検索

お知らせ
· 2025年2月5日

[Video] Open Source and Proprietary LLMs

Hi Community!

We're happy to share the next video in the series dedicated to Gen AI on our InterSystems Developers YouTube:

⏯ Open Source and Proprietary LLMs

Explore the comparison between open-source and proprietary AI models. Learn that proprietary models generally rank higher in performance on leaderboards like Chatbot Arena, though open-source models are not far behind. We will explain that open-source models tend to lag behind proprietary ones in terms of time but eventually reach comparable performance levels. This means businesses can start with proprietary models and later transition to open-source alternatives as they improve. The video emphasizes considering open-source models as part of an AI strategy due to their evolving capabilities and potential cost benefits.

🗣  Presenter@Don Woodlock, Vice President, Healthcare Solutions Development, InterSystems

Enjoy watching, and look forward to more videos! 👍

1 Comment
ディスカッション (1)2
続けるにはログインするか新規登録を行ってください
質問
· 2025年2月5日

How to check if the system user with the name "John" already exists in IRIS via SQL?

Hi Folks!

Have a very simple question, I hope :)

How can I make sure that user 'John' is already created in the IRIS system?

Preferabbly via SQL?

thanks a lot!

4 Comments
ディスカッション (4)3
続けるにはログインするか新規登録を行ってください
記事
· 2025年2月5日 2m read

Streamlining Interoperability with Embedded Python in InterSystems IRIS

Interoperability of systems ensures smooth workflow and management of data in today's connected digital world. InterSystems IRIS extends interoperability a notch higher with its Embedded Python feature, which lets developers seamlessly integrate Python scripts into the IRIS components, like services, operations, and custom functions.


Why Embedded Python in IRIS?
Embedded Python brings the flexibility of Python into the powerful world of InterSystems IRIS and provides the following: Access to Python Libraries: Leverage popular libraries such as pandas, NumPy, and requests for advanced data processing.
Ease of Use: Python simplifies the implementation of complex logic.
Efficient Interoperability: Develop services, operations, and transformations directly within IRIS, reducing the need for external tools.

Below is the code I tried to implement for a python business operation:

from iris import irisnative  

class SamplePythonOperation:  
    def on_message(self, request):  
        # Connect to IRIS  
        conn = irisnative.createConnection("localhost", 1972, "USER", "_SYSTEM", "SYS")  
        iris_obj = irisnative.createIRIS(conn)  

        # Log the request message  
        iris_obj.set("Received: " + request.get("MessageText"), "MyApp", "Log")  

        # Process the request  
        response = "Hello, " + request.get("MessageText")  
        return {"Result": response}  

Embedded Python in InterSystems IRIS allows developers to leverage the flexibility of Python with the robustness of IRIS for interoperability solutions. In transforming data, building APIs, or integrating external systems, traditional expectations are that Embedded Python is a revolution for modern applications.

Use Embedded Python today, and discover new avenues of seamless interoperability! 

4 Comments
ディスカッション (4)1
続けるにはログインするか新規登録を行ってください
記事
· 2025年2月5日 6m read

Présentation du framework web Python Stream



Salut la Communauté!

Dans cet article, je présenterai le framework web Python Streamlit.

Ci-dessous, vous trouverez les sujets que nous aborderons:

  • 1-Introduction au framework web Streamlit
  • 2-Installation du module Streamlit
  • 3-Lancement de l'application Streamlit
  • 4-Commandes de base de Streamlit
  • 5-Affichage du contenu multimédia 
  • 6-Widgets d'input
  • 7-Affichage des progrès et de l'état
  • 8-Barre latérale et conteneur
  • 9-Visualisation des données
  • 10-Affichage de DataFrame

 

Commençons donc par le premier sujet.

ディスカッション (0)2
続けるにはログインするか新規登録を行ってください
記事
· 2025年2月5日 2m read

Retornar imagens usando pesquisa de vetor (2)

Você precisa instalar o aplicativo primeiro. Se não estiver instalado, por favor, consulte o artigo anterior.

Demonstração do aplicativo

Após executar com sucesso o aplicativo de busca de vetores de imagem de íris, alguns dados precisam ser armazenados para suportar a recuperação de imagens, pois não são inicializados na biblioteca.

Armazenamento de imagens

Primeiramente, arraste e solte a imagem ou clique no ícone de upload, selecione a imagem e clique no botão de upload para enviar e vetorizá-la. Este processo pode ser um pouco lento.

Esse processo envolve o uso de Python incorporado para chamar o modelo CLIP e vetorizar a imagem em dados vetoriais de 512 dimensões.

ClassMethod LoadClip(imagePath) As %String [ Language = python ]
{
import torch
import clip
from PIL import Image

device = "cuda" if torch.cuda.is_available() else "cpu"
model, preprocess = clip.load("ViT-B/32", device=device)

image = preprocess(Image.open(imagePath)).unsqueeze(0).to(device)

with torch.no_grad():
    image_features = model.encode_image(image)
return str(image_features[0].tolist())
}

Ver dados vetoriais

Observando os dados vetoriais no Portal de Administração

Consulta de vetores

O processo de consulta de vetores utiliza VECTOR_COSINE para recuperar a similaridade dos dados vetoriais, o que pode ser realizado através do seguinte SQL.

select top ? id Image,VECTOR_COSINE(TO_VECTOR(ImageVector,double),(select TO_VECTOR(ImageVector,double) from VectorSearch_DB.ImageDB where id= ? )) Similarity from VectorSearch_DB.ImageDB order by Similarity desc

 

Após enviar várias imagens, selecione a consulta e a imagem com a maior similaridade à esquerda será exibida à direita. Clique em "próximo" para visualizar a próxima página. A similaridade entre as duas imagens diminuirá gradualmente a cada clique.

 

Esta aplicação é uma demonstração da recuperação de vetores de imagem. Se você estiver interessado, pode verificar oOpen Exchange ou visitar o Github tpara ver mais informações.

ディスカッション (0)1
続けるにはログインするか新規登録を行ってください