Skip to content
[Error - Langchain ...
 
Notifications
Clear all

[Error - Langchain Integration - AI Tool Agent] Tool input mismatch with DynamicStructuredTool.call schema

20 Posts
6 Users
0 Reactions
6 Views
Wayne
(@wayne)
Posts: 7
Active Member
 

Thanks Ria, is this included in release 1.64.1? There aren't any release notes available yet.

 
Posted : 23/10/2024 4:09 pm
ria
 ria
(@ria)
Posts: 48
Eminent Member
 

Hi

This should be released today and will then be in 1.64.2

:crossed_fingers:

 
Posted : 24/10/2024 7:53 am
Artur_Ferreira_Padua
(@artur_ferreira_padua)
Posts: 4
Active Member
Topic starter
 

Hello ria!
Thank you for reaching out.
Unfortunately, the fix you provided won't be helpful as I'm primarily using VertexAI and Anthropic as LLM Providers. Additionally, we updated to version 1.64.1 this week, and the bug persists.

Upon reviewing the logs, I've observed a consistent pattern even with different JSON Schemas and LLMs. On the initial run, the Agent successfully generates a correct JSON schema with the data structured properly. However, the agent executes the LLM a second time, which then leads to the error. The autofixing output parser has also been unable to resolve this. I will provide as much data as possible in the examples, and if you require anything further, I can send it via direct message.

Here is the workflow:

Here is the JSON the agent structured on the first run:
{
"action": "Final Answer",
"actioninput": "{n "structuredoutput": {n "structuredoutputarray": [n {n "name": "João da Silva",n "role": "Titular",n "maritalstatus": "Casado",n "healthplan": "ALICE CONFORTO",n "dentalplan": "",n "coparticipation": "false",n "phone": "1199709875",n "email": "joaosilva.96@hotmail.com",n "companyrole": "",n "dependentof": "",n "relationship": ""n },n {n "name": "Maria fatima",n "role": "Dependente",n "maritalstatus": "Casado",n "healthplan": "ALICE CONFORTO",n "dentalplan": "",n "coparticipation": "false",n "phone": "",n "email": "",n "companyrole": "",n "dependent_of": "João da Silva",n "relationship": ""n }n ]n }n}"
}

Here is the result after the second run and using the autofix output parser. By the way, I’ve already tried without using the autofix.
[
{
"output": [
{
},
{
}
]
}
]

The undesirable behavior is either an empty JSON as shown above or this generic error message found in the log of the second run on the LLM:
"". Error: SyntaxError: Unexpected token '`', "json
{
“… is not valid JSON”

Could you please assist us as soon as possible? This error is causing instability in our application, and based on my research, it appears to be related to an instruction on how the AI Agent parses JSON, which seems native to callin.io.

Please provide the rewritten markdown content *it should be in the markdown format.

 
Posted : 24/10/2024 5:20 pm
Antoni_Camilo
(@antoni_camilo)
Posts: 1
New Member
 

I resolved the issue by switching from Chat-GPT to Groq.

 
Posted : 13/11/2024 4:27 pm
system
(@system)
Posts: 332
Reputable Member
 

This thread was automatically closed 90 days after the last response. New responses are no longer permitted.

 
Posted : 11/02/2025 4:27 pm
Page 2 / 2
Share: