Hi Simon,
I have finally managed to get my IOC to run. Here's what I plan to do next: I will write a how-to documentation detailing the process. After completing the documentation, I will submit it and inform ChatGPT that this is one of the possible correct answers to
the question I asked. My definition of a possible correct answer is that it is a True Positive if there are no bugs, or a False Positive if there are hidden bugs.
This approach will improve the system as it relies on reinforcement learning with a human interface. Thank you for your advice. The old manual said 32 bytes. It was changed to 42 bytes but the manual was not updated. I can't blame AI for this.
Have a nice weekend,
Alex Tran
From: Simon Rose <Simon.Rose at ess.eu>
Sent: Friday, March 15, 2024 3:51 AM
To: Tran, Phi Dung <ptran1 at bnl.gov>; Timo Korhonen <Timo.Korhonen at ess.eu>; tech-talk at aps.anl.gov <tech-talk at aps.anl.gov>
Subject: Re: IOC databank
Hi Alex -
I know that this is not what you asked, but I would be _extremly_ careful about trusting ChatGPT and the like to help with novices writing IOCs. We have seen it generate completely false but seemingly
ok looking EPICS code that would be _very_ hard for a novice to understand why it is not correct.
Cheers,
Simon
From:
Tech-talk <tech-talk-bounces at aps.anl.gov> on behalf of "Tran, Phi Dung via Tech-talk" <tech-talk at aps.anl.gov>
Reply to: "Tran, Phi Dung" <ptran1 at bnl.gov>
Date: Thursday, 14 March 2024 at 17:33
To: Timo Korhonen <Timo.Korhonen at ess.eu>, "tech-talk at aps.anl.gov" <tech-talk at aps.anl.gov>
Subject: Re: IOC databank
Let's say I have a Libera Photon, and fortunately, it comes with a Libera Phone Epics Driver.
Let's assume that it's acceptable for the manufacturer to include it in the "OpenIOC databank," so customers don't have to search GitHub or the web for the IOC.
They can simply go to this databank and get the latest version.
For instance, if I'm working with an old Oxford Cryostream 700, I can visit the OpenIOC databank to download several IOCs developed by different people around
the world.
As a novice with EPICS, I'm thinking out loud. If ChatGPT can digest the data and the manual, then it can assist me in writing my IOCs. AlphaFold created after
protein databank was created.
From: Timo Korhonen <Timo.Korhonen at ess.eu>
Sent: Thursday, March 14, 2024 11:56 AM
To: Tran, Phi Dung <ptran1 at bnl.gov>; tech-talk at aps.anl.gov <tech-talk at aps.anl.gov>
Subject: Re: IOC databank
Hi Alex,
Could you elaborate a bit what you mean with a “centralized IOC databank” ?
This could mean (with my limited imagination):
- A central database to configure IOCs
- Some kind of master EPICS runtime database where “everything” resides
- Something like ChannelFinder that (with some configuration) creates a database of all records in your system
Or something else that I could not imagine right now.
Cheers,
Timo
From:
Tech-talk <tech-talk-bounces at aps.anl.gov> on behalf of "Tran, Phi Dung via Tech-talk" <tech-talk at aps.anl.gov>
Reply to: "Tran, Phi Dung" <ptran1 at bnl.gov>
Date: Thursday, 14 March 2024 at 16:40
To: "tech-talk at aps.anl.gov" <tech-talk at aps.anl.gov>
Subject: IOC databank
Is there a centralized IOC databank?
|