Aeolos

22 ms·

211.
by Aeolos 8y ago
Colemak seems to be working fine on MacOS and Windows.
212.
by Aeolos 9y ago
> This is where I don't understand. If the data scientists don't know the why of the decision making and the software cannot explain itself then how do they understand that it is accurate? They understand that by evaluating the
213.
by Aeolos 9y ago
Congratulations, Deeplab 3+ finally discovered that the U-net architecture, first proposed 3 years ago, is more efficient than the flat architecture they used before. Deeplab 3+ is still a wildly inefficient network structure, but it undeni
214.
by Aeolos 9y ago
There is one thing that stops me from using Safari (or Chrome/Opera) on Mac OS: the lack of middle-click autoscrolling. Firefox lets me do a triple-finger tap and then scroll a page smoothly at a consistent pace. This is wonderful for
215.
by Aeolos 9y ago
Because that allows you to capture a complete, scrollable website as a single png image. OS tools only let you capture the currently viewable area.
216.
by Aeolos 9y ago
One thing to note is that most cameras use 12 or 14 bit sensors, so in-sensor digital gain can still be better than post-processing.
217.
by Aeolos 10y ago
Indeed, but if performance is affected so much by the initialisation, then I would avoid random initialisation in the first place. There are various publications exploring different initialisation methods for various problems. I'm afra
218.
by Aeolos 10y ago
> 2. You usually flatten right before the final layers. The newer trend appears to be fully-convolutional networks even for classification, since they appear to overfit less, compared to flattening+dropout.
219.
by Aeolos 10y ago
> The randomness also doesn't help, as running the same model with different random seeds I have noticed that there is huge variance in accuracy. Set all your random seeds to something predefined, such as 42. Even though the exact r
220.
by Aeolos 10y ago
> Many C++ programmers don't use Boost and generic C++ libraries are unable to use Boost as a dependency. That is because a boost installation litters your hard drive with several hundred thousand tiny files. I can't in good co
221.
by Aeolos 10y ago
When using a single GPU, as is the case here, you will not be able to saturate a TB2 slot in any reasonable deep learning workload. For multiple GPUs, sure, PCIe performance matters, but for single GPUs it is not a bottleneck for any practi
222.
by Aeolos 10y ago
Exact same experience. Roughly 48 hours of uptime and then I have to reboot in order to maintain a SSH connection. That, and Finder that becomes unusably slow after working with images for a few hours. (Admittedly, tens of thousands of imag
223.
by Aeolos 10y ago
Tensorflow has a GPU-accelerated Random Forest implementation: https://github.com/tensorflow/tensorflow/blob/v0.10.0rc0/ten...
224.
by Aeolos 10y ago
This code suffers from a race condition: the file might be deleted/moved/renamed at any point, even if the file.is_open() check succeeds. Code that does not consider this possibility is technically fragile under a wide set of pret
225.
by Aeolos 10y ago
If the world evolved in linear terms, you would be right. However, the rate of scientific advancement has multiplied exponentially over the history of mankind: our scientific understanding has advanced more in the last 100 years than it has
226.
by Aeolos 10y ago
This is essential knowledge you can get from a number of different places: first semester of university, any sort of 3d or image/signal processing, or just plain old wikipedia. You could still use a deep learning framework as a black b
227.
by Aeolos 10y ago
> It doesn't really matter how many small-arrays you point at one target because it's the SNR at the individual detector-sites that matters. This is quite incorrect. Detector noise is a combination of thermal/electrical&#
228.
by Aeolos 10y ago
> The vast majority of them are terribly boring, by way of example, the very first one on the page: > [...] ἡ παρθένος ἐν γαστρὶ ἕξει καὶ τέξεται υἱόν [...] There is a more glaring problem here: "παρθένος" here can be transl
229.
by Aeolos 10y ago
For the record, this post is completely false, as even a cursory glance at the relevant Debian bug report will prove: https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=727708#672... We exercise our power to decide
230.
by Aeolos 10y ago
Arch + Gnome 3.20 is what I use on my desktop too. The UX is excellent. Too bad that linux battery life and gpu-switching is still not up to scratch for laptops.
231.
by Aeolos 10y ago
I'm sure there already is a complete, working port of MacOS on ARM. Application compatibility is the real sticker.
232.
by Aeolos 10y ago
There is a thing such as an amicable break up. You'd hope that the psych evaluation would select for people who can handle sex and relationships in a mature way without jeopardizing the mission.
233.
by Aeolos 10y ago
On the other hand, 6MB is the typical file size for the jpg-compressed images taken by my 20MP smartphone camera. It is all a matter of perspective.
234.
by Aeolos 10y ago
WinForms is a dead end. It doesn't support Windows modern or Mac OS X, and it looks completely out of place on Linux.
235.
by Aeolos 10y ago
Thank you for writing this. I had exactly the same experience, since primary school I would stay awake until 2-3AM every night staring at nothing. My biological clock just couldn't accept the 8PM sleep schedule. Then at some point I de
236.
by Aeolos 10y ago
Plus, burning trees generates a ton of other harmful particles. It's not a problem if a few people do it during winter time for fun, but it becomes a massive healthcare issue when it generates a suffocating low-lying brown smog carpeti
237.
by Aeolos 10y ago
What are you talking about? Most of this data is publicly available. You can download it and run your own analysis right now if you wish.
238.
by Aeolos 10y ago
I'd love to switch to PyPy, but Theano doesn't seem to be supported, Tensorflow is an unknown. Does anyone know what is the limiting factor for supporting deep learning frameworks on PyPy?
239.
by Aeolos 10y ago
I normally run the GPU headless for machine learning. In the rare case I need to access the desktop shell of the VM, I just connect via Chrome Remote Desktop. In the even rarer case that I wish to play a game, I have an HDMI connection betw
240.
by Aeolos 10y ago
Extremely so. My current (and only) working solution is to disable the Nvidia GPU on a host and use it PCIe passthrough to assign it to another Linux or Windows VM. That's the only stable solution that doesn't break the world when

More ›