Leonid

Hi @Uberi, thanks for getting back to this. I tried the change your proposed (Autocomplete.ahk, line 22) however that didn't affect the observed behavior. (Obviously I restarted Autocomplete after I...

Made the change in the .ini, too. But unfortunately still no autocomplete for words starting with parenthesis. Words that start with letters still work fine. The message box with `working`...

No effect, I am afraid! I tried checking if changing the `OtherKeyList` has any effect at all, by changing both the ini as well as the script to random values...

@nabenabe0928 Here is one of the logs, is it sufficient for you with the parameters removed? Let me know if not, then I'll come up with something to anonymize them....

Got it, first of all, here is the annonymized log: ``` [I 2024-08-23 08:17:44,494] A new study created in memory with name: no-name-d3e321d8-569e-4e5d-9896-4d2c27d6d736 [I 2024-08-23 08:26:14,720] Trial 0 finished with...

Thanks for the super quick response on this @nabenabe0928, thats really impressive! - You are right about your first observation, this study must have been from before I realized this...

I'm in the process of checking this, you noted absolutely correctly that `param10 < param11` should never happen. Can't reproduce that in my tests either. I think I'll have to...

I apologize for the delay, turns put there was a bug in the code that lead to the generation of conditional search spaces. After fixing it, the GPSampler ran through...

@nabenabe0928 That's what I get for trying to cut corners and use LLMs to do the annonymization. Sorry about that, I wrote an anonymization function now, so hopefully no more...

@nabenabe0928 I understand your frustration! The reason the log and the space do not match is that they are from different models and from different runs. As I mentioned, I...