mlx-bun library API
    Preparing search index...

    Interface Glm52MemoryPlan

    Common admission contract for every runtime. Representation-specific planners may expose richer fields, but serving consumes only this shape.

    interface Glm52MemoryPlan {
        allocatorLimitBytes?: number;
        batchSize: number;
        contextTokens: number;
        enableMtp: boolean;
        fits: boolean;
        kvBytes: number;
        lineItems: Glm52MemoryLineItems;
        machineBytes: number;
        machineHeadroomBytes: number;
        mainRequiredUnionSlots: number;
        mainResidentSlots: number;
        mainTotalSlots: number;
        mainWorkingSlots: number;
        maxGenerationTokens: number;
        maxSafeContext: number;
        mtpDraftTokens: number;
        mtpResidentSlots: number;
        mtpTotalSlots: number;
        mtpWorkingSlots: number;
        osReserveBytes: number;
        plannedMachineBytes: number;
        plannedProcessBytes: number;
        predictedDecodeTps: number | null;
        preset: "g5-32gb-quality";
        processHeadroomBytes: number;
        processLimitBytes: number;
        reserveBytes: number;
        runtimeReserveBytes: number;
        schemaVersion: 1;
        strategy: "glm52-colibri";
        streamedWeights?: Readonly<Record<string, number>>;
        totalBytes: number;
        transientBytes: number;
        usableBytes: number;
        verifyRows: number;
        weightsBytes: number;
    }

    Hierarchy (View Summary)

    Index
    allocatorLimitBytes? batchSize contextTokens enableMtp fits kvBytes lineItems machineBytes machineHeadroomBytes mainRequiredUnionSlots mainResidentSlots mainTotalSlots mainWorkingSlots maxGenerationTokens maxSafeContext mtpDraftTokens mtpResidentSlots mtpTotalSlots mtpWorkingSlots osReserveBytes plannedMachineBytes plannedProcessBytes predictedDecodeTps preset processHeadroomBytes processLimitBytes reserveBytes runtimeReserveBytes schemaVersion strategy streamedWeights? totalBytes transientBytes usableBytes verifyRows weightsBytes
    allocatorLimitBytes?: number

    mlx allocator-cache limit, distinct from the process admission limit.

    batchSize: number
    contextTokens: number
    enableMtp: boolean
    fits: boolean
    kvBytes: number
    lineItems: Glm52MemoryLineItems
    machineBytes: number
    machineHeadroomBytes: number

    RAM the plan leaves to the rest of the machine.

    mainRequiredUnionSlots: number
    mainResidentSlots: number
    mainTotalSlots: number
    mainWorkingSlots: number
    maxGenerationTokens: number

    Generated-token allowance reserved within contextTokens; absent when the plan reserves none.

    maxSafeContext: number
    mtpDraftTokens: number
    mtpResidentSlots: number
    mtpTotalSlots: number
    mtpWorkingSlots: number
    osReserveBytes: number

    Deliberately unavailable to this process.

    plannedMachineBytes: number
    plannedProcessBytes: number
    predictedDecodeTps: number | null
    preset: "g5-32gb-quality"
    processHeadroomBytes: number
    processLimitBytes: number
    reserveBytes: number
    runtimeReserveBytes: number
    schemaVersion: 1
    strategy: "glm52-colibri"

    Names the accounting the plan comes from; informational, never branched on.

    streamedWeights?: Readonly<Record<string, number>>

    Parts of weightsBytes held in bounded streaming tiers instead of resident memory, by display label. Resident weights are weightsBytes minus their sum.

    totalBytes: number
    transientBytes: number
    usableBytes: number
    verifyRows: number
    weightsBytes: number

    Settings
    Member Visibility

    On This Page
    Properties