Page 1 of 1

You can select the latest stable model in the ai agent settings

Posted: Thu Dec 05, 2024 6:28 am
by mahmud212
Make sure the tools listed match what you expect to be available in the node's decision-making process.
Make sure that the tool names are spelled correctly at your prompt to ensure that the node can successfully execute the specified action.
The autonomous node tools section.
The autonomous node tools section.


3. Check the Iterations tab
Typically, the autonomous node attempts to sri lanka phone numbers execute all instructions in one or two iterations. The number of iterations depends on the complexity of the statement and how the node parses it.

Image

For more complex tasks, the node may need several iterations to collect data, make decisions, or search for external information.

If you check the iterations tab (or all tab), you can understand:

how many iterations were necessary for the node to reach its final decision.
Which caused the node to take multiple steps (e.g. getting additional data from tools like knowledgeagent.knowledgequery or browser.websearch).
Why a specific result has been achieved.
You can check what happens in each iteration.
You can check what happens in each iteration.
Common problems
model size
the autonomous node could not be following your prompt, executing part of the prompt instead of all of it, or calling "workflowqueue" without calling the "workflowexecuteall" tools.

It makes sense to always change the size of the llm autonomous node to a smaller model - because it is cheaper - but that comes at a cost.

A smaller llm could result in parts of the prompt being truncated, specifically the definition wrapper that botpress adds to make sure the llm understands how the cards work, what parameters are needed, etc. Without this, the bot would not know how to act correctly.

llmz version
Always make sure you are using the latest stable version of llmz. It is the autonomous engine that directs the autonomous node to function.

It also contains bug fixes, making prompts more llms agnostic.




example: code creation diagnosis
Let's say that an autonomous node is generating code but is not following the prompt correctly. Here's how you can fix it:

inspect: Checks what instructions the node follows. Do you correctly understand the code generation request?
Tools: Verify that the node has access to the necessary tools (for example, code generation tools or knowledge base query tools). Make sure the notice explicitly refers to these tools.
Iterations: Look at the Iterations tab to see how the node got to the point of generating the code. Did you take one or several steps? Did you consult a knowledge base first or try to generate code immediately?
Solution: If the bot is not generating code correctly:

Make sure the tool being used for code generation is correctly referenced in the prompt.
Adjust the instructions so that the node is guided to follow specific steps, such as first retrieving relevant knowledge before attempting code generation.


sample code generated from an autonomous node.
Sample code generated from an autonomous node.


complete application example
**important: query knowledge base is to be used only for support questions related explicitly to student courses, and not for general features or pricing inquiries.