Module @mlx-bun/quantize

Interfaces

NamedTensor
SafetensorsIndex
ShardInfo
WriteOpts
WriteResult

Variables

DEFAULT_SHARD_BYTES

Functions

writeShardedSafetensors

References

analyzeSensitivityExact → analyzeSensitivityExact
AnalyzeSensitivityOptions → AnalyzeSensitivityOptions
automaticRotationWeightTransform → automaticRotationWeightTransform
buildQuantizationBlock → buildQuantizationBlock
computeBpw → computeBpw
CONVERT_DTYPES → CONVERT_DTYPES
ConvertDtype → ConvertDtype
convertModelDir → convertModelDir
ConvertOptions → ConvertOptions
ConvertResult → ConvertResult
DownAxis → DownAxis
drafterQuantizePredicate → drafterQuantizePredicate
FoldOptions → FoldOptions
isQuantizable → isQuantizable
klFromRef → klFromRef
klReduction → klReduction
LayerQuantConfig → LayerQuantConfig
LdlqEncoding → LdlqEncoding
llamaWeightTransform → llamaWeightTransform
LlamaWeightTransformPlan → LlamaWeightTransformPlan
loadHessianFactor → loadHessianFactor
loadLlmCalibration → loadLlmCalibration
mlxLmPath → mlxLmPath
OptimizationResult → OptimizationResult
optimizeMixedPrecision → optimizeMixedPrecision
OptimizeMixedPrecisionOptions → OptimizeMixedPrecisionOptions
OptiqMetadata → OptiqMetadata
PackedTrellisTensor → PackedTrellisTensor
parseKMap → parseKMap
PerLayerEntry → PerLayerEntry
planLlamaWeightTransform → planLlamaWeightTransform
planQwen35Fold → planQwen35Fold
planQwen35HfFold → planQwen35HfFold
planQwenMtpFold → planQwenMtpFold
PreparedProbe → PreparedProbe
ProbeSource → ProbeSource
ProgressEvent → ProgressEvent
QuantDef → QuantDef
QuantizationBlock → QuantizationBlock
quantizeDrafterDir → quantizeDrafterDir
QuantizeDrafterOptions → QuantizeDrafterOptions
quantizeModelDir → quantizeModelDir
QuantizeOptions → QuantizeOptions
QuantizeResult → QuantizeResult
quantizeTrellisModelDir → quantizeTrellisModelDir
Qwen35HfFoldPlan → Qwen35HfFoldPlan
Qwen35HfFoldSource → Qwen35HfFoldSource
qwen35WeightTransform → qwen35WeightTransform
QwenFoldOptions → QwenFoldOptions
qwenMtpWeightTransform → qwenMtpWeightTransform
QwenWeightTransformPlan → QwenWeightTransformPlan
SensitivityResult → SensitivityResult
Treatment → Treatment
treatmentFor → treatmentFor
TreatmentPolicy → TreatmentPolicy
trellisBpw → trellisBpw
TrellisEncoder → TrellisEncoder
TrellisEncoderOptions → TrellisEncoderOptions
TrellisKMap → TrellisKMap
TrellisQuantizeOptions → TrellisQuantizeOptions
TrellisQuantizeResult → TrellisQuantizeResult
UpgradeEntry → UpgradeEntry
UpgradeHeap → UpgradeHeap
WeightTransform → WeightTransform
WeightTransformContext → WeightTransformContext
WeightTransformMetadata → WeightTransformMetadata
WeightTransformPlan → WeightTransformPlan
withPreparedProbe → withPreparedProbe
writeQuantizedConfig → writeQuantizedConfig