/llamaBins/win-x64-cuda/llama-addon.node +0 -0
- package/llamaBins/win-x64-vulkan/_nlcBuildMetadata.json +1 -0
- package/llamaBins/win-x64-vulkan/llama-addon.exp +0 -0
- package/llamaBins/win-x64-vulkan/llama-addon.lib +0 -0
- package/llamaBins/win-x64-vulkan/llama-addon.node +0 -0
- package/package.json +61 -34
- package/templates/package.json +10 -0
- package/dist/AbortError.d.ts +0 -2
- package/dist/AbortError.js +0 -7
- package/dist/AbortError.js.map +0 -1
- package/dist/chatWrappers/LlamaChatWrapper.js.map +0 -1
- package/dist/chatWrappers/resolveChatWrapperBasedOnModel.d.ts +0 -13
- package/dist/chatWrappers/resolveChatWrapperBasedOnModel.js +0 -55
- package/dist/chatWrappers/resolveChatWrapperBasedOnModel.js.map +0 -1
- package/dist/llamaEvaluator/LlamaBins.d.ts +0 -18
- package/dist/llamaEvaluator/LlamaBins.js +0 -5
- package/dist/llamaEvaluator/LlamaBins.js.map +0 -1
- package/dist/llamaEvaluator/LlamaChat/LlamaChat.js.map +0 -1
- package/dist/llamaEvaluator/LlamaChat/utils/FunctionCallGrammar.js.map +0 -1
- package/dist/llamaEvaluator/LlamaChat/utils/contextShiftStrategies/eraseFirstResponseAndKeepFirstSystemChatContextShiftStrategy.js.map +0 -1
- package/dist/llamaEvaluator/LlamaChatSession/LlamaChatSession.js.map +0 -1
- package/dist/llamaEvaluator/LlamaChatSession/utils/defineChatSessionFunction.js.map +0 -1
- package/dist/llamaEvaluator/LlamaContext/LlamaContext.js.map +0 -1
- package/dist/llamaEvaluator/LlamaContext/types.d.ts +0 -86
- package/dist/llamaEvaluator/LlamaContext/types.js.map +0 -1
- package/dist/llamaEvaluator/LlamaContext/utils/batchItemsPrioritizingStrategies/firstInFirstOutStrategy.js.map +0 -1
- package/dist/llamaEvaluator/LlamaContext/utils/batchItemsPrioritizingStrategies/maximumParallelismStrategy.js.map +0 -1
- package/dist/llamaEvaluator/LlamaContext/utils/resolveBatchItemsPrioritizingStrategy.d.ts +0 -2
- package/dist/llamaEvaluator/LlamaContext/utils/resolveBatchItemsPrioritizingStrategy.js.map +0 -1
- package/dist/llamaEvaluator/LlamaGrammar.js.map +0 -1
- package/dist/llamaEvaluator/LlamaGrammarEvaluationState.js.map +0 -1
- package/dist/llamaEvaluator/LlamaJsonSchemaGrammar.js.map +0 -1
- package/dist/llamaEvaluator/LlamaModel.d.ts +0 -119
- package/dist/llamaEvaluator/LlamaModel.js +0 -322
- package/dist/llamaEvaluator/LlamaModel.js.map +0 -1
- package/dist/utils/binariesGithubRelease.js.map +0 -1
- package/dist/utils/clearLlamaBuild.d.ts +0 -1
- package/dist/utils/clearLlamaBuild.js +0 -12
- package/dist/utils/clearLlamaBuild.js.map +0 -1
- package/dist/utils/cloneLlamaCppRepo.d.ts +0 -2
- package/dist/utils/cloneLlamaCppRepo.js +0 -102
- package/dist/utils/cloneLlamaCppRepo.js.map +0 -1
- package/dist/utils/compileLLamaCpp.d.ts +0 -8
- package/dist/utils/compileLLamaCpp.js +0 -132
- package/dist/utils/compileLLamaCpp.js.map +0 -1
- package/dist/utils/getBin.js +0 -78
- package/dist/utils/getBin.js.map +0 -1
- package/dist/utils/getGbnfGrammarForGbnfJsonSchema.d.ts +0 -2
- package/dist/utils/getGbnfGrammarForGbnfJsonSchema.js +0 -9
- package/dist/utils/getGbnfGrammarForGbnfJsonSchema.js.map +0 -1
- package/dist/utils/getReleaseInfo.d.ts +0 -7
- package/dist/utils/getReleaseInfo.js +0 -30
- package/dist/utils/getReleaseInfo.js.map +0 -1
- package/dist/utils/parseModelTypeDescription.d.ts +0 -6
- package/dist/utils/parseModelTypeDescription.js +0 -9
- package/dist/utils/parseModelTypeDescription.js.map +0 -1
- package/dist/utils/resolveChatWrapper.d.ts +0 -4
- package/dist/utils/resolveChatWrapper.js +0 -16
- package/dist/utils/resolveChatWrapper.js.map +0 -1
- package/dist/utils/usedBinFlag.d.ts +0 -6
- package/dist/utils/usedBinFlag.js +0 -15
- package/dist/utils/usedBinFlag.js.map +0 -1
- package/llama/usedBin.json +0 -3
- package/llamaBins/mac-arm64/llama-addon.node +0 -0
- /package/dist/{utils → bindings/utils}/binariesGithubRelease.d.ts +0 -0
- /package/dist/{llamaEvaluator → evaluator}/LlamaChat/utils/contextShiftStrategies/eraseFirstResponseAndKeepFirstSystemChatContextShiftStrategy.d.ts +0 -0
- /package/dist/{llamaEvaluator → evaluator}/LlamaContext/types.js +0 -0
- /package/dist/{llamaEvaluator/LlamaContext/utils/batchItemsPrioritizingStrategies → evaluator/LlamaContext/utils/batchItemsPrioritizationStrategies}/firstInFirstOutStrategy.d.ts +0 -0
- /package/dist/{llamaEvaluator/LlamaContext/utils/batchItemsPrioritizingStrategies → evaluator/LlamaContext/utils/batchItemsPrioritizationStrategies}/firstInFirstOutStrategy.js +0 -0
- /package/dist/{llamaEvaluator/LlamaContext/utils/batchItemsPrioritizingStrategies → evaluator/LlamaContext/utils/batchItemsPrioritizationStrategies}/maximumParallelismStrategy.d.ts +0 -0
- /package/dist/{llamaEvaluator/LlamaContext/utils/batchItemsPrioritizingStrategies → evaluator/LlamaContext/utils/batchItemsPrioritizationStrategies}/maximumParallelismStrategy.js +0 -0
- /package/dist/{llamaEvaluator → evaluator}/LlamaGrammarEvaluationState.d.ts +0 -0
|
@@ -0,0 +1,76 @@
|
|
|
1
|
+
import { minAllowedContextSizeInCalculations } from "../../../config.js";
|
|
2
|
+
import { getDefaultContextBatchSize, getDefaultModelContextSize } from "../../../evaluator/LlamaContext/LlamaContext.js";
|
|
3
|
+
export function resolveContextContextSizeOption({ contextSize, batchSize, sequences, modelFileInsights, modelGpuLayers, modelTrainContextSize, getVramState, llamaGpu, ignoreMemorySafetyChecks = false, isEmbeddingContext = false }) {
|
|
4
|
+
if (contextSize == null)
|
|
5
|
+
contextSize = "auto";
|
|
6
|
+
if (typeof contextSize === "number") {
|
|
7
|
+
const resolvedContextSize = Math.max(1, Math.floor(contextSize));
|
|
8
|
+
if (ignoreMemorySafetyChecks)
|
|
9
|
+
return resolvedContextSize;
|
|
10
|
+
const vramState = getVramState();
|
|
11
|
+
const contextVram = modelFileInsights.estimateContextResourceRequirements({
|
|
12
|
+
contextSize: resolvedContextSize,
|
|
13
|
+
batchSize: batchSize ?? getDefaultContextBatchSize({ contextSize: resolvedContextSize, sequences }),
|
|
14
|
+
modelGpuLayers: modelGpuLayers,
|
|
15
|
+
sequences,
|
|
16
|
+
isEmbeddingContext
|
|
17
|
+
}).gpuVram;
|
|
18
|
+
if (contextVram > vramState.free)
|
|
19
|
+
