Replies: 1 comment 12 replies
-
|
The issue you're encountering seems to be related to the chaining of the two tools, where the output from the Here are a few steps to troubleshoot and resolve this issue:
By following these steps, you should be able to identify where the data is not being correctly passed between the tools and resolve the issue. |
Beta Was this translation helpful? Give feedback.
-
|
To integrate the execution logic within
Here's an example: from langchain_core.runnables import Runnable, RunnablePassthrough
from pydantic import BaseModel, Field
from typing import Dict, Any, Type
# Define the input and output schemas for each tool
class Tool1Input(BaseModel):
input_data: str
class Tool1Output(BaseModel):
processed_data: str
class Tool2Input(BaseModel):
input_data: str
# Define the first tool
class Tool1(BaseTool):
args_schema: Type[BaseModel] = Tool1Input
def _run(self, input_data: str) -> Tool1Output:
# Process the input data
processed_data = input_data.upper() # Example processing
return Tool1Output(processed_data=processed_data)
# Define the second tool
class Tool2(BaseTool):
args_schema: Type[BaseModel] = Tool2Input
def _process_dataflow(self, dataflow_payload: Dict[str, Any]) -> Dict[str, Any]:
# Use the dataflow_payload to perform some action
return {"status": "success", "payload": dataflow_payload}
async def _arun(self, input_data: str) -> Dict[str, Any]:
# Validate input using Tool2's schema
validated_input = self.args_schema(input_data=input_data)
# Create and execute the chain
tool1 = Tool1()
tool1_output = tool1._run(validated_input.input_data)
# Process the output from Tool1
result = self._process_dataflow({"dataflow_payload": tool1_output.dict()})
return result
# Example usage within a larger framework
async def execute_tool2(input_data: Dict[str, Any]):
tool2 = Tool2()
result = await tool2._arun(**input_data)
return result
# Example input
input_data = {"input_data": "example input"}
# Run the example
import asyncio
asyncio.run(execute_tool2(input_data))In this setup, |
Beta Was this translation helpful? Give feedback.
-
|
@dosu how do we make sure the output from tool1 passes to tool2 and we can define that inside tool2. Also I think langchain does a pydantic validation based on args_schema before the _run method. |
Beta Was this translation helpful? Give feedback.
-
|
To ensure that the output from
Here's an example: from langchain_core.runnables import Runnable
from pydantic import BaseModel, Field
from typing import Dict, Any, Type
# Define the input and output schemas for each tool
class Tool1Input(BaseModel):
input_data: str
class Tool1Output(BaseModel):
processed_data: str
class Tool2Input(BaseModel):
input_data: str
# Define the first tool
class Tool1(BaseTool):
args_schema: Type[BaseModel] = Tool1Input
def _run(self, input_data: str) -> Tool1Output:
# Process the input data
processed_data = input_data.upper() # Example processing
return Tool1Output(processed_data=processed_data)
# Define the second tool
class Tool2(BaseTool):
args_schema: Type[BaseModel] = Tool2Input
def _process_dataflow(self, dataflow_payload: Dict[str, Any]) -> Dict[str, Any]:
# Use the dataflow_payload to perform some action
return {"status": "success", "payload": dataflow_payload}
async def _arun(self, input_data: str) -> Dict[str, Any]:
# Validate input using Tool2's schema
validated_input = self.args_schema(input_data=input_data)
# Create and execute Tool1
tool1 = Tool1()
tool1_output = tool1._run(validated_input.input_data)
# Process the output from Tool1
result = self._process_dataflow({"dataflow_payload": tool1_output.dict()})
return result
# Example usage within a larger framework
async def execute_tool2(input_data: Dict[str, Any]):
tool2 = Tool2()
result = await tool2._arun(**input_data)
return result
# Example input
input_data = {"input_data": "example input"}
# Run the example
import asyncio
asyncio.run(execute_tool2(input_data))In this setup, |
Beta Was this translation helpful? Give feedback.
-
|
@dosu where can I look for more related information? |
Beta Was this translation helpful? Give feedback.
-
|
I didn't find any similar threads. |
Beta Was this translation helpful? Give feedback.
-
Consider these two tools below. I want to chain the output of payload builder tool to dataflow creator tool. However, creator tool does not seem to receive the payload created by the builder tool and receives an empty object. Payload builder tool creates the right payload as can be seen in the error logs below
Pasting the error trace for reference.
Beta Was this translation helpful? Give feedback.
All reactions