Condividi tramite

Integrare gli SDK di inferenza con Foundry Local

Importante

  • Foundry Local è disponibile in anteprima. Le versioni di anteprima pubblica consentono l'accesso anticipato alle funzionalità in fase di distribuzione attiva.
  • Funzionalità, approcci e processi possono modificare o avere funzionalità limitate, prima della disponibilità generale (GA).

Foundry Local si integra con vari SDK di inferenza, ad esempio OpenAI, Azure OpenAI, Langchain e così via. Questa guida illustra come connettere le applicazioni ai modelli di intelligenza artificiale in esecuzione in locale usando gli SDK più diffusi.

Prerequisiti

Installare pacchetti pip

Installare i pacchetti Python seguenti:

pip install openai
pip install foundry-local-sdk

Suggerimento

È consigliabile usare un ambiente virtuale per evitare conflitti di pacchetti. È possibile creare un ambiente virtuale usando venv o conda.

Usare OpenAI SDK con Foundry Local

L'esempio seguente illustra come usare OpenAI SDK con Foundry Local. Il codice inizializza il servizio locale Foundry, carica un modello e genera una risposta usando OpenAI SDK.

Copiare e incollare il codice seguente in un file Python denominato app.py:

import openai
from foundry_local import FoundryLocalManager

# By using an alias, the most suitable model will be downloaded 
# to your end-user's device. 
alias = "phi-3.5-mini"

# Create a FoundryLocalManager instance. This will start the Foundry
# Local service if it is not already running and load the specified model.
manager = FoundryLocalManager(alias)
# The remaining code uses the OpenAI Python SDK to interact with the local model.
# Configure the client to use the local Foundry service
client = openai.OpenAI(
    base_url=manager.endpoint,
    api_key=manager.api_key  # API key is not required for local usage
)
# Set the model to use and generate a response
response = client.chat.completions.create(
    model=manager.get_model_info(alias).id,
    messages=[{"role": "user", "content": "What is the golden ratio?"}]
)
print(response.choices[0].message.content)

Eseguire il codice usando il comando seguente:

python app.py

Streaming della risposta

Se si vuole ricevere una risposta di streaming, è possibile modificare il codice nel modo seguente:

import openai
from foundry_local import FoundryLocalManager

# By using an alias, the most suitable model will be downloaded 
# to your end-user's device.
alias = "phi-3.5-mini"

# Create a FoundryLocalManager instance. This will start the Foundry 
# Local service if it is not already running and load the specified model.
manager = FoundryLocalManager(alias)

# The remaining code us es the OpenAI Python SDK to interact with the local model.

# Configure the client to use the local Foundry service
client = openai.OpenAI(
    base_url=manager.endpoint,
    api_key=manager.api_key  # API key is not required for local usage
)

# Set the model to use and generate a streaming response
stream = client.chat.completions.create(
    model=manager.get_model_info(alias).id,
    messages=[{"role": "user", "content": "What is the golden ratio?"}],
    stream=True
)

# Print the streaming response
for chunk in stream:
    if chunk.choices[0].delta.content is not None:
        print(chunk.choices[0].delta.content, end="", flush=True)

È possibile eseguire il codice usando lo stesso comando di prima:

python app.py

Usare requests con Foundry Local

# Install with: pip install requests
import requests
import json
from foundry_local import FoundryLocalManager

# By using an alias, the most suitable model will be downloaded 
# to your end-user's device. 
alias = "phi-3.5-mini"

# Create a FoundryLocalManager instance. This will start the Foundry
# Local service if it is not already running and load the specified model.
manager = FoundryLocalManager(alias)

url = manager.endpoint + "/chat/completions"

payload = {
    "model": manager.get_model_info(alias).id,
    "messages": [
        {"role": "user", "content": "What is the golden ratio?"}
    ]
}

headers = {
    "Content-Type": "application/json"
}

response = requests.post(url, headers=headers, data=json.dumps(payload))
print(response.json()["choices"][0]["message"]["content"])

Installare pacchetti Node.js

È necessario installare i pacchetti di Node.js seguenti:

npm install openai
npm install foundry-local-sdk

Foundry Local SDK consente di gestire il servizio locale Foundry e i modelli.

Usare OpenAI SDK con Foundry Local

L'esempio seguente illustra come usare OpenAI SDK con Foundry Local. Il codice inizializza il servizio locale Foundry, carica un modello e genera una risposta usando OpenAI SDK.