throw new Error(`The context size of ${resolvedContextSize}${sequences > 1 ? ` with ${sequences} sequences` : ""} is too large for the available VRAM`);
|
|
20
|
+
return resolvedContextSize;
|
|
21
|
+
}
|
|
22
|
+
else if (contextSize === "auto" || typeof contextSize === "object") {
|
|
23
|
+
if (llamaGpu === false)
|
|
24
|
+
return modelTrainContextSize;
|
|
25
|
+
const vramState = getVramState();
|
|
26
|
+
if (vramState.total === 0)
|
|
27
|
+
return modelTrainContextSize;
|
|
28
|
+
const freeVram = vramState.free;
|
|
29
|
+
const maxContextSize = contextSize === "auto"
|
|
30
|
+
? getDefaultModelContextSize({ trainContextSize: modelTrainContextSize })
|
|
31
|
+
: Math.min(contextSize.max ?? getDefaultModelContextSize({ trainContextSize: modelTrainContextSize }), getDefaultModelContextSize({ trainContextSize: modelTrainContextSize }));
|
|
32
|
+
const minContextSize = contextSize === "auto"
|
|
33
|
+
? minAllowedContextSizeInCalculations
|
|
34
|
+
: Math.max(contextSize.min ?? minAllowedContextSizeInCalculations, minAllowedContextSizeInCalculations);
|
|
35
|
+
let highestCompatibleContextSize = null;
|
|
36
|
+
let step = -Math.max(1, Math.floor((maxContextSize - minContextSize) / 4));
|
|
37
|
+
for (let testContextSize = maxContextSize; testContextSize >= minContextSize && testContextSize <= maxContextSize;) {
|
|
38
|
+
const contextVram = modelFileInsights.estimateContextResourceRequirements({
|
|
39
|
+
contextSize: testContextSize,
|
|
40
|
+
batchSize: batchSize ?? getDefaultContextBatchSize({ contextSize: testContextSize, sequences }),
|
|
41
|
+
modelGpuLayers: modelGpuLayers,
|
|
42
|
+
sequences,
|
|
43
|
+
isEmbeddingContext
|
|
44
|
+
}).gpuVram;
|
|
45
|
+
if (contextVram <= freeVram) {
|
|
46
|
+
if (highestCompatibleContextSize == null || testContextSize > highestCompatibleContextSize) {
|
|
47
|
+
highestCompatibleContextSize = testContextSize;
|
|
48
|
+
if (step === -1)
|
|
49
|
+
break;
|
|
50
|
+
else if (step < 0)
|
|
51
|
+
step = Math.max(1, Math.floor(-step / 2));
|
|
52
|
+
}
|
|
53
|
+
}
|
|
54
|
+
else if (step > 0)
|
|
55
|
+
step = -Math.max(1, Math.floor(step / 2));
|
|
56
|
+
if (testContextSize == minContextSize && step === -1)
|
|
57
|
+
break;
|
|
58
|
+
testContextSize += step;
|
|
59
|
+
if (testContextSize < minContextSize) {
|
|
60
|
+
testContextSize = minContextSize;
|
|
61
|
+
step = Math.max(1, Math.floor(Math.abs(step) / 2));
|
|
62
|
+
}
|
|
63
|
+
else if (testContextSize > maxContextSize) {
|
|
64
|
+
testContextSize = maxContextSize;
|
|
65
|
+
step = -Math.max(1, Math.floor(Math.abs(step) / 2));
|
|
66
|
+
}
|
|
67
|
+
}
|
|
68
|
+
if (highestCompatibleContextSize != null)
|
|
69
|
+
return highestCompatibleContextSize;
|
|
70
|
+
if (ignoreMemorySafetyChecks)
|
|
71
|
+
return minContextSize;
|
|
72
|
+
throw new Error(`The available VRAM is too small to fit the context size of ${maxContextSize}${sequences > 1 ? ` with ${sequences} sequences` : ""}`);
|
|
73
|
+
}
|
|
74
|
+
throw new Error(`Invalid context size: "${contextSize}"`);
|
|
75
|
+
}
|
|
76
|
+
//# sourceMappingURL=resolveContextContextSizeOption.js.map
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"resolveContextContextSizeOption.js","sourceRoot":"","sources":["../../../../src/gguf/insights/utils/resolveContextContextSizeOption.ts"],"names":[],"mappings":"AAGA,OAAO,EAAC,mCAAmC,EAAC,MAAM,oBAAoB,CAAC;AACvE,OAAO,EAAC,0BAA0B,EAAE,0BAA0B,EAAC,MAAM,iDAAiD,CAAC;AAEvH,MAAM,UAAU,+BAA+B,CAAC,EAC5C,WAAW,EAAE,SAAS,EAAE,SAAS,EAAE,iBAAiB,EAAE,cAAc,EAAE,qBAAqB,EAAE,YAAY,EAAE,QAAQ,EACnH,wBAAwB,GAAG,KAAK,EAAE,kBAAkB,GAAG,KAAK,EAY/D;IACG,IAAI,WAAW,IAAI,IAAI;QACnB,WAAW,GAAG,MAAM,CAAC;IAEzB,IAAI,OAAO,WAAW,KAAK,QAAQ,EAAE;QACjC,MAAM,mBAAmB,GAAG,IAAI,CAAC,GAAG,CAAC,CAAC,EAAE,IAAI,CAAC,KAAK,CAAC,WAAW,CAAC,CAAC,CAAC;QAEjE,IAAI,wBAAwB;YACxB,OAAO,mBAAmB,CAAC;QAE/B,MAAM,SAAS,GAAG,YAAY,EAAE,CAAC;QACjC,MAAM,WAAW,GAAG,iBAAiB,CAAC,mCAAmC,CAAC;YACtE,WAAW,EAAE,mBAAmB;YAChC,SAAS,EAAE,SAAS,IAAI,0BAA0B,CAAC,EAAC,WAAW,EAAE,mBAAmB,EAAE,SAAS,EAAC,CAAC;YACjG,cAAc,EAAE,cAAc;YAC9B,SAAS;YACT,kBAAkB;SACrB,CAAC,CAAC,OAAO,CAAC;QAEX,IAAI,WAAW,GAAG,SAAS,CAAC,IAAI;YAC5B,MAAM,IAAI,KAAK,CAAC,uBAAuB,mBAAmB,GAAG,SAAS,GAAG,CAAC,CAAC,CAAC,CAAC,SAAS,SAAS,YAAY,CAAC,CAAC,CAAC,EAAE,sCAAsC,CAAC,CAAC;QAE5J,OAAO,mBAAmB,CAAC;KAC9B;SAAM,IAAI,WAAW,KAAK,MAAM,IAAI,OAAO,WAAW,KAAK,QAAQ,EAAE;QAClE,IAAI,QAAQ,KAAK,KAAK;YAClB,OAAO,qBAAqB,CAAC;QAEjC,MAAM,SAAS,GAAG,YAAY,EAAE,CAAC;QAEjC,IAAI,SAAS,CAAC,KAAK,KAAK,CAAC;YACrB,OAAO,qBAAqB,CAAC;QAEjC,MAAM,QAAQ,GAAG,SAAS,CAAC,IAAI,CAAC;QAEhC,MAAM,cAAc,GAAG,WAAW,KAAK,MAAM;YACzC,CAAC,CAAC,0BAA0B,CAAC,EAAC,gBAAgB,EAAE,qBAAqB,EAAC,CAAC;YACvE,CAAC,CAAC,IAAI,CAAC,GAAG,CACN,WAAW,CAAC,GAAG,IAAI,0BAA0B,CAAC,EAAC,gBAAgB,EAAE,qBAAqB,EAAC,CAAC,EACxF,0BAA0B,CAAC,EAAC,gBAAgB,EAAE,qBAAqB,EAAC,CAAC,CACxE,CAAC;QAEN,MAAM,cAAc,GAAG,WAAW,KAAK,MAAM;YACzC,CAAC,CAAC,mCAAmC;YACrC,CAAC,CAAC,IAAI,CAAC,GAAG,CACN,WAAW,CAAC,GAAG,IAAI,mCAAmC,EACtD,mCAAmC,CACtC,CAAC;QAEN,IAAI,4BAA4B,GAAkB,IAAI,CAAC;QACvD,IAAI,IAAI,GAAG,CAAC,IAAI,CAAC,GAAG,CAAC,CAAC,EAAE,IAAI,CAAC,KAAK,CAAC,CAAC,cAAc,GAAG,cAAc,CAAC,GAAG,CAAC,CAAC,CAAC,CAAC;QAC3E,KAAK,IAAI,eAAe,GAAG,cAAc,EAAE,eAAe,IAAI,cAAc,IAAI,eAAe,IAAI,cAAc,GAAG;YAChH,MAAM,WAAW,GAAG,iBAAiB,CAAC,mCAAmC,CAAC;gBACtE,WAAW,EAAE,eAAe;gBAC5B,SAAS,EAAE,SAAS,IAAI,0BAA0B,CAAC,EAAC,WAAW,EAAE,eAAe,EAAE,SAAS,EAAC,CAAC;gBAC7F,cAAc,EAAE,cAAc;gBAC9B,SAAS;gBACT,kBAAkB;aACrB,CAAC,CAAC,OAAO,CAAC;YAEX,IAAI,WAAW,IAAI,QAAQ,EAAE;gBACzB,IAAI,4BAA4B,IAAI,IAAI,IAAI,eAAe,GAAG,4BAA4B,EAAE;oBACxF,4BAA4B,GAAG,eAAe,CAAC;oBAE/C,IAAI,IAAI,KAAK,CAAC,CAAC;wBACX,MAAM;yBACL,IAAI,IAAI,GAAG,CAAC;wBACb,IAAI,GAAG,IAAI,CAAC,GAAG,CAAC,CAAC,EAAE,IAAI,CAAC,KAAK,CAAC,CAAC,IAAI,GAAG,CAAC,CAAC,CAAC,CAAC;iBACjD;aACJ;iBAAM,IAAI,IAAI,GAAG,CAAC;gBACf,IAAI,GAAG,CAAC,IAAI,CAAC,GAAG,CAAC,CAAC,EAAE,IAAI,CAAC,KAAK,CAAC,IAAI,GAAG,CAAC,CAAC,CAAC,CAAC;YAE9C,IAAI,eAAe,IAAI,cAAc,IAAI,IAAI,KAAK,CAAC,CAAC;gBAChD,MAAM;YAEV,eAAe,IAAI,IAAI,CAAC;YACxB,IAAI,eAAe,GAAG,cAAc,EAAE;gBAClC,eAAe,GAAG,cAAc,CAAC;gBACjC,IAAI,GAAG,IAAI,CAAC,GAAG,CAAC,CAAC,EAAE,IAAI,CAAC,KAAK,CAAC,IAAI,CAAC,GAAG,CAAC,IAAI,CAAC,GAAG,CAAC,CAAC,CAAC,CAAC;aACtD;iBAAM,IAAI,eAAe,GAAG,cAAc,EAAE;gBACzC,eAAe,GAAG,cAAc,CAAC;gBACjC,IAAI,GAAG,CAAC,IAAI,CAAC,GAAG,CAAC,CAAC,EAAE,IAAI,CAAC,KAAK,CAAC,IAAI,CAAC,GAAG,CAAC,IAAI,CAAC,GAAG,CAAC,CAAC,CAAC,CAAC;aACvD;SACJ;QAED,IAAI,4BAA4B,IAAI,IAAI;YACpC,OAAO,4BAA4B,CAAC;QAExC,IAAI,wBAAwB;YACxB,OAAO,cAAc,CAAC;QAE1B,MAAM,IAAI,KAAK,CAAC,8DAA8D,cAAc,GAAG,SAAS,GAAG,CAAC,CAAC,CAAC,CAAC,SAAS,SAAS,YAAY,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC;KACzJ;IAED,MAAM,IAAI,KAAK,CAAC,0BAA0B,WAAW,GAAG,CAAC,CAAC;AAC9D,CAAC"}
