chatmodel

package
v0.0.91 Latest Latest
Warning

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

 Go to latest Published: Dec 14, 2023 License: MIT Imports: 22 Imported by: 2

Documentation

Overview

Package chatmodel provides functionalities for working with Large Language Models (LLMs).

Index

Constants

This section is empty.

Variables

View Source 
var DefaultOpenAIOptions = OpenAIOptions{
	CallbackOptions: &schema.CallbackOptions{
		Verbose: golc.Verbose,
	},
	ModelName:        openai.GPT3Dot5Turbo,
	Temperature:      1,
	TopP:             1,
	PresencePenalty:  0,
	FrequencyPenalty: 0,
	MaxRetries:       3,
}

Functions

This section is empty.

Types

type Anthropic 

type Anthropic struct {
	schema.Tokenizer
	// contains filtered or unexported fields
}

Anthropic is a chat model based on the Anthropic API.

func NewAnthropic 

func NewAnthropic(apiKey string, optFns ...func(o *AnthropicOptions)) (*Anthropic, error)

NewAnthropic creates a new instance of the Anthropic chat model with the provided options.

func NewAnthropicFromClient added in v0.0.67 

func NewAnthropicFromClient(client AnthropicClient, optFns ...func(o *AnthropicOptions)) (*Anthropic, error)

NewAnthropicFromClient creates a new instance of the Anthropic chat model with the provided options.

func (*Anthropic) Callbacks 

func (cm *Anthropic) Callbacks() []schema.Callback

Callbacks returns the registered callbacks of the model.

func (*Anthropic) Generate 

func (cm *Anthropic) Generate(ctx context.Context, messages schema.ChatMessages, optFns ...func(o *schema.GenerateOptions)) (*schema.ModelResult, error)

Generate generates text based on the provided chat messages and options.

func (*Anthropic) InvocationParams added in v0.0.27 

func (cm *Anthropic) InvocationParams() map[string]any

InvocationParams returns the parameters used in the model invocation.

func (*Anthropic) Type 

func (cm *Anthropic) Type() string

Type returns the type of the model.

func (*Anthropic) Verbose 

func (cm *Anthropic) Verbose() bool

Verbose returns the verbosity setting of the model.

type AnthropicClient added in v0.0.67 

type AnthropicClient interface {
	CreateCompletion(ctx context.Context, request *anthropic.CompletionRequest) (*anthropic.CompletionResponse, error)
}

AnthropicClient is the interface for the Anthropic client.

type AnthropicOptions 

type AnthropicOptions struct {
	*schema.CallbackOptions `map:"-"`
	schema.Tokenizer        `map:"-"`

	// Model name to use.
	ModelName string `map:"model_name,omitempty"`

	// Temperature parameter controls the randomness of the generation output.
	Temperature float32 `map:"temperature,omitempty"`

	// Denotes the number of tokens to predict per generation.
	MaxTokens int `map:"max_tokens,omitempty"`

	// TopK parameter specifies the number of highest probability tokens to consider for generation.
	TopK int `map:"top_k,omitempty"`

	// TopP parameter specifies the cumulative probability threshold for generating tokens.
	TopP float32 `map:"top_p,omitempty"`
}

AnthropicOptions contains options for configuring the Anthropic chat model.

type AzureOpenAI added in v0.0.36 

type AzureOpenAI struct {
	*OpenAI
	// contains filtered or unexported fields
}

func NewAzureOpenAI added in v0.0.26 

func NewAzureOpenAI(apiKey, baseURL string, optFns ...func(o *AzureOpenAIOptions)) (*AzureOpenAI, error)

func (*AzureOpenAI) InvocationParams added in v0.0.55 

func (cm *AzureOpenAI) InvocationParams() map[string]any

InvocationParams returns the parameters used in the model invocation.

func (*AzureOpenAI) Type added in v0.0.36 

func (cm *AzureOpenAI) Type() string

Type returns the type of the model.

type AzureOpenAIOptions added in v0.0.26 

type AzureOpenAIOptions struct {
	OpenAIOptions
	Deployment string `map:"deployment,omitempty"`
}

type Bedrock added in v0.0.71 

type Bedrock struct {
	schema.Tokenizer
	// contains filtered or unexported fields
}

Bedrock is a model implementation of the schema.ChatModel interface for the Bedrock model.

func NewBedrock added in v0.0.71 

func NewBedrock(client BedrockRuntimeClient, optFns ...func(o *BedrockOptions)) (*Bedrock, error)

NewBedrock creates an instance of the Bedrock model.

