Add safeguards for infinite loops and unresolvable tasks #28441
Closed
+166
−3
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
added 4 commits
November 27, 2024 17:05
modified: libs/core/langchain_core/prompts/prompt.py modified: libs/langchain/langchain/agents/react/agent.py - Updated the prompt template to include explicit termination instructions: 1. Avoid generating its own "User Input." 2. Gracefully terminate when progress cannot be made. 3. Use a fallback message for incomplete tasks. - Enhanced the output parser: - Detects repetitive reasoning and circular logic. - Handles invalid outputs and iteration limits. - Returns appropriate error messages for unresolvable scenarios. - Configured the AgentExecutor: - Set a maximum iteration limit to prevent infinite loops. - Ensures the agent stops reasoning after reaching the limit.
…ct agent - Updated `output_parsers/base.py` to detect and handle repetitive reasoning. - Enhanced `prompts/prompt.py` with constraints to avoid infinite loops and user input generation. - Improved `agents/react/agent.py` with logic for graceful termination and fallback handling.
deleted: libs/community/langchain_community/tools/fallback.py new file: libs/core/tests/integration_tests/test_react_agent.py new file: libs/core/tests/unit_tests/output_parsers/test_base.py new file: libs/core/tests/unit_tests/output_parsers/test_prompt.py
dosubot
bot
added
the
size:L
|
The latest updates on your projects. Learn more about Vercel for Git ↗︎
|
dosubot
bot
added
Ɑ: agent
…hain into react_agent_test
modified: libs/langchain/langchain/agents/react/agent.py
modified: libs/langchain/langchain/agents/react/agent.py
modified: libs/langchain/langchain/agents/react/agent.py
added 8 commits
December 3, 2024 16:41
modified: libs/langchain/langchain/agents/output_parsers/react_single_input.py
modified: libs/core/langchain_core/prompts/prompt.py
modified: libs/core/langchain_core/prompts/prompt.py
modified: libs/core/langchain_core/prompts/prompt.py modified: libs/core/tests/unit_tests/output_parsers/test_prompt.py
added 2 commits
December 3, 2024 18:14
modified: libs/core/langchain_core/output_parsers/base.py modified: libs/core/langchain_core/prompts/prompt.py
added 4 commits
December 3, 2024 18:34
modified: libs/core/langchain_core/agents.py modified: libs/core/langchain_core/output_parsers/base.py modified: libs/core/langchain_core/prompts/prompt.py
modified: libs/core/langchain_core/agents.py modified: libs/core/langchain_core/output_parsers/base.py modified: libs/core/langchain_core/prompts/prompt.py
|
hey there! I believe this would be better implemented by the user in a langgraph agent: https://langchain-ai.github.io/langgraph/tutorials/introduction/ This out of the box AgentExecutor does have shortcomings, hence the greater degree of control in langgraph. I don't think this change to AgentExecutor and supporting data objects is desirable, so I will close this PR |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Description
In the issue referenced below, the agent generates it's own "User Input" prompts during execution. Rather than using provided tools it requests input data from the user whick it cannot receive. This results in Inifinite loops, ultimately reaching the iteration limit, and provides the following error message. "Agent stopped due to iteration limit". In this PR we provided a proposed solution with associated unit and integration tests, for the specific instances of infinite loops from user input.
Issue: #25696
Dependencies:
No additional dependencies
Twitter Handle:
None