|
|
@@ -0,0 +1,14 @@
|
|
|
1
|
+
import { LlamaModelOptions } from "../../../evaluator/LlamaModel.js";
|
|
2
|
+
import { BuildGpu } from "../../../bindings/types.js";
|
|
3
|
+
import type { GgufInsights } from "../GgufInsights.js";
|
|
4
|
+
export declare function resolveModelGpuLayersOption(gpuLayers: LlamaModelOptions["gpuLayers"], { ggufInsights, ignoreMemorySafetyChecks, getVramState, llamaVramPaddingSize, llamaGpu, llamaSupportsGpuOffloading }: {
|
|
5
|
+
ggufInsights: GgufInsights;
|
|
6
|
+
ignoreMemorySafetyChecks?: boolean;
|
|
7
|
+
getVramState(): {
|
|
8
|
+
total: number;
|
|
9
|
+
free: number;
|
|
10
|
+
};
|
|
11
|
+
llamaVramPaddingSize: number;
|
|
12
|
+
llamaGpu: BuildGpu;
|
|
13
|
+
llamaSupportsGpuOffloading: boolean;
|
|
14
|
+
}): number;
|
|
@@ -0,0 +1,177 @@
|
|
|
1
|
+
import { InsufficientMemoryError } from "../../../utils/InsufficientMemoryError.js";
|
|
2
|
+
import { findBestOption } from "../../../utils/findBestOption.js";
|
|
3
|
+
import { getDefaultContextBatchSize, getDefaultModelContextSize } from "../../../evaluator/LlamaContext/LlamaContext.js";
|
|
4
|
+
import { minAllowedContextSizeInCalculations } from "../../../config.js";
|
|
5
|
+
import { scoreLevels } from "./scoreLevels.js";
|
|
6
|
+
const fitContextExtraMemoryPaddingPercentage = 0.5;
|
|
7
|
+
export function resolveModelGpuLayersOption(gpuLayers, { ggufInsights, ignoreMemorySafetyChecks = false, getVramState, llamaVramPaddingSize, llamaGpu, llamaSupportsGpuOffloading }) {
|
|
8
|
+
if (gpuLayers == null)
|
|
9
|
+
gpuLayers = "auto";
|
|
10
|
+
if (!llamaSupportsGpuOffloading)
|
|
11
|
+
return 0;
|
|
12
|
+
if (gpuLayers === "max" || typeof gpuLayers === "number") {
|
|
13
|
+
const resolvedGpuLayers = typeof gpuLayers === "number"
|
|
14
|
+
? Math.max(0, Math.min(ggufInsights.totalLayers, gpuLayers))
|
|
15
|
+
: ggufInsights.totalLayers;
|
|
16
|
+
if (ignoreMemorySafetyChecks)
|
|
17
|
+
return resolvedGpuLayers;
|
|
18
|
+
const vramState = getVramState();
|
|
19
|
+
const maxLayersRequirements = getVramRequiredForGpuLayers({
|
|
20
|
+
gpuLayers: resolvedGpuLayers,
|
|
21
|
+
ggufInsights,
|
|
22
|
+
currentVram: vramState.free
|
|
23
|
+
});
|
|
24
|
+
if (maxLayersRequirements == null)
|
|
25
|
+
throw new InsufficientMemoryError("Not enough VRAM to fit the model with the specified settings");
|
|
26
|
+
return resolvedGpuLayers;
|
|
27
|
+
}
|
|
28
|
+
else if (gpuLayers === "auto" || typeof gpuLayers === "object") {
|
|
29
|
+
if (llamaGpu === false)
|
|
30
|
+
return 0;
|
|
31
|
+
const vramState = getVramState();
|
|
32
|
+
if (vramState.total === 0)
|
|
33
|
+
return 0;
|
|
34
|
+
let freeVram = vramState.free;
|
|
35
|
+
if (typeof gpuLayers === "object" && gpuLayers.fitContext?.contextSize != null) {
|
|
36
|
+
freeVram -= llamaVramPaddingSize * fitContextExtraMemoryPaddingPercentage;
|
|
37
|
+
if (freeVram < 0)
|
|
38
|
+
freeVram = 0;
|
|
39
|
+
}
|
|
40
|
+
const bestGpuLayersOption = getBestGpuLayersForFreeVram({
|
|
41
|
+
ggufInsights,
|
|
42
|
+
freeVram,
|
|
43
|
+
fitContext: typeof gpuLayers === "object"
|
|
44
|
+
? gpuLayers.fitContext
|
|
45
|
+
: undefined,
|
|
46
|
+
minGpuLayers: typeof gpuLayers === "object"
|
|
47
|
+
? gpuLayers.min
|
|
48
|
+
: undefined,
|
|
49
|
+
maxGpuLayers: typeof gpuLayers === "object"
|
|
50
|
+
? gpuLayers.max
|
|
51
|
+
: undefined
|
|
52
|
+
});
|
|
53
|
+
const hasGpuLayersRequirements = typeof gpuLayers === "object" &&
|
|
54
|
+
(gpuLayers.min != null || gpuLayers.max != null || gpuLayers.fitContext?.contextSize != null);
|
|
55
|
+
if (!ignoreMemorySafetyChecks && bestGpuLayersOption == null && hasGpuLayersRequirements)
|
|
56
|
+
throw new InsufficientMemoryError("Not enough VRAM to fit the model with the specified settings");
|
|
57
|
+
return bestGpuLayersOption ?? 0;
|
|
58
|
+
}
|
|
59
|
+
throw new Error(`Invalid gpuLayers value: ${gpuLayers}`);
|
|
60
|
+
}
|
|
61
|
+
function getBestGpuLayersForFreeVram({ ggufInsights, freeVram, fitContext, minGpuLayers, maxGpuLayers }) {
|
|
62
|
+
return findBestOption({
|
|
63
|
+
*generator() {
|
|
64
|
+
const minLayers = Math.floor(Math.max(0, minGpuLayers ?? 0));
|
|
65
|
+
const maxLayers = Math.floor(Math.min(ggufInsights.totalLayers, maxGpuLayers ?? ggufInsights.totalLayers));
|
|
66
|
+
for (let layers = maxLayers; layers >= minLayers; layers--) {
|
|
67
|
+
yield {
|
|
68
|
+
gpuLayers: layers
|
|
69
|
+
};
|
|
70
|
+
}
|
|
71
|
+
},
|
|
72
|
+
score(option) {
|
|
73
|
+
const layersRequirements = getVramRequiredForGpuLayers({
|
|
74
|
+
gpuLayers: option.gpuLayers,
|
|
75
|
+
ggufInsights,
|
|
76
|
+
currentVram: freeVram,
|
|
77
|
+
fitContext
|
|
78
|
+
});
|
|
79
|
+
if (layersRequirements == null)
|
|
80
|
+
return null;
|
|
81
|
+
return scoreGpuLayersAndContextCombination({ gpuLayers: option.gpuLayers, contextSize: layersRequirements.contextSize }, {
|
|
82
|
+
totalGpuLayers: ggufInsights.totalLayers,
|
|
83
|
+
trainContextSize: getDefaultModelContextSize({ trainContextSize: ggufInsights.trainContextSize })
|
|
84
|
+
});
|
|
85
|
+
}
|
|
86
|
+
})?.gpuLayers ?? null;
|
|
87
|
+
}
|
|
88
|
+
function scoreGpuLayersAndContextCombination({ gpuLayers, contextSize }, { totalGpuLayers, trainContextSize }) {
|
|
89
|
+
function scoreGpuLayers() {
|
|
90
|
+
return scoreLevels(gpuLayers, [{
|
|
91
|
+
start: 0,
|
|
92
|
+
points: 4
|
|
93
|
+
}, {
|
|
94
|
+
start: 1,
|
|
95
|
+
points: 26
|
|
96
|
+
}, {
|
|
97
|
+
start: totalGpuLayers,
|
|
98
|
+
points: 14,
|
|
99
|
+
end: totalGpuLayers
|
|
100
|
+
}]);
|
|
101
|
+
}
|
|
102
|
+
function scoreContextSize() {
|
|
103
|
+
const gpuLayersPercentage = gpuLayers / totalGpuLayers;
|
|
104
|
+
return scoreLevels(contextSize, [{
|
|
105
|
+
