Hello Su,
I could perhaps help you out with this—but from a slightly different approach. If you want to install the latest HDDM with all its functionality and without having to deal with difficult and time-consuming dependency and wheel-building issues, I would suggest installing the Docker HDDM image (dockerHDDM: A user-friendly environment for Bayesian Hierarchical Drift-Diffusion Modeling by Pan et al., 2024).
For this to work, I recommend first installing Docker for your desktop. I use Windows Subsystem for Linux 2 (WSL 2). Then, you can simply create a Docker image exactly as outlined by Pan et al. (2024) in their paper. If you install the latest version, you can convert output into InferenceData, get pointwise likelihood calculations, and perform posterior predictive checks with additional functionality. On top of that, the Arviz library is used for model plots.
When you want to run HDDM, I suggest running the Docker image and attaching to your container in VS Code. I’ve found that using HDDM this way has saved me a lot of time.
I believe this does not quite answer your question, but it might help with future HDDM use.
Best,
Veronika