textgeneration

Documentation

Index

• Constants
• Variables
• func DefaultModelForMachineTranslation(source, target string) string
• func PrepareInputForAbstractiveQuestionAnswering(question string, passages []string) string
• type Interface
• type Options
  • func DefaultOptions() *Options
  • func DefaultOptionsForTextParaphrasing() *Options
• type Response

Constants

const (
	// DefaultModelTemplateForMachineTranslation is the template for machine translations models.
	// You can use it as the basis for constructing the default model name for the desired source
	// and target languages pair (iso-a2), or you can use the convenient function DefaultModelForMachineTranslation.
	// Model card: https://huggingface.co/Helsinki-NLP
	DefaultModelTemplateForMachineTranslation = "Helsinki-NLP/opus-mt-%s-%s"

	// DefaultModelForTextParaphrasing is a summarization model fine-tuned for text paraphrasing.
	// Model card: https://huggingface.co/tuner007/pegasus_paraphrase
	DefaultModelForTextParaphrasing = "tuner007/pegasus_paraphrase"

	// DefaultModelForTextSummarization is a summarization model.
	// Model card: https://huggingface.co/facebook/bart-large-cnn
	DefaultModelForTextSummarization = "facebook/bart-large-cnn"

	// DefaultModelForTextSummarization2 is a summarization model.
	// Model card: https://huggingface.co/google/pegasus-multi_news
	DefaultModelForTextSummarization2 = "google/pegasus-multi_news"

	// DefaultModelForExtremeTextSummarization is a summarization model that tries to generate one-sentence answering
	// the question “What is the article about?”.
	// Model card: https://huggingface.co/facebook/bart-large-xsum
	DefaultModelForExtremeTextSummarization = "facebook/bart-large-xsum"

	// DefaultModelForAbstractiveQuestionAnswering is a summarization model fine-tuned for answer generation.
	// Model card: https://huggingface.co/vblagoje/bart_lfqa/tree/main
	DefaultModelForAbstractiveQuestionAnswering = "vblagoje/bart_lfqa"

	// DefaultModelForKeywordsGeneration is a text generation model that produces a concatenated sequence of keyphrases.
	// Model card: https://huggingface.co/bloomberg/KeyBART
	DefaultModelForKeywordsGeneration = "bloomberg/KeyBART"
)

Variables

var ErrInputSequenceTooLong = errors.New("input sequence too long")

ErrInputSequenceTooLong means that pre-processing the input text produced a sequence that exceeds the maximum allowed length.

Functions

func DefaultModelForMachineTranslation

func DefaultModelForMachineTranslation(source, target string) string

DefaultModelForMachineTranslation specializes the model template for the source and target languages (iso-a2).

func PrepareInputForAbstractiveQuestionAnswering

func PrepareInputForAbstractiveQuestionAnswering(question string, passages []string) string

PrepareInputForAbstractiveQuestionAnswering returns text to be input to the DefaultModelForAbstractiveQuestionAnswering.

Types

type Interface

type Interface interface {
	// Generate generates text (e.g. translation, summarization, paraphrase) from the given input.
	Generate(ctx context.Context, text string, opts *Options) (Response, error)
}

Interface defines the main functions for the TextGeneration task.

type Options

type Options struct {
	// Temperature is the temperature used for sampling.
	Temperature nullable.Type[float64]
	// Sample is whether to sample or greedy generation.
	Sample nullable.Type[bool]
	// TopK is the number of top-k candidates to be considered during generation.
	TopK nullable.Type[int]
	// TopP is the top-p candidates to be considered during generation.
	TopP nullable.Type[float64]
}

Options defines the options for generating text.

func DefaultOptions

func DefaultOptions() *Options

DefaultOptions returns the default options for generating text.

func DefaultOptionsForTextParaphrasing

func DefaultOptionsForTextParaphrasing() *Options

DefaultOptionsForTextParaphrasing returns the default options for generating text with sampling.

type Response

type Response struct {
	// Texts contains the generated texts.
	Texts []string
	// a list of floats that correspond the score of the generated text, in the same order as texts.
	Scores []float64
}

Response contains the result of the text generation.