start: 0,
|
|
106
|
+
points: 2
|
|
107
|
+
}, {
|
|
108
|
+
start: 1024,
|
|
109
|
+
points: 4
|
|
110
|
+
}, {
|
|
111
|
+
start: 2048,
|
|
112
|
+
points: gpuLayersPercentage < 0.1 ? 1 : 8
|
|
113
|
+
}, {
|
|
114
|
+
start: 4096,
|
|
115
|
+
points: gpuLayersPercentage < 0.3 ? 4 : 16
|
|
116
|
+
}, {
|
|
117
|
+
start: 8192,
|
|
118
|
+
points: gpuLayersPercentage < 0.6 ? 1 : 8,
|
|
119
|
+
end: Math.max(trainContextSize, 16384)
|
|
120
|
+
}]);
|
|
121
|
+
}
|
|
122
|
+
return scoreGpuLayers() + scoreContextSize();
|
|
123
|
+
}
|
|
124
|
+
function getVramRequiredForGpuLayers({ gpuLayers, ggufInsights, currentVram, fitContext }) {
|
|
125
|
+
const modelVram = ggufInsights.estimateModelResourceRequirements({ gpuLayers }).gpuVram;
|
|
126
|
+
if (modelVram > currentVram)
|
|
127
|
+
return null;
|
|
128
|
+
if (fitContext != null && fitContext.contextSize != null) {
|
|
129
|
+
const contextVram = ggufInsights.estimateContextResourceRequirements({
|
|
130
|
+
contextSize: fitContext.contextSize,
|
|
131
|
+
batchSize: getDefaultContextBatchSize({ contextSize: fitContext.contextSize, sequences: 1 }),
|
|
132
|
+
modelGpuLayers: gpuLayers,
|
|
133
|
+
sequences: 1,
|
|
134
|
+
isEmbeddingContext: fitContext.embeddingContext ?? false
|
|
135
|
+
}).gpuVram;
|
|
136
|
+
const totalVram = modelVram + contextVram;
|
|
137
|
+
if (totalVram > currentVram)
|
|
138
|
+
return null;
|
|
139
|
+
return {
|
|
140
|
+
contextSize: fitContext.contextSize,
|
|
141
|
+
contextVram,
|
|
142
|
+
totalVram
|
|
143
|
+
};
|
|
144
|
+
}
|
|
145
|
+
const maxContext = findMaxPossibleContextSizeForVram({
|
|
146
|
+
gpuLayers,
|
|
147
|
+
ggufInsights,
|
|
148
|
+
vram: currentVram - modelVram,
|
|
149
|
+
isEmbeddingContext: fitContext?.embeddingContext ?? false
|
|
150
|
+
});
|
|
151
|
+
if (maxContext == null || modelVram + maxContext.vram > currentVram)
|
|
152
|
+
return null;
|
|
153
|
+
return {
|
|
154
|
+
contextSize: maxContext.contextSize,
|
|
155
|
+
contextVram: maxContext.vram,
|
|
156
|
+
totalVram: modelVram + maxContext.vram
|
|
157
|
+
};
|
|
158
|
+
}
|
|
159
|
+
function findMaxPossibleContextSizeForVram({ gpuLayers, ggufInsights, vram, isEmbeddingContext }) {
|
|
160
|
+
const maxContextSize = getDefaultModelContextSize({ trainContextSize: ggufInsights.trainContextSize });
|
|
161
|
+
for (let contextSize = maxContextSize; contextSize >= minAllowedContextSizeInCalculations; contextSize--) {
|
|
162
|
+
const contextVram = ggufInsights.estimateContextResourceRequirements({
|
|
163
|
+
contextSize,
|
|
164
|
+
batchSize: getDefaultContextBatchSize({ contextSize, sequences: 1 }),
|
|
165
|
+
modelGpuLayers: gpuLayers,
|
|
166
|
+
sequences: 1,
|
|
167
|
+
isEmbeddingContext
|
|
168
|
+
}).gpuVram;
|
|
169
|
+
if (contextVram <= vram)
|
|
170
|
+
return {
|
|
171
|
+
contextSize,
|
|
172
|
+
vram: contextVram
|
|
173
|
+
};
|
|
174
|
+
}
|
|
175
|
+
return null;
|
|
176
|
+
}
|
|
177
|
+
//# sourceMappingURL=resolveModelGpuLayersOption.js.map
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"resolveModelGpuLayersOption.js","sourceRoot":"","sources":["../../../../src/gguf/insights/utils/resolveModelGpuLayersOption.ts"],"names":[],"mappings":"AAEA,OAAO,EAAC,uBAAuB,EAAC,MAAM,2CAA2C,CAAC;AAClF,OAAO,EAAC,cAAc,EAAC,MAAM,kCAAkC,CAAC;AAChE,OAAO,EAAC,0BAA0B,EAAE,0BAA0B,EAAC,MAAM,iDAAiD,CAAC;AACvH,OAAO,EAAC,mCAAmC,EAAC,MAAM,oBAAoB,CAAC;AACvE,OAAO,EAAC,WAAW,EAAC,MAAM,kBAAkB,CAAC;AAG7C,MAAM,sCAAsC,GAAG,GAAG,CAAC;AAEnD,MAAM,UAAU,2BAA2B,CAAC,SAAyC,EAAE,EACnF,YAAY,EAAE,wBAAwB,GAAG,KAAK,EAAE,YAAY,EAAE,oBAAoB,EAClF,QAAQ,EAAE,0BAA0B,EAIvC;IACG,IAAI,SAAS,IAAI,IAAI;QACjB,SAAS,GAAG,MAAM,CAAC;IAEvB,IAAI,CAAC,0BAA0B;QAC3B,OAAO,CAAC,CAAC;IAEb,IAAI,SAAS,KAAK,KAAK,IAAI,OAAO,SAAS,KAAK,QAAQ,EAAE;QACtD,MAAM,iBAAiB,GAAG,OAAO,SAAS,KAAK,QAAQ;YACnD,CAAC,CAAC,IAAI,CAAC,GAAG,CAAC,CAAC,EAAE,IAAI,CAAC,GAAG,CAAC,YAAY,CAAC,WAAW,EAAE,SAAS,CAAC,CAAC;YAC5D,CAAC,CAAC,YAAY,CAAC,WAAW,CAAC;QAE/B,IAAI,wBAAwB;YACxB,OAAO,iBAAiB,CAAC;QAE7B,MAAM,SAAS,GAAG,YAAY,EAAE,CAAC;QACjC,MAAM,qBAAqB,GAAG,2BAA2B,CAAC;YACtD,SAAS,EAAE,iBAAiB;YAC5B,YAAY;YACZ,WAAW,EAAE,SAAS,CAAC,IAAI;SAC9B,CAAC,CAAC;QAEH,IAAI,qBAAqB,IAAI,IAAI;YAC7B,MAAM,IAAI,uBAAuB,CAAC,8DAA8D,CAAC,CAAC;QAEtG,OAAO,iBAAiB,CAAC;KAC5B;SAAM,IAAI,SAAS,KAAK,MAAM,IAAI,OAAO,SAAS,KAAK,QAAQ,EAAE;QAC9D,IAAI,QAAQ,KAAK,KAAK;YAClB,OAAO,CAAC,CAAC;QAEb,MAAM,SAAS,GAAG,YAAY,EAAE,CAAC;QACjC,IAAI,SAAS,CAAC,KAAK,KAAK,CAAC;YACrB,OAAO,CAAC,CAAC;QAEb,IAAI,QAAQ,GAAG,SAAS,CAAC,IAAI,CAAC;QAC9B,IAAI,OAAO,SAAS,KAAK,QAAQ,IAAI,SAAS,CAAC,UAAU,EAAE,WAAW,IAAI,IAAI,EAAE;YAC5E,QAAQ,IAAI,oBAAoB,GAAG,sCAAsC,CAAC;YAE1E,IAAI,QAAQ,GAAG,CAAC;gBACZ,QAAQ,GAAG,CAAC,CAAC;SACpB;QAED,MAAM,mBAAmB,GAAG,2BAA2B,CAAC;YACpD,YAAY;YACZ,QAAQ;YACR,UAAU,EAAE,OAAO,SAAS,KAAK,QAAQ;gBACrC,CAAC,CAAC,SAAS,CAAC,UAAU;gBACtB,CAAC,CAAC,SAAS;YACf,YAAY,EAAE,OAAO,SAAS,KAAK,QAAQ;gBACvC,CAAC,CAAC,SAAS,CAAC,GAAG;gBACf,CAAC,CAAC,SAAS;YACf,YAAY,EAAE,OAAO,SAAS,KAAK,QAAQ;gBACvC,CAAC,CAAC,SAAS,CAAC,GAAG;gBACf,CAAC,CAAC,SAAS;SAClB,CAAC,CAAC;QAEH,MAAM,wBAAwB,GAAG,OAAO,SAAS,KAAK,QAAQ;YAC1D,CAAC,SAAS,CAAC,GAAG,IAAI,IAAI,IAAI,SAAS,CAAC,GAAG,IAAI,IAAI,IAAI,SAAS,CAAC,UAAU,EAAE,WAAW,IAAI,IAAI,CAAC,CAAC;QAElG,IAAI,CAAC,wBAAwB,IAAI,mBAAmB,IAAI,IAAI,IAAI,wBAAwB;YACpF,MAAM,IAAI,uBAAuB,CAAC,8DAA8D,CAAC,CAAC;QAEtG,OAAO,mBAAmB,IAAI,CAAC,CAAC;KACnC;IAED,MAAM,IAAI,KAAK,CAAC,4BAA4B,SAAS,EAAE,CAAC,CAAC;AAC7D,CAAC;AAED,SAAS,2BAA2B,CAAC,EACjC,YAAY,EACZ,QAAQ,EACR,UAAU,EACV,YAAY,EACZ,YAAY,EAOf;IACG,OAAO,cAAc,CAAC;QAClB,CAAC,SAAS;YACN,MAAM,SAAS,GAAG,IAAI,CAAC,KAAK,CAAC,IAAI,CAAC,GAAG,CAAC,CAAC,EAAE,YAAY,IAAI,CAAC,CAAC,CAAC,CAAC;YAC7D,MAAM,SAAS,GAAG,IAAI,CAAC,KAAK,CAAC,IAAI,CAAC,GAAG,CAAC,YAAY,CAAC,WAAW,EAAE,YAAY,IAAI,YAAY,CAAC,WAAW,CAAC,CAAC,CAAC;YAE3G,KAAK,IAAI,MAAM,GAAG,SAAS,EAAE,MAAM,IAAI,SAAS,EAAE,MAAM,EAAE,EAAE;gBACxD,MAAM;oBACF,SAAS,EAAE,MAAM;iBACpB,CAAC;aACL;QACL,CAAC;QACD,KAAK,CAAC,MAAM;YACR,MAAM,kBAAkB,GAAG,2BAA2B,CAAC;gBACnD,SAAS,EAAE,MAAM,CAAC,SAAS;gBAC3B,YAAY;gBACZ,WAAW,EAAE,QAAQ;gBACrB,UAAU;aACb,CAAC,CAAC;YAEH,IAAI,kBAAkB,IAAI,IAAI;gBAC1B,OAAO,IAAI,CAAC;YAEhB,OAAO,mCAAmC,CAAC,EAAC,SAAS,EAAE,MAAM,CAAC,SAAS,EAAE,WAAW,EAAE,kBAAkB,CAAC,WAAW,EAAC,EAAE;gBACnH,cAAc,EAAE,YAAY,CAAC,WAAW;gBACxC,gBAAgB,EAAE,0BAA0B,CAAC,EAAC,gBAAgB,EAAE,YAAY,CAAC,gBAAgB,EAAC,CAAC;aAClG,CAAC,CAAC;QACP,CAAC;KACJ,CAAC,EAAE,SAAS,IAAI,IAAI,CAAC;AAC1B,CAAC;AAED,SAAS,mCAAmC,CAAC,EAAC,SAAS,EAAE,WAAW,EAA2C,EAAE,EAC7G,cAAc,EAAE,gBAAgB,EAGnC;IACG,SAAS,cAAc;QACnB,OAAO,WAAW,CAAC,SAAS,EAAE,CAAC;gBAC3B,KAAK,EAAE,CAAC;gBACR,MAAM,EAAE,CAAC;aACZ,EAAE;gBACC,KAAK,EAAE,CAAC;gBACR,MAAM,EAAE,EAAE;aACb,EAAE;gBACC,KAAK,EAAE,cAAc;gBACrB,MAAM,EAAE,EAAE;gBACV,GAAG,EAAE,cAAc;aACtB,CAAC,CAAC,CAAC;IACR,CAAC;IAED,SAAS,gBAAgB;QACrB,MAAM,mBAAmB,GAAG,SAAS,GAAG,cAAc,CAAC;QAEvD,OAAO,WAAW,CAAC,WAAW,EAAE,CAAC;gBAC7B,KAAK,EAAE,CAAC;gBACR,MAAM,EAAE,CAAC;aACZ,EAAE;gBACC,KAAK,EAAE,IAAI;gBACX,MAAM,EAAE,CAAC;aACZ,EAAE;gBACC,KAAK,EAAE,IAAI;gBACX,MAAM,EAAE,mBAAmB,GAAG,GAAG,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC;aAC5C,EAAE;gBACC,KAAK,EAAE,IAAI;gBACX,MAAM,EAAE,mBAAmB,GAAG,GAAG,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,EAAE;aAC7C,EAAE;gBACC,KAAK,EAAE,IAAI;gBACX,MAAM,EAAE,mBAAmB,GAAG,GAAG,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC;gBACzC,GAAG,EAAE,IAAI,CAAC,GAAG,CAAC,gBAAgB,EAAE,KAAK,CAAC;aACzC,CAAC,CAAC,CAAC;IACR,CAAC;IAED,OAAO,cAAc,EAAE,GAAG,gBAAgB,EAAE,CAAC;AACjD,CAAC;AAED,SAAS,2BAA2B,CAAC,EACjC,SAAS,EAAE,YAAY,EAAE,WAAW,EAAE,UAAU,EAGnD;IACG,MAAM,SAAS,GAAG,YAAY,CAAC,iCAAiC,CAAC,EAAC,SAAS,EAAC,CAAC,CAAC,OAAO,CAAC;IAEtF,IAAI,SAAS,GAAG,WAAW;QACvB,OAAO,IAAI,CAAC;IAEhB,IAAI,UAAU,IAAI,IAAI,IAAI,UAAU,CAAC,WAAW,IAAI,IAAI,EAAE;QACtD,MAAM,WAAW,GAAG,YAAY,CAAC,mCAAmC,CAAC;YACjE,WAAW,EAAE,UAAU,CAAC,WAAW;YACnC,SAAS,EAAE,0BAA0B,CAAC,EAAC,WAAW,EAAE,UAAU,CAAC,WAAW,EAAE,SAAS,EAAE,CAAC,EAAC,CAAC;YAC1F,cAAc,EAAE,SAAS;YACzB,SAAS,EAAE,CAAC;YACZ,kBAAkB,EAAE,UAAU,CAAC,gBAAgB,IAAI,KAAK;SAC3D,CAAC,CAAC,OAAO,CAAC;QAEX,MAAM,SAAS,GAAG,SAAS,GAAG,WAAW,CAAC;QAC1C,IAAI,SAAS,GAAG,WAAW;YACvB,OAAO,IAAI,CAAC;QAEhB,OAAO;YACH,WAAW,EAAE,UAAU,CAAC,WAAW;YACnC,WAAW;YACX,SAAS;SACZ,CAAC;KACL;IAED,MAAM,UAAU,GAAG,iCAAiC,CAAC;QACjD,SAAS;QACT,YAAY;QACZ,IAAI,EAAE,WAAW,GAAG,SAAS;QAC7B,kBAAkB,EAAE,UAAU,EAAE,gBAAgB,IAAI,KAAK;KAC5D,CAAC,CAAC;IAEH,IAAI,UAAU,IAAI,IAAI,IAAI,SAAS,GAAG,UAAU,CAAC,IAAI,GAAG,WAAW;QAC/D,OAAO,IAAI,CAAC;IAEhB,OAAO;QACH,WAAW,EAAE,UAAU,CAAC,WAAW;QACnC,WAAW,EAAE,UAAU,CAAC,IAAI;QAC5B,SAAS,EAAE,SAAS,GAAG,UAAU,CAAC,IAAI;KACzC,CAAC;AACN,CAAC;AAED,SAAS,iCAAiC,CAAC,EAAC,SAAS,EAAE,YAAY,EAAE,IAAI,EAAE,kBAAkB,EAE5F;IACG,MAAM,cAAc,GAAG,0BAA0B,CAAC,EAAC,gBAAgB,EAAE,YAAY,CAAC,gBAAgB,EAAC,CAAC,CAAC;IAErG,KAAK,IAAI,WAAW,GAAG,cAAc,EAAE,WAAW,IAAI,mCAAmC,EAAE,WAAW,EAAE,EAAE;QACtG,MAAM,WAAW,GAAG,YAAY,CAAC,mCAAmC,CAAC;YACjE,WAAW;YACX,SAAS,EAAE,0BAA0B,CAAC,EAAC,WAAW,EAAE,SAAS,EAAE,CAAC,EAAC,CAAC;YAClE,cAAc,EAAE,SAAS;YACzB,SAAS,EAAE,CAAC;YACZ,kBAAkB;SACrB,CAAC,CAAC,OAAO,CAAC;QAEX,IAAI,WAAW,IAAI,IAAI;YACnB,OAAO;gBACH,WAAW;gBACX,IAAI,EAAE,WAAW;aACpB,CAAC;KACT;IAED,OAAO,IAAI,CAAC;AAChB,CAAC"}
|
|
@@ -0,0 +1,16 @@
|
|
|
1
|
+
export function scoreLevels(num, levels) {
|
|
2
|
+
let res = 0;
|
|
3
|
+
for (let i = 0; i < levels.length; i++) {
|
|
4
|
+
const level = levels[i];
|
|
5
|
+
const start = level.start;
|
|
6
|
+
const end = level.end ?? levels[i + 1]?.start ?? Math.max(start, num);
|
|
7
|
+
if (num < start)
|
|
8
|
+
break;
|
|
9
|
+
else if (num >= end)
|
|
10
|
+
res += level.points;
|
|
11
|
+
else
|
|
12
|
+
res += level.points * ((num - start) / (end - start));
|
|
13
|
+
}
|
|
14
|
+
return res;
|
|
15
|
+
}
|
|
16
|
+
//# sourceMappingURL=scoreLevels.js.map
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"scoreLevels.js","sourceRoot":"","sources":["../../../../src/gguf/insights/utils/scoreLevels.ts"],"names":[],"mappings":"AAAA,MAAM,UAAU,WAAW,CAAC,GAAW,EAAE,MAAyD;IAC9F,IAAI,GAAG,GAAG,CAAC,CAAC;IAEZ,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,MAAM,CAAC,MAAM,EAAE,CAAC,EAAE,EAAE;QACpC,MAAM,KAAK,GAAG,MAAM,CAAC,CAAC,CAAC,CAAC;QACxB,MAAM,KAAK,GAAG,KAAK,CAAC,KAAK,CAAC;QAC1B,MAAM,GAAG,GAAG,KAAK,CAAC,GAAG,IAAI,MAAM,CAAC,CAAC,GAAG,CAAC,CAAC,EAAE,KAAK,IAAI,IAAI,CAAC,GAAG,CAAC,KAAK,EAAE,GAAG,CAAC,CAAC;QAEtE,IAAI,GAAG,GAAG,KAAK;YACX,MAAM;aACL,IAAI,GAAG,IAAI,GAAG;YACf,GAAG,IAAI,KAAK,CAAC,MAAM,CAAC;;YAEpB,GAAG,IAAI,KAAK,CAAC,MAAM,GAAG,CAAC,CAAC,GAAG,GAAG,KAAK,CAAC,GAAG,CAAC,GAAG,GAAG,KAAK,CAAC,CAAC,CAAC;KAC7D;IAED,OAAO,GAAG,CAAC;AACf,CAAC"}
|
|
@@ -0,0 +1,19 @@
|
|
|
1
|
+
import { GgufReadOffset } from "../utils/GgufReadOffset.js";
|
|
2
|
+
import { GgufValueType, GgufVersionParserOptions, GgufVersionParserResult, MetadataKeyValueRecord, MetadataValue } from "../types/GgufFileInfoTypes.js";
|
|
3
|
+
export declare class GgufV2Parser {
|
|
4
|
+
private readonly _fileReader;
|
|
5
|
+
private readonly _shouldReadTensorInfo;
|
|
6
|
+
private readonly _ignoreKeys;
|
|
7
|
+
private readonly _readOffset;
|
|
8
|
+
private readonly _logWarnings;
|
|
9
|
+
constructor({ fileReader, readTensorInfo, ignoreKeys, readOffset, logWarnings }: GgufVersionParserOptions);
|
|
10
|
+
parse(): Promise<GgufVersionParserResult>;
|
|
11
|
+
protected _readGgufValue(type: GgufValueType, offset: number | GgufReadOffset): Promise<MetadataValue>;
|
|
12
|
+
protected _readStringValue(offset: number | GgufReadOffset): Promise<string>;
|
|
13
|
+
protected _readRawHeader(readOffset: GgufReadOffset): Promise<{
|
|
14
|
+
tensorCount: number | bigint;
|
|
15
|
+
metadata: MetadataKeyValueRecord;
|
|
16
|
+
headerSize: number;
|
|
17
|
+
}>;
|
|
18
|
+
private _readTensorInfo;
|
|
19
|
+
}
|
|
@@ -0,0 +1,115 @@
|
|
|
1
|
+
import { GgufFileReader, valueTypeToBytesToRead } from "../fileReaders/GgufFileReader.js";
|
|
2
|
+
import { GgufReadOffset } from "../utils/GgufReadOffset.js";
|
|
3
|
+
import { UnsupportedGgufValueTypeError } from "../errors/UnsupportedGgufValueTypeError.js";
|
|
4
|
+
import { GgufValueType } from "../types/GgufFileInfoTypes.js";
|
|
5
|
+
import { convertMetadataKeyValueRecordToNestedObject } from "../utils/convertMetadataKeyValueRecordToNestedObject.js";
|
|
6
|
+
export class GgufV2Parser {
|
|
7
|
+
_fileReader;
|
|
8
|
+
_shouldReadTensorInfo;
|
|
9
|
+
_ignoreKeys;
|
|
10
|
+
_readOffset;
|
|
11
|
+
_logWarnings;
|
|
12
|
+
constructor({ fileReader, readTensorInfo = true, ignoreKeys = [], readOffset, logWarnings }) {
|
|
13
|
+
this._fileReader = fileReader;
|
|
14
|
+
this._shouldReadTensorInfo = readTensorInfo;
|
|
