Karma AI Package Documentation

This package provides an interface for interacting with various AI models and services. It includes functionality for chat completions, single prompt generation, and streaming chat completions.

Installation

go get github.com/MelloB1989/karma/ai

Usage

Creating a KarmaAI Instance

To use the AI functionalities, first create a KarmaAI instance:

import "github.com/MelloB1989/karma/ai"
 
kai := ai.NewKarmaAI(ai.Llama3_8B,
    ai.WithMaxTokens(100),
    ai.WithTemperature(0.5),
    ai.WithTopP(0.9),
    ai.WithSystemMessage("Your system message here"))

Chat Completion

To perform a chat completion:

import (
    "fmt"
    "github.com/MelloB1989/karma/ai"
    "github.com/MelloB1989/karma/models"
)
 
kai := ai.NewKarmaAI(ai.ApacClaude3_5Sonnet20240620V1,
    ai.WithSystemMessage("I am Kartik Deshmukh."),
    ai.WithTemperature(0.5),
    ai.WithMaxTokens(10),
    ai.WithTopP(0.9))
 
response, err := kai.ChatCompletion(models.AIChatHistory{
    Messages: []models.AIMessage{
        {
            Message: "Hello",
            Role:    models.User,
        },
    },
})
if err != nil {
    panic(err)
}
fmt.Println(response.AIResponse)

Single Prompt Generation

To generate a response from a single prompt:

kai := ai.NewKarmaAI(ai.Llama3_8B,
    ai.WithUserPrePrompt("This is Kartik Deshmukh."),
    ai.WithTemperature(0.5),
    ai.WithMaxTokens(10),
    ai.WithTopP(0.9))
 
response, err := kai.GenerateFromSinglePrompt("Hello!")
if err != nil {
    panic(err)
}
fmt.Println(response.AIResponse)

Streaming Chat Completion

To perform a streaming chat completion:

import (
    "fmt"
    "github.com/sashabaranov/go-openai"
    "github.com/MelloB1989/karma/ai"
    "github.com/MelloB1989/karma/models"
)
 
chunkHandler := func(chunk openai.ChatCompletionChunk) {
    fmt.Print(chunk.Choices[0].Delta.Content)
}
 
kai := ai.NewKarmaAI(ai.ChatModelChatgpt4oLatest, ai.WithUserPrePrompt("I am Kartik Deshmukh."))
 
response, err := kai.ChatCompletionStream(models.AIChatHistory{
    Messages: []models.AIMessage{
        {
            Message: "Hello, how to create a new file in Go?",
            Role:    models.User,
        },
    },
}, chunkHandler)
 
if err != nil {
    panic(err)
}
fmt.Println(response.AIResponse)

CLI Chat Implementation

Here's an example of how to implement a CLI chat interface:

func cliChatImplementation() {
    kai := ai.NewKarmaAI(ai.Llama3_8B,
        ai.WithMaxTokens(100),
        ai.WithTemperature(0.5),
        ai.WithTopP(0.9),
        ai.WithSystemMessage("Your system message here"))
 
    chatHistory := models.AIChatHistory{Messages: []models.AIMessage{}}
    var message string
 
    for message != "exit" {
        fmt.Print("You: ")
        fmt.Scanln(&message)
        chatHistory.Messages = append(chatHistory.Messages, models.AIMessage{Message: message, Role: models.User})
 
        response, err := kai.ChatCompletion(chatHistory)
        if err != nil {
            panic(err)
        }
 
        fmt.Println("AI: ", response.AIResponse)
        chatHistory.Messages = append(chatHistory.Messages, models.AIMessage{Message: response.AIResponse, Role: models.Assistant})
    }
 
    fmt.Println(chatHistory)
}

This implementation creates a simple command-line interface for chatting with the AI. The chat continues until the user types "exit".

Note: The example provided in the code includes content that may not be suitable for all audiences. Please adjust the system message and behavior as needed for your specific use case.

Updatefamilymemberdetails Llm Context