ML development risks · Question 42 of 60
After a PyTorch update, the same model with the same data delivers different results. Which risk is present, and which mitigations fit?
2 answers are correct. Hard
Show answer
Answer
- B Risk: a defect in a library used by the framework
- D Mitigations: ML functional performance testing and back-to-back testing
Explanation
A defect in a library used by the framework, for example PyTorch, is a typical ML development risk. ML functional performance testing and back-to-back testing help against it; a back-to-back test can compare the results of the old and the new version.