func NewBedrockAntrophic added in v0.0.78 

func NewBedrockAntrophic(client BedrockRuntimeClient, optFns ...func(o *BedrockAnthropicOptions)) (*Bedrock, error)

NewBedrockAntrophic creates a new instance of Bedrock for the "anthropic" provider.

func NewBedrockMeta added in v0.0.87 

func NewBedrockMeta(client BedrockRuntimeClient, optFns ...func(o *BedrockMetaOptions)) (*Bedrock, error)

NewBedrockMeta creates a new instance of Bedrock for the "meta" provider.

func (*Bedrock) Callbacks added in v0.0.71 

func (cm *Bedrock) Callbacks() []schema.Callback

Callbacks returns the registered callbacks of the model.

func (*Bedrock) Generate added in v0.0.71 

func (cm *Bedrock) Generate(ctx context.Context, messages schema.ChatMessages, optFns ...func(o *schema.GenerateOptions)) (*schema.ModelResult, error)

Generate generates text based on the provided chat messages and options.

func (*Bedrock) InvocationParams added in v0.0.71 

func (cm *Bedrock) InvocationParams() map[string]any

InvocationParams returns the parameters used in the model invocation.

func (*Bedrock) Type added in v0.0.71 

func (cm *Bedrock) Type() string

Type returns the type of the model.

func (*Bedrock) Verbose added in v0.0.71 

func (cm *Bedrock) Verbose() bool

Verbose returns the verbosity setting of the model.

type BedrockAnthropicOptions added in v0.0.78 

type BedrockAnthropicOptions struct {
	*schema.CallbackOptions `map:"-"`
	schema.Tokenizer        `map:"-"`

	// Model id to use.
	ModelID string `map:"model_id,omitempty"`

	// MaxTokensToSmaple sets the maximum number of tokens in the generated text.
	MaxTokensToSample int `map:"max_tokens_to_sample"`

	// Temperature controls the randomness of text generation. Higher values make it more random.
	Temperature float32 `map:"temperature"`

	// TopP is the total probability mass of tokens to consider at each step.
	TopP float32 `map:"top_p,omitempty"`

	// TopK determines how the model selects tokens for output.
	TopK int `map:"top_k"`

	// Stream indicates whether to stream the results or not.
	Stream bool `map:"stream,omitempty"`
}

BedrockAnthropicOptions contains options for configuring the Bedrock model with the "anthropic" provider.

type BedrockInputOutputAdapter added in v0.0.71 

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

BedrockInputOutputAdapter is a helper struct for preparing input and handling output for Bedrock model.

func NewBedrockInputOutputAdapter added in v0.0.71 

func NewBedrockInputOutputAdapter(provider string) *BedrockInputOutputAdapter

NewBedrockInputOutputAdpter creates a new instance of BedrockInputOutputAdpter.

func (*BedrockInputOutputAdapter) PrepareInput added in v0.0.71 

func (bioa *BedrockInputOutputAdapter) PrepareInput(messages schema.ChatMessages, modelParams map[string]any, stop []string) ([]byte, error)

PrepareInput prepares the input for the Bedrock model based on the specified provider.

func (*BedrockInputOutputAdapter) PrepareOutput added in v0.0.71 

func (bioa *BedrockInputOutputAdapter) PrepareOutput(response []byte) (string, error)

PrepareOutput prepares the output for the Bedrock model based on the specified provider.

type BedrockMetaOptions added in v0.0.87 

type BedrockMetaOptions struct {
	*schema.CallbackOptions `map:"-"`
	schema.Tokenizer        `map:"-"`

	// Model id to use.
	ModelID string `map:"model_id,omitempty"`

	// Temperature controls the randomness of text generation. Higher values make it more random.
	Temperature float32 `map:"temperature"`

	// TopP is the total probability mass of tokens to consider at each step.
	TopP float32 `map:"top_p,omitempty"`

	// MaxGenLen specify the maximum number of tokens to use in the generated response.
	MaxGenLen int `map:"max_gen_len"`

	// Stream indicates whether to stream the results or not.
	Stream bool `map:"stream,omitempty"`
}

BedrockMetaOptions contains options for configuring the Bedrock model with the "meta" provider.

type BedrockOptions added in v0.0.71 

type BedrockOptions struct {
	*schema.CallbackOptions `map:"-"`
	schema.Tokenizer        `map:"-"`

	// Model id to use.
	ModelID string `map:"model_id,omitempty"`

	// Model params to use.
	ModelParams map[string]any `map:"model_params,omitempty"`

	// Stream indicates whether to stream the results or not.
	Stream bool `map:"stream,omitempty"`
}

BedrockOptions contains options for configuring the Bedrock model.

