batch

package
v1.27.23 Latest Latest
Warning

This package is not in the latest version of its module.

Go to latest
Published: May 7, 2025 License: BSD-3-Clause Imports: 17 Imported by: 1

Documentation

Index

Constants

View Source
const BatchChannelSize = 100

Variables

This section is empty.

Functions

func VectorizeBatch

func VectorizeBatch(ctx context.Context, objs []*models.Object, skipObject []bool, cfg moduletools.ClassConfig, logger logrus.FieldLogger, objectVectorizer objectVectorizer) ([][]float32, []models.AdditionalProperties, map[int]error)

Types

type Batch

type Batch struct {
	Label string
	// contains filtered or unexported fields
}

func NewBatchVectorizer

func NewBatchVectorizer(client BatchClient, maxBatchTime time.Duration, settings Settings, logger logrus.FieldLogger, label string) *Batch

func (*Batch) SubmitBatchAndWait

func (b *Batch) SubmitBatchAndWait(ctx context.Context, cfg moduletools.ClassConfig, skipObject []bool, tokenCounts []int, texts []string) ([][]float32, map[int]error)

type BatchClient

type BatchClient interface {
	Vectorize(ctx context.Context, input []string,
		config moduletools.ClassConfig) (*modulecomponents.VectorizationResult, *modulecomponents.RateLimits, int, error)
	GetVectorizerRateLimit(ctx context.Context, config moduletools.ClassConfig) *modulecomponents.RateLimits
	GetApiKeyHash(ctx context.Context, config moduletools.ClassConfig) [32]byte
}

type BatchJob

type BatchJob struct {
	// contains filtered or unexported fields
}

type EncoderCache added in v1.27.18

type EncoderCache struct {
	// contains filtered or unexported fields
}

func NewEncoderCache added in v1.27.18

func NewEncoderCache() *EncoderCache

func (*EncoderCache) Get added in v1.27.18

func (e *EncoderCache) Get(model string) (*tiktoken.Tiktoken, bool)

func (*EncoderCache) Set added in v1.27.18

func (e *EncoderCache) Set(model string, tk *tiktoken.Tiktoken)

type Settings added in v1.25.25

type Settings struct {
	TokenMultiplier    float32
	MaxTimePerBatch    float64
	MaxObjectsPerBatch int
	MaxTokensPerBatch  func(cfg moduletools.ClassConfig) int
	HasTokenLimit      bool
	ReturnsRateLimit   bool
}

type TokenizerFuncType added in v1.25.25

type TokenizerFuncType func(ctx context.Context, objects []*models.Object, skipObject []bool, cfg moduletools.ClassConfig, objectVectorizer *objectsvectorizer.ObjectVectorizer, encoderCache *EncoderCache) ([]string, []int, bool, error)

func ReturnBatchTokenizer added in v1.25.25

func ReturnBatchTokenizer(multiplier float32, moduleName string, lowerCaseInput bool) TokenizerFuncType

Jump to

Keyboard shortcuts

? : This menu
/ : Search site
f or F : Jump to
y or Y : Canonical URL