CrewAI Training: Is the Crew Really Using the .pkl Files?

CrewAI Community Support
1

What the documentation claims:

“I went through this post: How does training & evaluation actually work in CrewAI?
When you run training on a crew, it will prompt you for feedback. It will then capture that feedback and store it in a pickle file. And then the next time it runs, it will inject that file into your system prompts.”

What actually happens:

  1. After training my crew and generating the training_data.pkl and usecase.pkl files, I ran the crew—but it doesn’t seem to utilize the training data.
  2. When I train the same crew a second time, the previous training data appears to be overwritten/reset, losing prior feedback.

My questions:

  • How can I verify that the crew is actually using the training data stored in these .pkl files?
  • Is there a way to persist and accumulate training data across multiple sessions?
  • If training_data.pkl is being reset, how can I prevent this and make sure my crew continues learning from previous feedback?

Any insights or workarounds would be appreciated! @crewAI_Staff

2

Shivang, first off, welcome!

In this other thread, you can read some more recent and maybe more complete information about CrewAI training.

1 Like
3

This topic was automatically closed 24 hours after the last reply. New replies are no longer allowed.