Module @mlx-bun/inference/generation/speculative

Classes

AdaptiveDraftGate
AssistantProvider
DeepspecProvider
Dflash2Provider
DflashProvider
DraftAcceptance
DraftVocabularyHead
Glm52NativeMtpProvider
NgramProvider
QwenMtpProvider
TwoModelProvider

Interfaces

AdaptiveDraftGateOptions
DraftOutputRow
DraftPrefillGroup
DraftProvider
DraftRowConstraints
DraftRowGroup
DraftRowSampling
GroupedDraftProvider
NgramOptions
ProjectedDraftMethod
QuantizedHeadState

Type Aliases

DraftOutputResult
DraftSelection
SubsetDraftSampler
VerifyCostModel

Functions

configuredDraftVocabulary
constraintDraftProvider
deliverDraftOutputs
draftSamplerOptions
makeSubsetDraftSampler
projectedDraftGroups

References

AssistantRowsTarget → AssistantRowsTarget
defaultDraftProviders → defaultDraftProviders
detectDraftKind → detectDraftKind
DRAFT_KINDS → DRAFT_KINDS
DRAFT_VOCAB_FILE → DRAFT_VOCAB_FILE
draftArtifact → draftArtifact
DraftArtifact → DraftArtifact
DraftKind → DraftKind
DraftLoadRequest → DraftLoadRequest
DraftProjection → DraftProjection
DraftProviderKind → DraftProviderKind
DraftProviderRegistry → DraftProviderRegistry
DraftRowCheckpoint → DraftRowCheckpoint
DraftTarget → DraftTarget
HiddenLayerTapsTarget → HiddenLayerTapsTarget
LoadedDraft → LoadedDraft
NativeDraftKind → NativeDraftKind
RecurrentMtpTarget → RecurrentMtpTarget
targetLacks → targetLacks
TargetPort → TargetPort
TargetView → TargetView