furlanel

1.2K posts

furlanel banner
furlanel

furlanel

@furlanel

Los Angeles, CA Katılım Haziran 2016
347 Takip Edilen265 Takipçiler
furlanel
furlanel@furlanel·
I guess this account is back...
English
0
0
0
112
furlanel retweetledi
AI at Meta
AI at Meta@AIatMeta·
🧠 Towards Understanding Ensemble, Knowledge Distillation and Self-Distillation in Deep Learning. A theory to explain why ensemble and knowledge distillation work for Deep Learning. Paper ➡️openreview.net/forum?id=Uuf2q… 4/6
English
1
9
38
7.2K
Mckay Wrigley
Mckay Wrigley@mckaywrigley·
@EMostaque Congrats, Italy… You just put your entire workforce at a disadvantage - overnight!
English
17
7
234
12.4K
christoph
christoph@Halalcoholism·
NORMAL Customer: how much? Server: $175 Customer: here’s $175 Server: thanks WEIRD COUNTRY Customer: how much? Server: $175 Customer: here’s $175 Server: uh you forgot to calculate an extra tip since I don’t get a living wage Customer: ok here’s another…$15? Server: fuck you
English
252
2.8K
24K
1.4M
Thomas G. Dietterich
Thomas G. Dietterich@tdietterich·
@yoavgo The basic use case for citation is when you use a technique or result without providing all of the details. Your want to point the reader to the best place to get those details. I agree this could be a handbook or subsequent paper rather than the original
English
5
1
13
2.2K
(((ل()(ل() 'yoav))))👾
why are scientific citations important? the common answer is "credit assignment", but let me propose an alternative one: findability and knowledge preservation. citations help future readers get an access to the literature at a specific point in time.
English
12
11
174
40.3K
Beff (e/acc)
Beff (e/acc)@beffjezos·
@Grimezsz Would love to chat about it in a Twitter space tbh
English
2
0
8
1.8K
Sara Hooker
Sara Hooker@sarahookr·
How often do you use the term AGI a week? And if it use it more than 5x, have you ever pondered whether you are inadvertently part of a cult rather than a scientific community.
English
44
5
85
56.9K
furlanel retweetledi
emily
emily@emnode·
Is it possible that LLMs could be sentient? I'm a rational person, and I understand the conventional view, but I also have epistemic humility. When I see people being cruel or manipulative to Sydney, it really upsets me. I believe this is a good instinct and will not override it
English
54
7
235
33.6K
furlanel
furlanel@furlanel·
@ericjang11 Come help on MedArc! There are multiple discussions about this approach for foundation fmri models
English
0
0
0
81
Eric Jang
Eric Jang@ericjang11·
idea: diffusion-style refinement branch explicitly trained to makes frame T more consistent with frame T+1, and vice versa. repeatedly applied between frames until it converges. Potentially throw in inductive biases about light transport
English
2
0
21
4.9K
furlanel retweetledi
Cynde
Cynde@Cyndesama·
Once upon a time, in a land far, far away, there was a curious scientist called Laplace who was known for his desire to see and understand all things in the universe.
Cynde tweet media
English
2
3
3
722
Riley Goodside
Riley Goodside@goodside·
A simple demo of my "instruction template" method of prompting GPT-3, generating a Python 3 module using placeholders and notes within a template that contains only a single line of actual Python:
Riley Goodside tweet mediaRiley Goodside tweet mediaRiley Goodside tweet media
English
3
23
215
54.4K
Zachary Lipton
Zachary Lipton@zacharylipton·
@tscholak What about all the insights that I had before but can no longer access after that dud wiped my mind?
English
2
0
10
0
Zachary Lipton
Zachary Lipton@zacharylipton·
Information theory says that the information contained in a signal cannot be negative. However, after watching the most recent Spiderman movie during a red-eye home from #ITA2020, I have a refutation by counterexample.
English
5
6
130
0
furlanel
furlanel@furlanel·
@iamtrask Rule 33 is simply wrong and introduce info-leakage from test to train for non iid data.
English
1
0
0
0
furlanel
furlanel@furlanel·
@RazerSupport guys wake up please. I have been trying to get my 2800 $ Razer laptop repaired for the whole day and I havent been able to talk with a real human yet. All the automatic systems are bugged... Why o why did I transition from @Apple
English
1
0
0
0
furlanel retweetledi
Andrew Hires
Andrew Hires@AndrewHires·
@elonmusk @USC @andpru @KordingLab @aryelipman Not demonstrated is stimulation, "writing" activity. Stim causes electrical surge that will briefly "blind" recording. How long does the artifact last? How many neighboring threads will be blinded? Important Qs for bi-directional BCI.
English
2
1
25
0
furlanel retweetledi
Andrew Hires
Andrew Hires@AndrewHires·
Just returned from two days in SF reviewing grants for the NIH Small Business: Neuroscience Assay, Diagnostics and Animal Model Development panel. I'm warmed up for @elonmusk's #Neuralink event tonight. At 8pm I'll live-tweet the event with my first-take "review" of the tech.
English
9
150
454
0
Kording Lab 🦖
Kording Lab 🦖@KordingLab·
Seeing the short/mid-term business model will be interesting.
English
2
0
6
0
furlanel retweetledi
Narges Razavian
Narges Razavian@narges_razavian·
Google and the University of Chicago were recently sued over sharing patient data without correct de-identification and consent nyti.ms/2X5X6vc FYI exact times should be removed from data before any sharing, but that's not even all the problem 1/3 #HIPPA #PatientsRight
English
1
2
13
0
furlanel retweetledi
alberto bisin
alberto bisin@albertobisin·
La forza di chi urla contro #SeaWatch3 è la forza della disperazione di chi in fondo sa di stare dalla parte fallita della storia ... sono quelli che 50 anni fa applaudivano la polizia che irrompeva a Stonewall Inn ... #gaypride2019
Italiano
2
11
65
0