15
|
+
this._ignoreKeys = ignoreKeys;
|
|
16
|
+
this._readOffset = readOffset;
|
|
17
|
+
this._logWarnings = logWarnings;
|
|
18
|
+
}
|
|
19
|
+
async parse() {
|
|
20
|
+
const readOffset = this._readOffset;
|
|
21
|
+
const initialOffset = readOffset.offset;
|
|
22
|
+
const headerReadResult = await this._readRawHeader(readOffset);
|
|
23
|
+
const tensorReadResult = this._shouldReadTensorInfo
|
|
24
|
+
? await this._readTensorInfo(headerReadResult.tensorCount, readOffset)
|
|
25
|
+
: null;
|
|
26
|
+
const metadata = convertMetadataKeyValueRecordToNestedObject(headerReadResult.metadata, {
|
|
27
|
+
logOverrideWarnings: this._logWarnings,
|
|
28
|
+
ignoreKeys: this._ignoreKeys
|
|
29
|
+
});
|
|
30
|
+
return {
|
|
31
|
+
tensorCount: headerReadResult.tensorCount,
|
|
32
|
+
metadata: metadata,
|
|
33
|
+
tensorInfo: tensorReadResult?.tensorInfo,
|
|
34
|
+
metadataSize: headerReadResult.headerSize + initialOffset,
|
|
35
|
+
tensorInfoSize: tensorReadResult?.tensorInfoSize
|
|
36
|
+
};
|
|
37
|
+
}
|
|
38
|
+
async _readGgufValue(type, offset) {
|
|
39
|
+
const readOffset = GgufReadOffset.resolveReadOffset(offset);
|
|
40
|
+
switch (type) {
|
|
41
|
+
case GgufValueType.Uint8: return await this._fileReader.readUint8(readOffset);
|
|
42
|
+
case GgufValueType.Int8: return await this._fileReader.readInt8(readOffset);
|
|
43
|
+
case GgufValueType.Uint16: return await this._fileReader.readUint16(readOffset);
|
|
44
|
+
case GgufValueType.Int16: return await this._fileReader.readInt16(readOffset);
|
|
45
|
+
case GgufValueType.Uint32: return await this._fileReader.readUint32(readOffset);
|
|
46
|
+
case GgufValueType.Int32: return await this._fileReader.readInt32(readOffset);
|
|
47
|
+
case GgufValueType.Float32: return await this._fileReader.readFloat32(readOffset);
|
|
48
|
+
case GgufValueType.Bool: return await this._fileReader.readBool(readOffset);
|
|
49
|
+
case GgufValueType.String: return await this._readStringValue(readOffset);
|
|
50
|
+
case GgufValueType.Uint64: return await this._fileReader.readUint64(readOffset);
|
|
51
|
+
case GgufValueType.Int64: return await this._fileReader.readInt64(readOffset);
|
|
52
|
+
case GgufValueType.Float64: return await this._fileReader.readFloat64(readOffset);
|
|
53
|
+
}
|
|
54
|
+
if (type === GgufValueType.Array) {
|
|
55
|
+
const arrayType = await this._fileReader.readUint32(readOffset);
|
|
56
|
+
const arrayLength = await this._fileReader.readUint64(readOffset);
|
|
57
|
+
const arrayValues = [];
|
|
58
|
+
for (let i = 0; i < arrayLength; i++) {
|
|
59
|
+
const value = await this._readGgufValue(arrayType, readOffset);
|
|
60
|
+
arrayValues.push(value);
|
|
61
|
+
}
|
|
62
|
+
return arrayValues;
|
|
63
|
+
}
|
|
64
|
+
throw new UnsupportedGgufValueTypeError(type);
|
|
65
|
+
}
|
|
66
|
+
async _readStringValue(offset) {
|
|
67
|
+
const readOffset = GgufReadOffset.resolveReadOffset(offset);
|
|
68
|
+
const length = Number(await this._fileReader.readUint64(readOffset));
|
|
69
|
+
const readLength = valueTypeToBytesToRead.uint8 * length;
|
|
70
|
+
const stringBytes = await this._fileReader.readByteRange(readOffset, readLength);
|
|
71
|
+
return String.fromCharCode(...stringBytes);
|
|
72
|
+
}
|
|
73
|
+
async _readRawHeader(readOffset) {
|
|
74
|
+
const initialOffset = readOffset.offset;
|
|
75
|
+
const tensorCount = await this._fileReader.readUint64(readOffset);
|
|
76
|
+
const metadataKVCount = Number(await this._fileReader.readUint64(readOffset));
|
|
77
|
+
const metadata = {};
|
|
78
|
+
for (let i = 0; i < metadataKVCount; i++) {
|
|
79
|
+
const keyResult = await this._readStringValue(readOffset);
|
|
80
|
+
const valueType = await this._fileReader.readUint32(readOffset);
|
|
81
|
+
metadata[keyResult] = await this._readGgufValue(valueType, readOffset);
|
|
82
|
+
}
|
|
83
|
+
return {
|
|
84
|
+
tensorCount: GgufFileReader.castNumberIfSafe(tensorCount),
|
|
85
|
+
metadata: metadata,
|
|
86
|
+
headerSize: readOffset.offset - initialOffset
|
|
87
|
+
};
|
|
88
|
+
}
|
|
89
|
+
async _readTensorInfo(tensorCount, readOffset) {
|
|
90
|
+
const initialOffset = readOffset.offset;
|
|
91
|
+
const tensorInfo = [];
|
|
92
|
+
for (let i = 0n; i < BigInt(tensorCount); i++) {
|
|
93
|
+
const name = await this._readStringValue(readOffset);
|
|
94
|
+
const dimensionsNumber = await this._fileReader.readUint32(readOffset);
|
|
95
|
+
const dimensions = [];
|
|
96
|
+
for (let i = 0; i < dimensionsNumber; i++) {
|
|
97
|
+
const dimension = await this._fileReader.readUint64(readOffset);
|
|
98
|
+
dimensions.push(GgufFileReader.castNumberIfSafe(dimension));
|
|
99
|
+
}
|
|
100
|
+
const ggmlType = await this._fileReader.readUint32(readOffset);
|
|
101
|
+
const offset = await this._fileReader.readUint64(readOffset);
|
|
102
|
+
tensorInfo.push({
|
|
103
|
+
name,
|
|
104
|
+
dimensions,
|
|
105
|
+
ggmlType: ggmlType,
|
|
106
|
+
offset: GgufFileReader.castNumberIfSafe(offset)
|
|
107
|
+
});
|
|
108
|
+
}
|
|
109
|
+
return {
|
|
110
|
+
tensorInfo,
|
|
111
|
+
tensorInfoSize: readOffset.offset - initialOffset
|
|
112
|
+
};
|
|
113
|
+
}
|
|
114
|
+
}
|
|
115
|
+
//# sourceMappingURL=GgufV2Parser.js.map
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"GgufV2Parser.js","sourceRoot":"","sources":["../../../src/gguf/parser/GgufV2Parser.ts"],"names":[],"mappings":"AAAA,OAAO,EAAC,cAAc,EAAE,sBAAsB,EAAC,MAAM,kCAAkC,CAAC;AACxF,OAAO,EAAC,cAAc,EAAC,MAAM,4BAA4B,CAAC;AAC1D,OAAO,EAAC,6BAA6B,EAAC,MAAM,4CAA4C,CAAC;AACzF,OAAO,EACH,aAAa,EAChB,MAAM,+BAA+B,CAAC;AAGvC,OAAO,EAAC,2CAA2C,EAAC,MAAM,yDAAyD,CAAC;AAEpH,MAAM,OAAO,YAAY;IACJ,WAAW,CAAiB;IAC5B,qBAAqB,CAAU;IAC/B,WAAW,CAAW;IACtB,WAAW,CAAiB;IAC5B,YAAY,CAAU;IAEvC,YAAmB,EAAC,UAAU,EAAE,cAAc,GAAG,IAAI,EAAE,UAAU,GAAG,EAAE,EAAE,UAAU,EAAE,WAAW,EAA2B;QACtH,IAAI,CAAC,WAAW,GAAG,UAAU,CAAC;QAC9B,IAAI,CAAC,qBAAqB,GAAG,cAAc,CAAC;QAC5C,IAAI,CAAC,WAAW,GAAG,UAAU,CAAC;QAC9B,IAAI,CAAC,WAAW,GAAG,UAAU,CAAC;QAC9B,IAAI,CAAC,YAAY,GAAG,WAAW,CAAC;IACpC,CAAC;IAEM,KAAK,CAAC,KAAK;QACd,MAAM,UAAU,GAAG,IAAI,CAAC,WAAW,CAAC;QACpC,MAAM,aAAa,GAAG,UAAU,CAAC,MAAM,CAAC;QAExC,MAAM,gBAAgB,GAAG,MAAM,IAAI,CAAC,cAAc,CAAC,UAAU,CAAC,CAAC;QAC/D,MAAM,gBAAgB,GAAG,IAAI,CAAC,qBAAqB;YAC/C,CAAC,CAAC,MAAM,IAAI,CAAC,eAAe,CAAC,gBAAgB,CAAC,WAAW,EAAE,UAAU,CAAC;YACtE,CAAC,CAAC,IAAI,CAAC;QACX,MAAM,QAAQ,GAAG,2CAA2C,CAAC,gBAAgB,CAAC,QAAQ,EAAE;YACpF,mBAAmB,EAAE,IAAI,CAAC,YAAY;YACtC,UAAU,EAAE,IAAI,CAAC,WAAW;SAC/B,CAAC,CAAC;QAEH,OAAO;YACH,WAAW,EAAE,gBAAgB,CAAC,WAAW;YACzC,QAAQ,EAAE,QAA+B;YACzC,UAAU,EAAE,gBAAgB,EAAE,UAAU;YACxC,YAAY,EAAE,gBAAgB,CAAC,UAAU,GAAG,aAAa;YACzD,cAAc,EAAE,gBAAgB,EAAE,cAAc;SACnD,CAAC;IACN,CAAC;IAES,KAAK,CAAC,cAAc,CAAC,IAAmB,EAAE,MAA+B;QAC/E,MAAM,UAAU,GAAG,cAAc,CAAC,iBAAiB,CAAC,MAAM,CAAC,CAAC;QAE5D,QAAQ,IAAI,EAAE;YACV,KAAK,aAAa,CAAC,KAAK,CAAC,CAAC,OAAO,MAAM,IAAI,CAAC,WAAW,CAAC,SAAS,CAAC,UAAU,CAAC,CAAC;YAC9E,KAAK,aAAa,CAAC,IAAI,CAAC,CAAC,OAAO,MAAM,IAAI,CAAC,WAAW,CAAC,QAAQ,CAAC,UAAU,CAAC,CAAC;YAC5E,KAAK,aAAa,CAAC,MAAM,CAAC,CAAC,OAAO,MAAM,IAAI,CAAC,WAAW,CAAC,UAAU,CAAC,UAAU,CAAC,CAAC;YAChF,KAAK,aAAa,CAAC,KAAK,CAAC,CAAC,OAAO,MAAM,IAAI,CAAC,WAAW,CAAC,SAAS,CAAC,UAAU,CAAC,CAAC;YAC9E,KAAK,aAAa,CAAC,MAAM,CAAC,CAAC,OAAO,MAAM,IAAI,CAAC,WAAW,CAAC,UAAU,CAAC,UAAU,CAAC,CAAC;YAChF,KAAK,aAAa,CAAC,KAAK,CAAC,CAAC,OAAO,MAAM,IAAI,CAAC,WAAW,CAAC,SAAS,CAAC,UAAU,CAAC,CAAC;YAC9E,KAAK,aAAa,CAAC,OAAO,CAAC,CAAC,OAAO,MAAM,IAAI,CAAC,WAAW,CAAC,WAAW,CAAC,UAAU,CAAC,CAAC;YAClF,KAAK,aAAa,CAAC,IAAI,CAAC,CAAC,OAAO,MAAM,IAAI,CAAC,WAAW,CAAC,QAAQ,CAAC,UAAU,CAAC,CAAC;YAC5E,KAAK,aAAa,CAAC,MAAM,CAAC,CAAC,OAAO,MAAM,IAAI,CAAC,gBAAgB,CAAC,UAAU,CAAC,CAAC;YAC1E,KAAK,aAAa,CAAC,MAAM,CAAC,CAAC,OAAO,MAAM,IAAI,CAAC,WAAW,CAAC,UAAU,CAAC,UAAU,CAAC,CAAC;YAChF,KAAK,aAAa,CAAC,KAAK,CAAC,CAAC,OAAO,MAAM,IAAI,CAAC,WAAW,CAAC,SAAS,CAAC,UAAU,CAAC,CAAC;YAC9E,KAAK,aAAa,CAAC,OAAO,CAAC,CAAC,OAAO,MAAM,IAAI,CAAC,WAAW,CAAC,WAAW,CAAC,UAAU,CAAC,CAAC;SACrF;QAED,IAAI,IAAI,KAAK,aAAa,CAAC,KAAK,EAAE;YAC9B,MAAM,SAAS,GAAG,MAAM,IAAI,CAAC,WAAW,CAAC,UAAU,CAAC,UAAU,CAAC,CAAC;YAChE,MAAM,WAAW,GAAG,MAAM,IAAI,CAAC,WAAW,CAAC,UAAU,CAAC,UAAU,CAAC,CAAC;YAElE,MAAM,WAAW,GAAoB,EAAE,CAAC;YACxC,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,WAAW,EAAE,CAAC,EAAE,EAAE;gBAClC,MAAM,KAAK,GAAG,MAAM,IAAI,CAAC,cAAc,CAAC,SAAS,EAAE,UAAU,CAAC,CAAC;gBAC/D,WAAW,CAAC,IAAI,CAAC,KAAK,CAAC,CAAC;aAC3B;YACD,OAAO,WAAW,CAAC;SACtB;QAED,MAAM,IAAI,6BAA6B,CAAC,IAAI,CAAC,CAAC;IAClD,CAAC;IAES,KAAK,CAAC,gBAAgB,CAAC,MAA+B;QAC5D,MAAM,UAAU,GAAG,cAAc,CAAC,iBAAiB,CAAC,MAAM,CAAC,CAAC;QAC5D,MAAM,MAAM,GAAG,MAAM,CAAC,MAAM,IAAI,CAAC,WAAW,CAAC,UAAU,CAAC,UAAU,CAAC,CAAC,CAAC;QAErE,MAAM,UAAU,GAAG,sBAAsB,CAAC,KAAK,GAAG,MAAM,CAAC;QACzD,MAAM,WAAW,GAAG,MAAM,IAAI,CAAC,WAAW,CAAC,aAAa,CAAC,UAAU,EAAE,UAAU,CAAC,CAAC;QAEjF,OAAO,MAAM,CAAC,YAAY,CAAC,GAAG,WAAW,CAAC,CAAC;IAC/C,CAAC;IAES,KAAK,CAAC,cAAc,CAAC,UAA0B;QACrD,MAAM,aAAa,GAAG,UAAU,CAAC,MAAM,CAAC;QAExC,MAAM,WAAW,GAAG,MAAM,IAAI,CAAC,WAAW,CAAC,UAAU,CAAC,UAAU,CAAC,CAAC;QAClE,MAAM,eAAe,GAAG,MAAM,CAAC,MAAM,IAAI,CAAC,WAAW,CAAC,UAAU,CAAC,UAAU,CAAC,CAAC,CAAC;QAE9E,MAAM,QAAQ,GAA2B,EAAE,CAAC;QAE5C,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,eAAe,EAAE,CAAC,EAAE,EAAE;YACtC,MAAM,SAAS,GAAG,MAAM,IAAI,CAAC,gBAAgB,CAAC,UAAU,CAAC,CAAC;YAC1D,MAAM,SAAS,GAAG,MAAM,IAAI,CAAC,WAAW,CAAC,UAAU,CAAC,UAAU,CAAC,CAAC;YAChE,QAAQ,CAAC,SAAS,CAAC,GAAG,MAAM,IAAI,CAAC,cAAc,CAAC,SAAS,EAAE,UAAU,CAAC,CAAC;SAC1E;QAED,OAAO;YACH,WAAW,EAAE,cAAc,CAAC,gBAAgB,CAAC,WAAW,CAAC;YACzD,QAAQ,EAAE,QAAQ;YAClB,UAAU,EAAE,UAAU,CAAC,MAAM,GAAG,aAAa;SAChD,CAAC;IACN,CAAC;IAEO,KAAK,CAAC,eAAe,CAAC,WAA4B,EAAE,UAA0B;QAClF,MAAM,aAAa,GAAG,UAAU,CAAC,MAAM,CAAC;QACxC,MAAM,UAAU,GAAqB,EAAE,CAAC;QAExC,KAAK,IAAI,CAAC,GAAG,EAAE,EAAE,CAAC,GAAG,MAAM,CAAC,WAAW,CAAC,EAAE,CAAC,EAAE,EAAE;YAC3C,MAAM,IAAI,GAAG,MAAM,IAAI,CAAC,gBAAgB,CAAC,UAAU,CAAC,CAAC;YACrD,MAAM,gBAAgB,GAAG,MAAM,IAAI,CAAC,WAAW,CAAC,UAAU,CAAC,UAAU,CAAC,CAAC;YACvE,MAAM,UAAU,GAAwB,EAAE,CAAC;YAE3C,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,gBAAgB,EAAE,CAAC,EAAE,EAAE;gBACvC,MAAM,SAAS,GAAG,MAAM,IAAI,CAAC,WAAW,CAAC,UAAU,CAAC,UAAU,CAAC,CAAC;gBAChE,UAAU,CAAC,IAAI,CAAC,cAAc,CAAC,gBAAgB,CAAC,SAAS,CAAC,CAAC,CAAC;aAC/D;YAED,MAAM,QAAQ,GAAG,MAAM,IAAI,CAAC,WAAW,CAAC,UAAU,CAAC,UAAU,CAAC,CAAC;YAC/D,MAAM,MAAM,GAAG,MAAM,IAAI,CAAC,WAAW,CAAC,UAAU,CAAC,UAAU,CAAC,CAAC;YAE7D,UAAU,CAAC,IAAI,CAAC;gBACZ,IAAI;gBACJ,UAAU;gBACV,QAAQ,EAAE,QAAoB;gBAC9B,MAAM,EAAE,cAAc,CAAC,gBAAgB,CAAC,MAAM,CAAC;aAClD,CAAC,CAAC;SACN;QAED,OAAO;YACH,UAAU;YACV,cAAc,EAAE,UAAU,CAAC,MAAM,GAAG,aAAa;SACpD,CAAC;IACN,CAAC;CACJ"}
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"GgufV3Parser.js","sourceRoot":"","sources":["../../../src/gguf/parser/GgufV3Parser.ts"],"names":[],"mappings":"AAAA,OAAO,EAAC,YAAY,EAAC,MAAM,mBAAmB,CAAC;AAE/C,MAAM,OAAO,YAAa,SAAQ,YAAY;CAE7C"}
|
|
@@ -0,0 +1,8 @@
|
|
|
1
|
+
import { GgufFileReader } from "../fileReaders/GgufFileReader.js";
|
|
2
|
+
import { GgufFileInfo } from "../types/GgufFileInfoTypes.js";
|
|
3
|
+
export declare function parseGguf({ fileReader, readTensorInfo, ignoreKeys, logWarnings }: {
|
|
4
|
+
fileReader: GgufFileReader;
|
|
5
|
+
readTensorInfo?: boolean;
|
|
6
|
+
ignoreKeys?: string[];
|
|
7
|
+
logWarnings?: boolean;
|
|
8
|
+
}): Promise<GgufFileInfo>;
|
|
@@ -0,0 +1,63 @@
|
|
|
1
|
+
import { InvalidGgufMagicError } from "../errors/InvalidGgufMagicError.js";
|
|
2
|
+
import { getConsoleLogPrefix } from "../../utils/getConsoleLogPrefix.js";
|
|
3
|
+
import { UnsupportedError } from "../../utils/UnsupportedError.js";
|
|
4
|
+
import { GgufReadOffset } from "../utils/GgufReadOffset.js";
|
|
5
|
+
import { valueTypeToBytesToRead } from "../fileReaders/GgufFileReader.js";
|
|
6
|
+
import { getGgufMetadataArchitectureData } from "../utils/getGgufMetadataArchitectureData.js";
|
|
7
|
+
import { GgufV2Parser } from "./GgufV2Parser.js";
|
|
8
|
+
import { GgufV3Parser } from "./GgufV3Parser.js";
|
|
9
|
+
const ggufMagic = "GGUF";
|
|
10
|
+
export async function parseGguf({ fileReader, readTensorInfo = true, ignoreKeys = [], logWarnings = true }) {
|
|
11
|
+
const readOffset = new GgufReadOffset(0);
|
|
12
|
+
const magicAndVersion = await parseMagicAndVersion(fileReader, readOffset);
|
|
13
|
+
const gguifInfo = await parseGgufUsingASpecificVersionParser({
|
|
14
|
+
fileReader,
|
|
15
|
+
readTensorInfo,
|
|
16
|
+
ignoreKeys,
|
|
17
|
+
version: magicAndVersion.version,
|
|
18
|
+
readOffset,
|
|
19
|
+
logWarnings
|
|
20
|
+
});
|
|
21
|
+