type BedrockRuntimeClient added in v0.0.71 

type BedrockRuntimeClient interface {
	InvokeModel(ctx context.Context, params *bedrockruntime.InvokeModelInput, optFns ...func(*bedrockruntime.Options)) (*bedrockruntime.InvokeModelOutput, error)
	InvokeModelWithResponseStream(ctx context.Context, params *bedrockruntime.InvokeModelWithResponseStreamInput, optFns ...func(*bedrockruntime.Options)) (*bedrockruntime.InvokeModelWithResponseStreamOutput, error)
}

BedrockRuntimeClient is an interface for the Bedrock model runtime client.

type Ernie added in v0.0.67 

type Ernie struct {
	schema.Tokenizer
	// contains filtered or unexported fields
}

Ernie is a struct representing the Ernie language model.

func NewErnie added in v0.0.67 

func NewErnie(clientID, clientSecret string, optFns ...func(o *ErnieOptions)) (*Ernie, error)

NewErnie creates a new instance of the Ernie chat model.

func NewErnieFromClient added in v0.0.67 

func NewErnieFromClient(client ErnieClient, optFns ...func(o *ErnieOptions)) (*Ernie, error)

NewErnieFromClient creates a new instance of the Ernie chat model from a custom ErnieClient.

func (*Ernie) Callbacks added in v0.0.67 

func (cm *Ernie) Callbacks() []schema.Callback

Callbacks returns the registered callbacks of the model.

func (*Ernie) Generate added in v0.0.67 

func (cm *Ernie) Generate(ctx context.Context, messages schema.ChatMessages, optFns ...func(o *schema.GenerateOptions)) (*schema.ModelResult, error)

Generate generates text based on the provided chat messages and options.

func (*Ernie) InvocationParams added in v0.0.67 

func (cm *Ernie) InvocationParams() map[string]any

InvocationParams returns the parameters used in the model invocation.

func (*Ernie) Type added in v0.0.67 

func (cm *Ernie) Type() string

Type returns the type of the model.

func (*Ernie) Verbose added in v0.0.67 

func (cm *Ernie) Verbose() bool

Verbose returns the verbosity setting of the model.

type ErnieClient added in v0.0.67 

type ErnieClient interface {
	CreateChatCompletion(ctx context.Context, model string, request *ernie.ChatCompletionRequest) (*ernie.ChatCompletionResponse, error)
}

ErnieClient is the interface for the Ernie client.

type ErnieOptions added in v0.0.67 

type ErnieOptions struct {
	*schema.CallbackOptions `map:"-"`
	schema.Tokenizer        `map:"-"`

	// ModelName is the name of the Ernie chat model to use.
	ModelName string `map:"model_name,omitempty"`

	// Temperature is the sampling temperature to use during text generation.
	Temperature float64 `map:"temperature,omitempty"`

	// TopP is the total probability mass of tokens to consider at each step.
	TopP float64 `map:"top_p,omitempty"`

	// PenaltyScore is a parameter used during text generation to apply a penalty for generating longer responses.
	PenaltyScore float64 `map:"penalty_score"`
}

ErnieOptions is the options struct for the Ernie chat model.

type Fake added in v0.0.14 

type Fake struct {
	schema.Tokenizer
	// contains filtered or unexported fields
}

Fake is a mock implementation of the schema.ChatModel interface for testing purposes.

func NewFake added in v0.0.14 

func NewFake(fakeResultFunc FakeResultFunc, optFns ...func(o *FakeOptions)) *Fake

NewFake creates an instance of the Fake model with the provided custom result function.

func NewSimpleFake added in v0.0.58 

func NewSimpleFake(response string, optFns ...func(o *FakeOptions)) *Fake

NewSimpleFake creates a simple instance of the Fake model with a fixed response for all inputs.

func (*Fake) Callbacks added in v0.0.14 

func (cm *Fake) Callbacks() []schema.Callback

Callbacks returns the registered callbacks of the model.

func (*Fake) Generate added in v0.0.14 

func (cm *Fake) Generate(ctx context.Context, messages schema.ChatMessages, optFns ...func(o *schema.GenerateOptions)) (*schema.ModelResult, error)

Generate generates text based on the provided chat messages and options.

func (*Fake) InvocationParams added in v0.0.27 

func (cm *Fake) InvocationParams() map[string]any

InvocationParams returns the parameters used in the model invocation.

func (*Fake) Type added in v0.0.14 

func (cm *Fake) Type() string

Type returns the type of the model.

func (*Fake) Verbose added in v0.0.14 

func (cm *Fake) Verbose() bool

Verbose returns the verbosity setting of the model.

