[BUG] Crewai 0.120.0 Doesn’t Think

by ADMIN 35 views

Introduction

In this article, we will discuss a bug in the crewai 0.120.0 version, which is causing an error when using Claude AI sonet 3.7 with reasoning enabled. The error is related to the expectation of "thinking" but receiving text instead. We will go through the steps to reproduce the issue, the expected behavior, and possible solutions.

Description of the Issue

When using the crewai 0.120.0 version with Claude AI sonet 3.7 and reasoning enabled, an error is returned. The error message indicates that the LLM (Large Language Model) expects "thinking" but got text instead. This issue is not present in previous stable versions of crewai.

Steps to Reproduce the Issue

To reproduce the issue, follow these steps:

Step 1: Install crewai 0.120.0

!pip install “crewai==0.120.0”

Step 2: Instantiate LLM with Claude Sonnet 3.7

from crewai import LLM

model = LLM("claude:sonnet-3.7")
model.set_reasoning("high")

Step 3: Kick off the crew

Once the LLM is instantiated, kick off the crew. The crew will fail with the error mentioned above.

Expected Behavior

The expected behavior is that the crew should run normally, just like in previous stable versions of crewai. However, with the crewai 0.120.0 version, the crew fails with the error mentioned above.

Possible Solution

One possible solution to this issue is to investigate the difference between the 0.120.0 and 0.114.0 versions of crewai. By comparing the code and functionality of these two versions, we may be able to identify the cause of the issue and implement a fix.

Additional Context

This issue was also observed on AWS SageMaker, which suggests that the issue is not specific to a particular environment or setup.

Conclusion

In conclusion, the crewai 0.120.0 version has a bug that causes an error when using Claude AI sonet 3.7 with reasoning enabled. The error is related to the expectation of "thinking" but receiving text instead. By following the steps to reproduce the issue and investigating the difference between the 0.120.0 and 0.114.0 versions of crewai, we may be able to identify the cause of the issue and implement a fix.

Troubleshooting Tips

  • Make sure to install the correct version of crewai using pip.
  • Verify that the LLM is instantiated correctly with the desired model and reasoning settings.
  • Check the error message for any additional information that may help identify the cause of the issue.

Related Issues

Additional Resources

Frequently Asked Questions

In this article, we will address some of the frequently asked questions related to the crewai 0.120.0 version not thinking issue.

Q: What is the crewai 0.120.0 version not thinking issue?

A: The crewai 0.120.0 version not thinking issue is a bug that causes an error when using Claude AI sonet 3.7 with reasoning enabled. The error message indicates that the LLM expects "thinking" but got text instead.

Q: What are the steps to reproduce the issue?

A: To reproduce the issue, follow these steps:

  1. Install the crewai 0.120.0 version using pip.
  2. Instantiate the LLM with Claude AI sonet 3.7 and set reasoning to any value (e.g., "high").
  3. Kick off the crew. The crew will fail with the error mentioned above.

Q: What is the expected behavior?

A: The expected behavior is that the crew should run normally, just like in previous stable versions of crewai.

Q: What is the possible solution to this issue?

A: One possible solution to this issue is to investigate the difference between the 0.120.0 and 0.114.0 versions of crewai. By comparing the code and functionality of these two versions, we may be able to identify the cause of the issue and implement a fix.

Q: Is this issue specific to a particular environment or setup?

A: No, this issue was also observed on AWS SageMaker, which suggests that the issue is not specific to a particular environment or setup.

Q: How can I troubleshoot this issue?

A: To troubleshoot this issue, make sure to:

  • Install the correct version of crewai using pip.
  • Verify that the LLM is instantiated correctly with the desired model and reasoning settings.
  • Check the error message for any additional information that may help identify the cause of the issue.

Q: Are there any related issues?

A: Yes, there are related issues, including:

Q: Where can I find additional resources?

A: Additional resources can be found at:

Conclusion

In conclusion, the crewai 0.120.0 version not thinking issue is a bug that causes an error when using Claude AI sonet 3.7 with reasoning enabled. By following the steps to reproduce the issue and investigating the difference between the 0.120.0 and 0.114.0 versions of crewai, we may be able to identify the cause of the issue and implement a fix.