Sharing variables between nodes

[sofyanajridi]

Hello,

I wanted to know if it is possible to share variables between nodes. E.g lets say I have an initial flow that sets up a AzureOpenAI connection (not using the existing LLM node):

Node 1:

client = AzureOpenAI(....)
r = client.embeddings.create(...)

Now for Node 2, I would like to reuse the same client from Node 1. Is this possible or should I always have to redefine the variable?

If this is not possible, I would think a feature request like the following could be interesting?

Node 1:

client = AzureOpenAI(....)
global_config["client"] = client

r = client.embeddings.create(...)

Node 2:

client =global_config["client"] 
r = client.embeddings.create(...)

[D-W-]

Hi @sofyanajridi , currently it's not supported in DAG flow. But you can try our class-based flex flow and define your AzureOpenAI client inside __init__ method, each line run can share same client to avoid create duplicate clients.

[sofyanajridi]

Thank you, will have a look!