type FakeOptions added in v0.0.58 

type FakeOptions struct {
	*schema.CallbackOptions `map:"-"`
	schema.Tokenizer        `map:"-"`
	ChatModelType           string `map:"-"`
}

FakeOptions contains options for configuring the Fake model.

type FakeResultFunc added in v0.0.58 

type FakeResultFunc func(ctx context.Context, messages schema.ChatMessages) (*schema.ModelResult, error)

FakeResultFunc is a function type used for providing custom model results in the Fake model.

type Ollama added in v0.0.91 

type Ollama struct {
	schema.Tokenizer
	// contains filtered or unexported fields
}

Ollama is a struct representing the Ollama generative model.

func NewOllama added in v0.0.91 

func NewOllama(client OllamaClient, optFns ...func(o *OllamaOptions)) (*Ollama, error)

NewOllama creates a new instance of the Ollama model with the provided client and options.

func (*Ollama) Callbacks added in v0.0.91 

func (cm *Ollama) Callbacks() []schema.Callback

Callbacks returns the registered callbacks of the model.

func (*Ollama) Generate added in v0.0.91 

func (cm *Ollama) Generate(ctx context.Context, messages schema.ChatMessages, optFns ...func(o *schema.GenerateOptions)) (*schema.ModelResult, error)

Generate generates text based on the provided chat messages and options.

func (*Ollama) InvocationParams added in v0.0.91 

func (cm *Ollama) InvocationParams() map[string]any

InvocationParams returns the parameters used in the model invocation.

func (*Ollama) Type added in v0.0.91 

func (cm *Ollama) Type() string

Type returns the type of the model.

func (*Ollama) Verbose added in v0.0.91 

func (cm *Ollama) Verbose() bool

Verbose returns the verbosity setting of the model.

type OllamaClient added in v0.0.91 

type OllamaClient interface {
	// GenerateChat produces a single request and response for the Ollama generative model.
	GenerateChat(ctx context.Context, req *ollama.ChatRequest) (*ollama.ChatResponse, error)
}

OllamaClient is an interface for the Ollama generative model client.

type OllamaOptions added in v0.0.91 

type OllamaOptions struct {
	// CallbackOptions specify options for handling callbacks during text generation.
	*schema.CallbackOptions `map:"-"`
	// Tokenizer represents the tokenizer to be used with the LLM model.
	schema.Tokenizer `map:"-"`
	// ModelName is the name of the Gemini model to use.
	ModelName string `map:"model_name,omitempty"`
	// Temperature controls the randomness of the generation. Higher values make the output more random.
	Temperature float32 `map:"temperature,omitempty"`
	// MaxTokens is the maximum number of tokens to generate in the completion.
	MaxTokens int `map:"max_tokens,omitempty"`
	// TopP is the nucleus sampling parameter. It controls the cumulative probability of the most likely tokens to sample from.
	TopP float32 `map:"top_p,omitempty"`
	// TopK is the number of top tokens to consider for sampling.
	TopK int `map:"top_k,omitempty"`
	// PresencePenalty penalizes repeated tokens.
	PresencePenalty float32 `map:"presence_penalty,omitempty"`
	// FrequencyPenalty penalizes repeated tokens according to frequency.
	FrequencyPenalty float32 `map:"frequency_penalty,omitempty"`
}

OllamaOptions contains options for the Ollama model.

type OpenAI 

type OpenAI struct {
	schema.Tokenizer
	// contains filtered or unexported fields
}

OpenAI represents the OpenAI chat model.

func NewOpenAI 

func NewOpenAI(apiKey string, optFns ...func(o *OpenAIOptions)) (*OpenAI, error)

NewOpenAI creates a new instance of the OpenAI chat model.

func NewOpenAIFromClient added in v0.0.36 

func NewOpenAIFromClient(client OpenAIClient, optFns ...func(o *OpenAIOptions)) (*OpenAI, error)

NewOpenAIFromClient creates a new instance of the OpenAI chat model with the provided client and options.

func (*OpenAI) Callbacks 

func (cm *OpenAI) Callbacks() []schema.Callback

Callbacks returns the registered callbacks of the model.

func (*OpenAI) Generate 

func (cm *OpenAI) Generate(ctx context.Context, messages schema.ChatMessages, optFns ...func(o *schema.GenerateOptions)) (*schema.ModelResult, error)

Generate generates text based on the provided chat messages and options.

func (*OpenAI) InvocationParams added in v0.0.27 

func (cm *OpenAI) InvocationParams() map[string]any

InvocationParams returns the parameters used in the model invocation.

