Notifications
Clear all
Topic starter
I've utilized Cole Medin’s local-ai-packaged callin.io workflow, which I've customized for local file uploads.
I currently have two small, simple text documents uploaded to my Qdrant vector store for testing accuracy and other aspects.
Curiously, the initial chat response consistently appears as:
< pythontag >tool call: getchristmas_plans()
If I submit the same chat question or prompt again, the behavior seems to normalize.
When I ask a new question or prompt, the first response consistently includes some leading text, such as (this is an example of one such response):
Tool call output: John’s present ideas are:
- Lego Creator Expert - Ferrari 250 GT Berlinetta
- Lego Ideas - Old Fishing Store
This information is actually incorrect. If I resubmit the prompt, the leading text is absent, and the correct information is retrieved from my text files.
I made a mistake earlier. The correct output from the tool call is Ferrari Lego Technic and iPad Mini.
What exactly are these tool calls? And why do they seem to occur only on the first question or prompt?
Information on your callin.io setup
- callin.io version: 1.67.1
- Database (default: SQLite): Postgres + Qdrant
- callin.io EXECUTIONS_PROCESS setting (default: own, main): ???
- Running callin.io via (Docker, npm, callin.io cloud, desktop app): Docker
- Operating system: Macbook M1 with macOS 18.1
Please provide the rewritten markdown content *it should be in the markdown format.
Posted : 13/12/2024 4:21 pm
Hi there,
The output from the Agent will depend on the model you are utilizing and the instructions you provide.
In your current workflow, the agent is not receiving any instructions, so it will simply respond based on the base model's default instructions, which appear to be to call the get_christmas_plans() function. Which model are you currently using?
The tool calling functionality is what the agent uses to execute a connected tool (in your case, the qdrant vector store) and return the output it receives from that tool. However, the agent needs to be instructed on how to use the tool correctly; otherwise, you might encounter unexpected behavior like this.
If you want the agent to call the vector store tool correctly, ensure that the name and description are sufficiently descriptive for the model to understand when to invoke it.
You can learn more about the concept of tool calling here:
Posted : 19/12/2024 6:20 am
This thread was automatically closed 90 days following the last response. New replies are no longer permitted.
Posted : 19/03/2025 6:20 am