I run very very quickly very very far
Best posts made by yop217
-
RE: ChatGPT's answer about the flight model of BMS
If the off topic is still allowed, and being an AI practitioner for many years now, I will make 2 comments on LLMs (large langage models): (i do not know / master well other GenR models):
-
LLMs are not smart. As Mav JP stated, they spit out information that was injected before. The way they do it is really interesting. But this is not smart.
-
Where I see very interesting application is that LLMs can simplify the interface between the machine and humans. I see it as some sort of modern age “Windows”.
e.g.: Feed a custom LLMs with a BMS manual, train it & customize it to minimize hallucination, and then, instead of spamming BMS forum with: Why is my aircraft not turning on the ground? and risk a “RTFM” answer (i miss these good old days :p), the LLM will provide the NWS info, even with the default shortcut to it.
And this is cool… I think.
And this would be possible ONLY because there is a BMS manual, well written, with all the relevant info inside it (it does not need to be super well written btw, it’s quite robust of mixed up ideas within documents). So the smartness is within the BMS dev team, LLM is just the tool to help you access the info (in case you are so freaking lazy & can’t freaking read a well written manual).
Joke aside, this has many interesting applications when you have crazy complicated norms to follow, with updated version on top of each other and crazy badly done versioning…
-
-
RE: 4.38
We do know: two more weeks ! It has been two more weeks since so much time now. However, this should no prevent you to enjoy 4.37 in the meantime
-
RE: which AI is best for reading BMS manuals?
@bbostjan would sound something like
“RTFM” -
RE: which AI is best for reading BMS manuals?
It will work.
Using RAG, and proper prompts, you can avoid hallucinations. You can also use distillation to only use a smaller model (and make it work locally if dockerised https://github.com/google-research/distilling-step-by-step).
Now… I really doubt about the added value. A CTLR+F is more than enough to find the info on a super well structure documentation.
And if you think about the planet, you will avoid using a LLM when a CTRL+F is enough (elastic search if you want to be fancy).
We have this debate at work on a regular basis, the hype on LLM is really going crazy. It’s cool, but use it when it’s needed…
-
RE: Bms + windows 11?
@alu332 BMS is the best for Windows 11 according to my tests
-
RE: Bms + windows 11?
@wazza69 on windows? yeah it makes it useful otherwise, nope
-
RE: Engine fire
@dumba In the documentation, it is written when tests are “eye candy”
Latest posts made by yop217
-
RE: Easter Hollidays comes with U4?
@Mike_Bravo Used to be 2 more weeks
-
RE: Enemy jamming works
@Croa We (as player) cannot burn through until very close but is the AI able to burn through player jammer at “normal” distance ?
-
RE: Engine fire
@Micro_440th I don’t get it: does this mean in U4 tests are no longer optional ? I think there is value in knowing the test has no effect on the sim. Personally, I always skip them and feel safe as the manual tells me I can do it. If there is no longer any note in the doc, by default everything becomes “mandatory”
-
RE: Engine fire
@dumba In the documentation, it is written when tests are “eye candy”
-
RE: 4.38
We do know: two more weeks ! It has been two more weeks since so much time now. However, this should no prevent you to enjoy 4.37 in the meantime
-
RE: Bms + windows 11?
@wazza69 on windows? yeah it makes it useful otherwise, nope
-
RE: Bms + windows 11?
@alu332 BMS is the best for Windows 11 according to my tests