func (*OpenAI) Type 

func (cm *OpenAI) Type() string

Type returns the type of the model.

func (*OpenAI) Verbose 

func (cm *OpenAI) Verbose() bool

Verbose returns the verbosity setting of the model.

type OpenAIClient added in v0.0.36 

type OpenAIClient interface {
	CreateChatCompletion(ctx context.Context, request openai.ChatCompletionRequest) (response openai.ChatCompletionResponse, err error)
	CreateChatCompletionStream(ctx context.Context, request openai.ChatCompletionRequest) (stream *openai.ChatCompletionStream, err error)
}

OpenAIClient is an interface for the OpenAI chat model client.

type OpenAIOptions 

type OpenAIOptions struct {
	*schema.CallbackOptions `map:"-"`
	schema.Tokenizer        `map:"-"`
	// Model name to use.
	ModelName string `map:"model_name,omitempty"`
	// Sampling temperature to use.
	Temperature float32 `map:"temperature,omitempty"`
	// The maximum number of tokens to generate in the completion.
	// -1 returns as many tokens as possible given the prompt and
	//the models maximal context size.
	MaxTokens int `map:"max_tokens,omitempty"`
	// Total probability mass of tokens to consider at each step.
	TopP float32 `map:"top_p,omitempty"`
	// Penalizes repeated tokens.
	PresencePenalty float32 `map:"presence_penalty,omitempty"`
	// Penalizes repeated tokens according to frequency.
	FrequencyPenalty float32 `map:"frequency_penalty,omitempty"`
	// How many completions to generate for each prompt.
	N int `map:"n,omitempty"`
	// BaseURL is the base URL of the OpenAI service.
	BaseURL string `map:"base_url,omitempty"`
	// OrgID is the organization ID for accessing the OpenAI service.
	OrgID string `map:"org_id,omitempty"`
	// Stream indicates whether to stream the results or not.
	Stream bool `map:"stream,omitempty"`
	// MaxRetries represents the maximum number of retries to make when generating.
	MaxRetries uint `map:"max_retries,omitempty"`
}

OpenAIOptions contains the options for the OpenAI chat model.

type Palm added in v0.0.36 

type Palm struct {
	schema.Tokenizer
	// contains filtered or unexported fields
}

Palm is a struct representing the PALM language model.

func NewPalm added in v0.0.36 

func NewPalm(client PalmClient, optFns ...func(o *PalmOptions)) (*Palm, error)

NewPalm creates a new instance of the PALM language model.

func (*Palm) Callbacks added in v0.0.36 

func (cm *Palm) Callbacks() []schema.Callback

Callbacks returns the registered callbacks of the model.

func (*Palm) Generate added in v0.0.36 

func (cm *Palm) Generate(ctx context.Context, messages schema.ChatMessages, optFns ...func(o *schema.GenerateOptions)) (*schema.ModelResult, error)

Generate generates text based on the provided chat messages and options.

func (*Palm) InvocationParams added in v0.0.36 

func (cm *Palm) InvocationParams() map[string]any

InvocationParams returns the parameters used in the model invocation.

func (*Palm) Type added in v0.0.36 

func (cm *Palm) Type() string

Type returns the type of the model.

func (*Palm) Verbose added in v0.0.36 

func (cm *Palm) Verbose() bool

Verbose returns the verbosity setting of the model.

type PalmClient added in v0.0.36 

type PalmClient interface {
	GenerateMessage(ctx context.Context, req *generativelanguagepb.GenerateMessageRequest, opts ...gax.CallOption) (*generativelanguagepb.GenerateMessageResponse, error)
}

PalmClient is the interface for the PALM client.

type PalmOptions added in v0.0.36 

type PalmOptions struct {
	*schema.CallbackOptions `map:"-"`
	schema.Tokenizer        `map:"-"`

	// ModelName is the name of the Palm chat model to use.
	ModelName string `map:"model_name,omitempty"`

	// Temperature is the sampling temperature to use during text generation.
	Temperature float32 `map:"temperature,omitempty"`

	// TopP is the total probability mass of tokens to consider at each step.
	TopP float32 `map:"top_p,omitempty"`

	// TopK determines how the model selects tokens for output.
	TopK int32 `map:"top_k"`

	// CandidateCount specifies the number of candidates to generate during text completion.
	CandidateCount int32 `map:"candidate_count"`
}

PalmOptions is the options struct for the PALM chat model.

Source Files

View all Source files

Jump to

Keyboard shortcuts

? : This menu
/ : Search site
f or F : Jump to
y or Y : Canonical URL
go.dev uses cookies from Google to deliver and enhance the quality of its services and to analyze traffic. Learn more.