const architectureMetadata = getGgufMetadataArchitectureData(gguifInfo.metadata);
|
|
22
|
+
return {
|
|
23
|
+
version: magicAndVersion.version,
|
|
24
|
+
tensorCount: gguifInfo.tensorCount,
|
|
25
|
+
metadata: gguifInfo.metadata,
|
|
26
|
+
architectureMetadata: architectureMetadata,
|
|
27
|
+
tensorInfo: gguifInfo.tensorInfo,
|
|
28
|
+
metadataSize: gguifInfo.metadataSize,
|
|
29
|
+
splicedParts: 1,
|
|
30
|
+
totalTensorInfoSize: gguifInfo.tensorInfoSize,
|
|
31
|
+
totalTensorCount: gguifInfo.tensorCount,
|
|
32
|
+
totalMetadataSize: gguifInfo.metadataSize,
|
|
33
|
+
fullTensorInfo: gguifInfo.tensorInfo,
|
|
34
|
+
tensorInfoSize: gguifInfo.tensorInfoSize
|
|
35
|
+
};
|
|
36
|
+
}
|
|
37
|
+
async function parseMagicAndVersion(fileReader, readOffset) {
|
|
38
|
+
const fileMagicBytes = await fileReader.readByteRange(readOffset, valueTypeToBytesToRead.uint8 * ggufMagic.length);
|
|
39
|
+
const fileMagicText = String.fromCharCode(...fileMagicBytes);
|
|
40
|
+
if (fileMagicText !== ggufMagic)
|
|
41
|
+
throw new InvalidGgufMagicError(ggufMagic, fileMagicText);
|
|
42
|
+
const version = await fileReader.readUint32(readOffset);
|
|
43
|
+
return {
|
|
44
|
+
magic: ggufMagic,
|
|
45
|
+
version
|
|
46
|
+
};
|
|
47
|
+
}
|
|
48
|
+
async function parseGgufUsingASpecificVersionParser(specificVersionParserOptions) {
|
|
49
|
+
switch (specificVersionParserOptions.version) {
|
|
50
|
+
case 1:
|
|
51
|
+
throw new UnsupportedError("GGUF version 1 is not supported by llama.cpp anymore");
|
|
52
|
+
case 2:
|
|
53
|
+
return await (new GgufV2Parser(specificVersionParserOptions)).parse();
|
|
54
|
+
case 3:
|
|
55
|
+
return await (new GgufV3Parser(specificVersionParserOptions)).parse();
|
|
56
|
+
default:
|
|
57
|
+
if (specificVersionParserOptions.logWarnings)
|
|
58
|
+
console.warn(getConsoleLogPrefix() +
|
|
59
|
+
`Unsupported GGUF version "${specificVersionParserOptions.version}". Reading the file as GGUF version 3`);
|
|
60
|
+
return await (new GgufV3Parser(specificVersionParserOptions)).parse();
|
|
61
|
+
}
|
|
62
|
+
}
|
|
63
|
+
//# sourceMappingURL=parseGguf.js.map
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"parseGguf.js","sourceRoot":"","sources":["../../../src/gguf/parser/parseGguf.ts"],"names":[],"mappings":"AAAA,OAAO,EAAC,qBAAqB,EAAC,MAAM,oCAAoC,CAAC;AACzE,OAAO,EAAC,mBAAmB,EAAC,MAAM,oCAAoC,CAAC;AACvE,OAAO,EAAC,gBAAgB,EAAC,MAAM,iCAAiC,CAAC;AACjE,OAAO,EAAC,cAAc,EAAC,MAAM,4BAA4B,CAAC;AAC1D,OAAO,EAAiB,sBAAsB,EAAC,MAAM,kCAAkC,CAAC;AAExF,OAAO,EAAC,+BAA+B,EAAC,MAAM,6CAA6C,CAAC;AAC5F,OAAO,EAAC,YAAY,EAAC,MAAM,mBAAmB,CAAC;AAC/C,OAAO,EAAC,YAAY,EAAC,MAAM,mBAAmB,CAAC;AAE/C,MAAM,SAAS,GAAG,MAAM,CAAC;AAEzB,MAAM,CAAC,KAAK,UAAU,SAAS,CAAC,EAC5B,UAAU,EACV,cAAc,GAAG,IAAI,EACrB,UAAU,GAAG,EAAE,EACf,WAAW,GAAG,IAAI,EAMrB;IACG,MAAM,UAAU,GAAG,IAAI,cAAc,CAAC,CAAC,CAAC,CAAC;IACzC,MAAM,eAAe,GAAG,MAAM,oBAAoB,CAAC,UAAU,EAAE,UAAU,CAAC,CAAC;IAC3E,MAAM,SAAS,GAAG,MAAM,oCAAoC,CAAC;QACzD,UAAU;QACV,cAAc;QACd,UAAU;QAEV,OAAO,EAAE,eAAe,CAAC,OAAO;QAChC,UAAU;QACV,WAAW;KACd,CAAC,CAAC;IACH,MAAM,oBAAoB,GAAG,+BAA+B,CAAC,SAAS,CAAC,QAAQ,CAAC,CAAC;IAEjF,OAAO;QACH,OAAO,EAAE,eAAe,CAAC,OAAO;QAChC,WAAW,EAAE,SAAS,CAAC,WAAW;QAClC,QAAQ,EAAE,SAAS,CAAC,QAAQ;QAC5B,oBAAoB,EAAE,oBAAoB;QAC1C,UAAU,EAAE,SAAS,CAAC,UAAU;QAChC,YAAY,EAAE,SAAS,CAAC,YAAY;QACpC,YAAY,EAAE,CAAC;QACf,mBAAmB,EAAE,SAAS,CAAC,cAAc;QAC7C,gBAAgB,EAAE,SAAS,CAAC,WAAW;QACvC,iBAAiB,EAAE,SAAS,CAAC,YAAY;QACzC,cAAc,EAAE,SAAS,CAAC,UAAU;QACpC,cAAc,EAAE,SAAS,CAAC,cAAc;KAC3C,CAAC;AACN,CAAC;AAED,KAAK,UAAU,oBAAoB,CAAC,UAA0B,EAAE,UAA0B;IACtF,MAAM,cAAc,GAAG,MAAM,UAAU,CAAC,aAAa,CAAC,UAAU,EAAE,sBAAsB,CAAC,KAAK,GAAG,SAAS,CAAC,MAAM,CAAC,CAAC;IACnH,MAAM,aAAa,GAAG,MAAM,CAAC,YAAY,CAAC,GAAG,cAAc,CAAC,CAAC;IAE7D,IAAI,aAAa,KAAK,SAAS;QAC3B,MAAM,IAAI,qBAAqB,CAAC,SAAS,EAAE,aAAa,CAAC,CAAC;IAE9D,MAAM,OAAO,GAAG,MAAM,UAAU,CAAC,UAAU,CAAC,UAAU,CAAC,CAAC;IAExD,OAAO;QACH,KAAK,EAAE,SAAS;QAChB,OAAO;KACV,CAAC;AACN,CAAC;AAED,KAAK,UAAU,oCAAoC,CAC/C,4BAAsD;IAEtD,QAAQ,4BAA4B,CAAC,OAAO,EAAE;QAC1C,KAAK,CAAC;YACF,MAAM,IAAI,gBAAgB,CAAC,sDAAsD,CAAC,CAAC;QAEvF,KAAK,CAAC;YACF,OAAO,MAAM,CAAC,IAAI,YAAY,CAAC,4BAA4B,CAAC,CAAC,CAAC,KAAK,EAAE,CAAC;QAE1E,KAAK,CAAC;YACF,OAAO,MAAM,CAAC,IAAI,YAAY,CAAC,4BAA4B,CAAC,CAAC,CAAC,KAAK,EAAE,CAAC;QAE1E;YACI,IAAI,4BAA4B,CAAC,WAAW;gBACxC,OAAO,CAAC,IAAI,CACR,mBAAmB,EAAE;oBACrB,6BAA6B,4BAA4B,CAAC,OAAO,uCAAuC,CAC3G,CAAC;YAEN,OAAO,MAAM,CAAC,IAAI,YAAY,CAAC,4BAA4B,CAAC,CAAC,CAAC,KAAK,EAAE,CAAC;KAC7E;AACL,CAAC"}
|
|
@@ -0,0 +1,33 @@
|
|
|
1
|
+
/// <reference types="node" />
|
|
2
|
+
import retry from "async-retry";
|
|
3
|
+
import { GgufFileInfo } from "./types/GgufFileInfoTypes.js";
|
|
4
|
+
/**
|
|
5
|
+
* Read a GGUF file and return its metadata and tensor info (unless `readTensorInfo` is set to `false`).
|
|
6
|
+
* Only the parts of the file required for the metadata and tensor info are read.
|
|
7
|
+
*/
|
|
8
|
+
export declare function readGgufFileInfo(pathOrUrl: string, { readTensorInfo, sourceType, ignoreKeys, logWarnings, fetchRetryOptions, fetchHeaders, spliceSplitFiles, signal }?: {
|
|
9
|
+
/**
|
|
10
|
+
* Whether to read the tensor info from the file's header
|
|
11
|
+
* Enabled by default.
|
|
12
|
+
*/
|
|
13
|
+
readTensorInfo?: boolean;
|
|
14
|
+
/**
|
|
15
|
+
* Set to a specific value to force it to only use that source type.
|
|
16
|
+
* By default, it detects whether the path is a network URL or a filesystem path and uses the appropriate reader accordingly.
|
|
17
|
+
*/
|
|
18
|
+
sourceType?: "network" | "filesystem";
|
|
19
|
+
/**
|
|
20
|
+
* Metadata keys to ignore when parsing the metadata.
|
|
21
|
+
* For example, `["tokenizer.ggml.tokens"]`
|
|
22
|
+
*/
|
|
23
|
+
ignoreKeys?: string[];
|
|
24
|
+
/** Whether to log warnings */
|
|
25
|
+
logWarnings?: boolean;
|
|
26
|
+
/** Relevant only when fetching from a network */
|
|
27
|
+
fetchRetryOptions?: retry.Options;
|
|
28
|
+
/** Relevant only when fetching from a network */
|
|
29
|
+
fetchHeaders?: Record<string, string>;
|
|
30
|
+
/** When split files are detected, read the metadata of the first file and splice the tensor info from all the parts */
|
|
31
|
+
spliceSplitFiles?: boolean;
|
|
32
|
+
signal?: AbortSignal;
|
|
33
|
+
}): Promise<GgufFileInfo>;
|