After deploying n8n locally, I encountered a problem as title when calling a local large model. The workflow is a simple dialogue workflow, that is, a message and an AI Agent. In the model call section, it is already displayed: connection tested successfully, so I guess the problem lies in the AI Agent module. I hope someone can give some suggestions for modification. Thank you!
#Problem in node 'AI Agent':Cannot read properties of undefined (reading 'content')
signal grotto
lean pebble
This means your LLM didn't return a proper content property as the response
So most likely, your LLM responded without content (an error in the generation possibly)
signal grotto
This means your LLM didn't return a proper content property as the response
Thx for ur reply. Actually, no matter what I change the model, even if I write the wrong info in model id, it still shows the problem as the title, so that's why I think the issue happens in AI Agent , it did not transfer the meaasge to the model from AI Agent
merry zephyr
After deploying n8n locally, I encountered a problem as title when calling a loc...
can you show the workflow and tell which model exactly?
signal grotto
can you show the workflow and tell which model exactly?
It's very simple and we use the OpenAI chat model to load our local model, which has shown connection tested successfully after fulfilling the API-Key and Base-URL
merry zephyr
It's very simple and we use the OpenAI chat model to load our local model, which...
oh so something wrong with your config? i use lm studio for this purpose
or maybe small context window
signal grotto
or maybe small context window
No,sry cuz i cannot show u the Intranet. it's only the workflow and in Intrannet, everything is ok but when it runs, the issue happens as the title.
merry zephyr
No,sry cuz i cannot show u the Intranet. it's only the workflow and in Intrannet...
i am not sure if i get you but can you show me the actual error?
tough gulch
What is the message and what are the prompts being used
signal grotto
What is the message and what are the prompts being used
just hello and no prompts
merry zephyr
no system prompt? thats why maybe
signal grotto
no system prompt? thats why maybe
Only test it can work in Intranet
and the problem is showed in picture:
signal grotto
The workflow is as same as in the Intranet. We use the OpenAI chat model to load our local model, which has shown connection tested successfully. When it runs, only the AI Agent, which I use the red circle to show, has the issue as the title, and the detailed info has been shown in the second pic.