Copiare e incollare il codice seguente in un file JavaScript denominato app.js:

import { OpenAI } from "openai";
import { FoundryLocalManager } from "foundry-local-sdk";

// By using an alias, the most suitable model will be downloaded 
// to your end-user's device.
// TIP: You can find a list of available models by running the 
// following command in your terminal: `foundry model list`.
const alias = "phi-3.5-mini";

// Create a FoundryLocalManager instance. This will start the Foundry 
// Local service if it is not already running.
const foundryLocalManager = new FoundryLocalManager()

// Initialize the manager with a model. This will download the model 
// if it is not already present on the user's device.
const modelInfo = await foundryLocalManager.init(alias)
console.log("Model Info:", modelInfo)

const openai = new OpenAI({
  baseURL: foundryLocalManager.endpoint,
  apiKey: foundryLocalManager.apiKey,
});

async function generateText() {
  const response = await openai.chat.completions.create({
    model: modelInfo.id,
    messages: [
      {
        role: "user",
        content: "What is the golden ratio?",
      },
    ],
  });

  console.log(response.choices[0].message.content);
}

generateText();

Eseguire il codice usando il comando seguente:

node app.js

Risposte in streaming

Se si vogliono ricevere risposte in streaming, è possibile modificare il codice nel modo seguente:

import { OpenAI } from "openai";
import { FoundryLocalManager } from "foundry-local-sdk";

// By using an alias, the most suitable model will be downloaded 
// to your end-user's device.
// TIP: You can find a list of available models by running the 
// following command in your terminal: `foundry model list`.
const alias = "phi-3.5-mini";

// Create a FoundryLocalManager instance. This will start the Foundry 
// Local service if it is not already running.
const foundryLocalManager = new FoundryLocalManager()

// Initialize the manager with a model. This will download the model 
// if it is not already present on the user's device.
const modelInfo = await foundryLocalManager.init(alias)
console.log("Model Info:", modelInfo)

const openai = new OpenAI({
  baseURL: foundryLocalManager.endpoint,
  apiKey: foundryLocalManager.apiKey,
});

async function streamCompletion() {
    const stream = await openai.chat.completions.create({
      model: modelInfo.id,
      messages: [{ role: "user", content: "What is the golden ratio?" }],
      stream: true,
    });
  
    for await (const chunk of stream) {
      if (chunk.choices[0]?.delta?.content) {
        process.stdout.write(chunk.choices[0].delta.content);
      }
    }
}
  
streamCompletion();

Eseguire il codice usando il comando seguente:

node app.js

Usare l'API Fetch con Foundry Local

Se si preferisce usare un client HTTP come fetch, è possibile farlo come segue:

import { FoundryLocalManager } from "foundry-local-sdk";

// By using an alias, the most suitable model will be downloaded 
// to your end-user's device.
// TIP: You can find a list of available models by running the 
// following command in your terminal: `foundry model list`.
const alias = "phi-3.5-mini";

// Create a FoundryLocalManager instance. This will start the Foundry 
// Local service if it is not already running.
const foundryLocalManager = new FoundryLocalManager()

// Initialize the manager with a model. This will download the model 
// if it is not already present on the user's device.
const modelInfo = await foundryLocalManager.init(alias)
console.log("Model Info:", modelInfo)

async function queryModel() {
    const response = await fetch(foundryLocalManager.endpoint + "/chat/completions", {
        method: "POST",
        headers: {
            "Content-Type": "application/json",
        },
        body: JSON.stringify({
            model: modelInfo.id,
            messages: [
                { role: "user", content: "What is the golden ratio?" },
            ],
        }),
    });

    const data = await response.json();
    console.log(data.choices[0].message.content);
}

queryModel();

Risposte in streaming

Se si vogliono ricevere risposte in streaming usando l'API Fetch, è possibile modificare il codice nel modo seguente:

import { FoundryLocalManager } from "foundry-local-sdk";

// By using an alias, the most suitable model will be downloaded 
// to your end-user's device.
// TIP: You can find a list of available models by running the 
// following command in your terminal: `foundry model list`.
const alias = "phi-3.5-mini";

// Create a FoundryLocalManager instance. This will start the Foundry 
// Local service if it is not already running.
const foundryLocalManager = new FoundryLocalManager()

