icecoffee

4 ms·

1.
Future of Scientific Problem Solving? Sharing Pre-Built AI Solvers Like Recipes(forums.developer.nvidia.com)
2 points by icecoffee 2y ago|1 comments
2.
by icecoffee 2y ago
An interesting find. In scientific computing, I've learned to accept the performance/accuracy as it, slow and iterative, many runs being similar - Would pre-built solver templates with AI models reduce repetitive compute? There a
3.
by icecoffee 3y ago
The cv reading here is really a POC base on off-the-self and will require some work. Fix-camera vs Spot, I can definitely see the pros and cons. So at what price point do you think industry will adopt Spot for doing rounds?
4.
by icecoffee 3y ago
I suspect models probably need "reasoning" ability - very lacking in current models. Perhaps, AlphaGeometry can do better.
5.
by icecoffee 3y ago
Though I haven't effort into analyzing or tweaking the training, some of the perditions does look like it points at the shadow instead of the tip. This synthetic dataset is tiny (not even 100k) - but still larger than what I could find
6.
by icecoffee 3y ago
Yes, I encountered a similar issues with the first iteration of dataset. The prediction location of the needle tip would be wildly off when the view angle was high. The fix was easy, just generated high viewing angle images for the 2nd data
7.
by icecoffee 3y ago
Exactly, I use to think the same thing! Many legacy system, the industrial machines that run factories, infrastructure, etc. all have hundreds of gauges and some of them do get "watched". Cost of digitization via smart gauges just
8.
by icecoffee 3y ago
Yea... the model wasn't trained on speedometers. I curious how this might help with your "eCVT issue", can you elaborate?
9.
by icecoffee 3y ago
The error messaging was non-existent. Add some and redeployed, hopefully its a bit helpful.
10.
by icecoffee 3y ago
Yup, the components can be replaced to your liking and expand the data to your match your environment better. I found it really interesting how, no matter how I prompted foundational models they couldn't read gauges.
11.
Show HN: Simply Reading Analog Gauges – GPT4, CogVLM Can't(huggingface.co)
66 points by icecoffee 3y ago|41 comments