brilee

13 ms·

121.
by brilee 4y ago
I find that if you're getting this response, you're not talking high up enough in the org chart. Eventually, somebody does care about the overall good of the company, over the turf wars/budgetary bureaucracy of the middle man
122.
Using Linear Counting, LogLog, and HyperLogLog to Estimate Cardinality(moderndescartes.com)
1 points by brilee 4y ago|0 comments
123.
The Chemistry of Refrigerants(moderndescartes.com)
2 points by brilee 4y ago|0 comments
124.
The Chemistry of Refrigerants(moderndescartes.com)
5 points by brilee 4y ago|0 comments
125.
Notes on the Earth's Carbon Cycle(moderndescartes.com)
3 points by brilee 4y ago|0 comments
126.
by brilee 4y ago
HFOs are third generation refrigerants, replacing HFCs, which in turn replaced CFCs. In 1985, CFCs were recognized as destroying the ozone layer, so they were banned in 1989 and replaced them with HFCs that didn't do that. The problem
127.
Deep learning on electronic medical records is doomed to fail(moderndescartes.com)
227 points by brilee 5y ago|141 comments
128.
Unexpected Shunts in the Supply Chain(moderndescartes.com)
29 points by brilee 5y ago|5 comments
129.
by brilee 5y ago
Great read! Another phenomenon that's not explicitly mentioned, but is also important: the industry lab. DeepMind, Google Brain, FAIR, IBM Research, MSR all have big industry labs that mostly focus on CS/ML research, but the small
130.
by brilee 5y ago
This article suggests this etymology is made up... https://www.nytimes.com/2005/04/06/dining/the-raw-truth-dont...
131.
Unexpected Shunts in the Supply Chain(moderndescartes.com)
1 points by brilee 5y ago|0 comments
132.
2021 Advent of Code in Haskell(moderndescartes.com)
3 points by brilee 5y ago|0 comments
133.
by brilee 5y ago
This isn't a bad article; the title is just completely misleading. More appropriate would be: "How to implement the simplest possible version of TensorFlow Probability [1], a probabilistic programming framework". [1] https:&
134.
by brilee 5y ago
This is a really great overview of JAX - the best I've seen outside of primary Google/DeepMind sources. Glad to see people besides Googlers getting familiar with its capabilities.
135.
by brilee 5y ago
NumPy does not automatically use GPU, so my guess is that the speedup is due to the CPU/GPU differential. This is specifically noted later on in the "Should I use JAX" section - if you just want numpy for accelerators, JAX is
136.
by brilee 5y ago
if you draw out the 10x10 multiplication table, anything 0 * x or x * 0 ends in 0 (19 out of 100 cases), and any even number times 5 also ends in 0 (8 out of 100 cases), which lines up with the 27% you're seeing. Since this is on the u
137.
by brilee 5y ago
https://en.wikipedia.org/wiki/FICO how about 1956?
138.
Bulk and Cut as a Metaphor for Codebase Growth(moderndescartes.com)
2 points by brilee 5y ago|0 comments
139.
by brilee 5y ago
Figuring out how to defer taxes is equivalent to receiving a free loan from the government, which can be used to profit on the float.
140.
by brilee 5y ago
Assuming risk is evenly distributed between birth and death, you would expect a 1 in 80 chance of dying in any given year. But since the risk is loaded towards the later years of life, 1 in 200 during the earlier years sounds about right. P
141.
by brilee 5y ago
FYI - Tangent evolved into TF2's AutoGraph.
142.
by brilee 5y ago
You wrote this... All the comparisons like > are better written using their TensorFlow equivalent (e.g tf.greater). Autograph can convert them (you could write >), but it’s less idiomatic and I recommend to do not relying upon the
143.
by brilee 5y ago
a typo in the collatz description: C (2k + 1) = C (3k + 2) should be C (2k + 1) = C (6k + 4)
144.
by brilee 5y ago
The reason spammers don't remove the + automatically, is because their goal isn't to deliver spam, it's to get the right people to read the spam. The people who use the + trick are exactly the set of people who would never re
145.
by brilee 5y ago
If you're doing grid search on more than 4 hyperparameters, that probably explains why you think computation wins. Even random search with a constrained budget will do pretty well without breaking the bank.
146.
by brilee 5y ago
Thanks for the comment, I also had to dust off some mental cobwebs for this analysis :) You're pointing out the crux of the issue: "Ultimately making CaCO3 this way (and CaO) winds up substituting a 400$/tonne product in NaOH
147.
by brilee 5y ago
The secret is apparently to be slightly cagey about your tech and nerd snipe half of HN into trying to figure out if your thing actually works or not...
148.
by brilee 5y ago
I did some lengthy analysis behind the chemistry / economics and wrote it up here. https://www.moderndescartes.com/essays/carbon_neutral_concre... TL;DR: ignoring all sources of overhead/inefficiency and pure
149.
by brilee 5y ago
A followup question... how do you regenerate the alkaline sorbent? It gets acidified in the process of extracting CaO from the water and needs to be made alkaline again somehow.
150.
by brilee 5y ago
I'm not sure I understand where the carbon savings comes from. The cement industry specifically needs lime, CaO. Lime is most easily obtained by burning CO2 off of limestone, CaCO3. As you point out, this is effectively "burning o

More ›