[GENERAL SUPPORT]: Duplicate parametrizations (different objective values) give NaN errors (V 0.5.0)

Open WinC3 opened this issue 1 month ago • 1 comments

Question

https://colab.research.google.com/drive/1i615PxRrTxg0psP36knRJmgWFV1fVl32?usp=sharing Here is the colab notebook for reproducing the NaN issue. Reproduction instructions: Run all cells Last cell should stop at a manually raised ValueError Open generated log file in the files section and scroll to bottom where it should contain the full trial dataframe containing NaNs

From my testing I believe it is an issue with different trials having same parametrizations, though I do want to complete the trial despite that because I have data with different values for the objective despite same parametrizations (parameter space doesn't fully define resulting function).

Please provide any relevant code snippet if applicable.

Code of Conduct

[x] I agree to follow this Ax's Code of Conduct

Nov 20 '25 21:11 WinC3

Hello there! Running your provided code with Ax's latest version (1.2.1) doesn't reproduce the issue. Can you try upgrading Ax?

Nov 26 '25 22:11 Cesar-Cardoso