ekelsen

15 ms·

181.
by ekelsen 5y ago
I can't tell from the article if this is enriched uranium (which would be really scary), or just uranium as it comes out of the ground (which would be much less scary, although still not great).
182.
by ekelsen 6y ago
Blame the people at the company who didn't make sure the right processes and tools were in place? It seems like while CTO made the error, he was also responsible for making sure he wouldn't be able to make such an error. He faile
183.
by ekelsen 6y ago
Could not understand that sample at all, I don't think this makes your point.
184.
by ekelsen 6y ago
Standard convolutions are not memory bandwidth bound for common sizes/shapes. Convolutions run much faster using the TC than not using them. Depthwise convolutions are different though - they usually are bandwidth bound.
185.
by ekelsen 6y ago
For those interested in more detail / explanation, the paper below does a good job explaining the basic idea of how this works, but with only a 1st order approximation for the mantissa bits. The first order approximation reduces the c
186.
by ekelsen 6y ago
The closest open source project is probably https://github.com/mozilla/LPCNet/
187.
by ekelsen 6y ago
WaveRNN. https://arxiv.org/abs/1802.08435
188.
by ekelsen 6y ago
If you buy a call option, someone has to be on the other side of the contract. They are the short position.
189.
by ekelsen 6y ago
Serves in tennis don't go directly down the line, they go cross court. Returning players will often be standing behind the baseline. Additionally balls start ~2.5-3m above the ground, bounce and then come up again. The total distanc
190.
by ekelsen 6y ago
Last I checked none of those things are fruits or vegetables. Suggesting they were enough to get all necessary vitamins and minerals was simply inaccurate and bad health advice. When I Google "mushroom B12" it seems to suggest mus
191.
by ekelsen 6y ago
> Fruits and Vegetables have extremely high densities of all your minerals and vitamins. I don't think they have B12, you need to get that from animal sources or a pill.
192.
by ekelsen 6y ago
One minor nit: The recent FB blogpost wasn't really "research" - it was mostly a reimplementation of this paper: https://arxiv.org/abs/1802.08435
193.
by ekelsen 6y ago
Exciting to see our research making broad impact across the industry! https://arxiv.org/abs/1802.08435
194.
by ekelsen 6y ago
I think reading through all the examples https://github.com/seanbaxter/circle/blob/master/examples/RE... makes it clear that the author has put quite a deal of thought into how these features make i
195.
by ekelsen 7y ago
I can't find any studies that confirm cholera or tuberculosis. Do you have any citations? For example, I found: https://www.ncbi.nlm.nih.gov/pubmed/26527495 which is a single author hypothesis.
196.
by ekelsen 7y ago
It's unclear this algorithm would be useful in practice. Training the weights will lead to a more accurate network for the same amount of work at inference time.
197.
by ekelsen 7y ago
The implications are unclear to me. We already know how to prune models for inference. For example https://arxiv.org/abs/1710.01878 , along with earlier work (and more recent work). There's also work showing tha
198.
by ekelsen 7y ago
What infections does having one copy of the CF gene protect against? The wikipedia page suggests this is an unproven hypothesis.
199.
by ekelsen 7y ago
I'm generally unclear about the utility here - goSLP which generates the training data seems to be significantly better at ~1.04. Is it because the NN might be faster? It would be illuminating to see approximate compile times. The 1.
200.
by ekelsen 7y ago
Transcribing solo piano to MIDI, even very complex polyphonic pieces, is fairly reasonable these days. You can try a demo from a not quite state of the art system here: https://piano-scribe.glitch.me/ . (full disclosure, I
201.
by ekelsen 7y ago
TLDR; Make 1x1 convolutions sparse, write fast Sparse Matrix Multiplication kernels, get a nearly 2x speedup with smaller models.
202.
by ekelsen 7y ago
I wasn't questioning whether they did the right thing. I wanted to understand why the obvious solution of descending (which is what commercial airliners have done successfully on numerous occasions when losing cabin pressurization) di
203.
by ekelsen 7y ago
Why didn't the pilots just descend when they encountered a failure with the environmental system?
204.
by ekelsen 7y ago
It seems like they got the order wrong though.
205.
by ekelsen 8y ago
They aren't always as fast...
206.
by ekelsen 8y ago
I think the goal should be use the smallest dense network possible as the baseline. For MNIST, this might be a LeNet style convnet with [3, 9, 50] instead of the [20, 50, 500] network which is standard (and way overkill). I haven't ex
207.
by ekelsen 8y ago
We rigorously evaluate three state-of-the-art techniques for inducing sparsity in deep neural networks on two large-scale learning tasks: Transformer trained on WMT 2014 English-to-German, and ResNet-50 trained on ImageNet. Across thousands
208.
The State of Sparsity in Neural Networks(arxiv.org)
4 points by ekelsen 8y ago|3 comments
209.
by ekelsen 8y ago
Each program got 1 minute per move in the original match. Did it find the blunders within 1 minute? If you really feel like their original setup handicapped stockfish, it seems like the best way to know would be to have that setup play you
210.
by ekelsen 8y ago
They also beat it with only 1/10th the time that stockfish got. That should more or less negate any advantage of the extra processing power.

More ›