// Initialize the manager with a model. This will download the model 
// if it is not already present on the user's device.
const modelInfo = await foundryLocalManager.init(alias)
console.log("Model Info:", modelInfo)

async function streamWithFetch() {
    const response = await fetch(foundryLocalManager.endpoint + "/chat/completions", {
        method: "POST",
        headers: {
            "Content-Type": "application/json",
            Accept: "text/event-stream",
        },
        body: JSON.stringify({
            model: modelInfo.id,
            messages: [{ role: "user", content: "what is the golden ratio?" }],
            stream: true,
        }),
    });

    const reader = response.body.getReader();
    const decoder = new TextDecoder();

    while (true) {
        const { done, value } = await reader.read();
        if (done) break;

        const chunk = decoder.decode(value);
        const lines = chunk.split("\n").filter((line) => line.trim() !== "");

        for (const line of lines) {
            if (line.startsWith("data: ")) {
                const data = line.substring(6);
                if (data === "[DONE]") continue;

                try {
                    const json = JSON.parse(data);
                    const content = json.choices[0]?.delta?.content || "";
                    if (content) {
                        // Print to console without line breaks, similar to process.stdout.write
                        process.stdout.write(content);
                    }
                } catch (e) {
                    console.error("Error parsing JSON:", e);
                }
            }
        }
    }
}

// Call the function to start streaming
streamWithFetch();

Creare un progetto

Creare un nuovo progetto C# e accederne:

dotnet new console -n hello-foundry-local
cd hello-foundry-local

Installare pacchetti NuGet

Installare i pacchetti NuGet seguenti nella cartella del progetto:

dotnet add package Microsoft.AI.Foundry.Local --version 0.1.0
dotnet add package OpenAI --version 2.2.0-beta.4

Usare OpenAI SDK con Foundry Local

L'esempio seguente illustra come usare OpenAI SDK con Foundry Local. Il codice inizializza il servizio locale Foundry, carica un modello e genera una risposta usando OpenAI SDK.

Copiare e incollare il codice seguente in un file C# denominato Program.cs:

using Microsoft.AI.Foundry.Local;
using OpenAI;
using OpenAI.Chat;
using System.ClientModel;
using System.Diagnostics.Metrics;

var alias = "phi-3.5-mini";

var manager = await FoundryLocalManager.StartModelAsync(aliasOrModelId: alias);

var model = await manager.GetModelInfoAsync(aliasOrModelId: alias);
ApiKeyCredential key = new ApiKeyCredential(manager.ApiKey);
OpenAIClient client = new OpenAIClient(key, new OpenAIClientOptions
{
    Endpoint = manager.Endpoint
});

var chatClient = client.GetChatClient(model?.ModelId);

var completionUpdates = chatClient.CompleteChatStreaming("Why is the sky blue'");

Console.Write($"[ASSISTANT]: ");
foreach (var completionUpdate in completionUpdates)
{
    if (completionUpdate.ContentUpdate.Count > 0)
    {
        Console.Write(completionUpdate.ContentUpdate[0].Text);
    }
}

Eseguire il codice usando il comando seguente:

dotnet run

Creare un progetto

Creare un nuovo progetto Rust e accederne:

cargo new hello-foundry-local
cd hello-foundry-local

Installare crate

Installare i seguenti crate Rust usando Cargo:

cargo add foundry-local anyhow env_logger serde_json
cargo add reqwest --features json
cargo add tokio --features full

Aggiornare il main.rs file

Nell'esempio seguente viene illustrato come effettuare un'inferenza tramite una richiesta al servizio Foundry Locale. Il codice inizializza il servizio locale Foundry, carica un modello e genera una risposta usando la reqwest libreria.

Copiare e incollare il codice seguente nel file Rust denominato main.rs:

use foundry_local::FoundryLocalManager;
use anyhow::Result;

#[tokio::main]
async fn main() -> Result<()> {
    // Create a FoundryLocalManager instance with default options
    let mut manager = FoundryLocalManager::builder()
        .alias_or_model_id("qwen2.5-0.5b") // Specify the model to use   
        .bootstrap(true) // Start the service if not running
        .build()
        .await?;
    
    // Use the OpenAI compatible API to interact with the model
    let client = reqwest::Client::new();
    let endpoint = manager.endpoint()?;
    let response = client.post(format!("{}/chat/completions", endpoint))
        .header("Content-Type", "application/json")
        .header("Authorization", format!("Bearer {}", manager.api_key()))
        .json(&serde_json::json!({
            "model": manager.get_model_info("qwen2.5-0.5b", true).await?.id,
            "messages": [{"role": "user", "content": "What is the golden ratio?"}],
        }))
        .send()
        .await?;

    let result = response.json::<serde_json::Value>().await?;
    println!("{}", result["choices"][0]["message"]["content"]);
    
    Ok(())
}

Eseguire il codice usando il comando seguente:

cargo run

Passaggi successivi