fallingmeat

10 ms·

61.
by fallingmeat 4y ago
Just tried this with ChaGPT for giggles. It spit out words, but I wouldn't call it sound reasoning. __user__: - A system in composed of two elevators. - Each have their own unique velocity parameter. - At each time step in the system,
62.
by fallingmeat 4y ago
Yes, if they were to move any number of floors at the same time step. But the language model isn't necessarily going to automatically discover those rules for you. Even if it did, the model won't necessarily be able to frame the p
63.
by fallingmeat 4y ago
sorry, that wasn't my point. I just meant, there are tools for that job. LLMs don't seem to be the most appropriate tool. Thanks for helping me to clarify
64.
by fallingmeat 4y ago
No it's not, it's a language model. It manipulates words. If you want reasoning, you should find the right tool. For example, encode your problem for an SMT solver (like Z3, NuSMV, etc). Let's say you have an elevator start o
65.
by fallingmeat 4y ago
do you foresee sysml like diagrams in your roadmap? if you could do activity diagrams with swimlanes…dude. that'd be huge.
66.
by fallingmeat 4y ago
I like where your head is at! make that the vision. /fire suit on
67.
by fallingmeat 4y ago
“… TTP/Global Entry sort of fixes this…” wish that were true, but the variance of wait time, even with tsa precheck, etc defeats the purpose. needs to be reliably consistent (and short) for me to take advantage of a 30min show time.
68.
by fallingmeat 4y ago
haha well they did get that cool nickname
69.
by fallingmeat 4y ago
This is why you should always synchronize you cloud with your other cloud, and the NAS in your closet, and floss after every meal, and take daily vitamins.
70.
by fallingmeat 4y ago
this is so much cheaper than fine-tuned model; it would make sense to try and use the multi-shot prompt in the 3.5turbo model. plus, there's the hundreds/thousands of training items that you wouldn't need to create....fml
71.
by fallingmeat 4y ago
I’ve tried to identify parts of the prompt in the output, hopefully recognizing someone trying to dump the prompt, but that’s pretty tough to do. I’m sure even fine tuned models are susceptible to spilling proprietary training data
72.
by fallingmeat 4y ago
i think that would be an even more interest post.
73.
by fallingmeat 4y ago
im also curious
74.
by fallingmeat 4y ago
not bad! eager to see how the economics work out. is openAi endpoint cost scalable for a "free" or "free-ish" service like this (seems like there will be a very liberal use of queries)?
75.
by fallingmeat 4y ago
oh it's nondeterministic! yeah this is fun
76.
by fallingmeat 4y ago
Do your part- make a review and put a dent in
77.
by fallingmeat 4y ago
isn't this what mbse (i.e. represented in sysml) is for?
78.
by fallingmeat 4y ago
whoa! this is amazing. nice work my friend
79.
Generative “Choose your own adventure” with ChatGPT prompt
2 points by fallingmeat 4y ago|2 comments
80.
by fallingmeat 4y ago
Wi-Fi jammer?
81.
by fallingmeat 4y ago
Location: Atlanta, GA Remote: Yes or commute (~quarterly) Willing to relocate: No Technologies: Engineering management; Avionics embedded systems engineering (RTCA/DO-178C); Autonomous systems; Formal methods Resume: https:/&#x2
82.
by fallingmeat 4y ago
^ this guy knows what he is talking about :). sup man
83.
by fallingmeat 4y ago
They do have a flight test program.
84.
by fallingmeat 4y ago
what are the killer apps for this? maybe someone more bullish can summarize who this customer is?
85.
Kittyhawk, Larry Page’s Flying-Car Company, Will Shut Down(bloomberg.com)
1 points by fallingmeat 4y ago|0 comments
86.
by fallingmeat 4y ago
shouldn’t matter who does it or for how much, as long as the sw lifecycle process is followed (which would include supplier oversight). Actually writing code, btw, is the cheap/easy part- designing correct system and sw requirements, a
87.
by fallingmeat 4y ago
+1
88.
by fallingmeat 4y ago
To anyone who has experience with Arc24; How does this compare to the MBSE/SysML framework from OMG? https://www.omgsysml.org/what-is-sysml.htm
89.
by fallingmeat 4y ago
Have you explored using a formal verification tool to prove out the safety/liveness properties of your (or the end user's) autonomy specifications? Could you model the system in something like Spin/NuSMV/PRISM and then e
90.
by fallingmeat 4y ago
(1) Model checking first (using a tool most appropriate for the domain): UPPAAL, SPIN, NuSMV, Z3, TLA+ (2) Model based design (less time spent on less value-added concerns...plus, capturing even more benefit from (1))

More ›