Sam Stites @sam@types.pl (@SamStites) header
Sam Stites @sam@types.pl (@SamStites) profile picture
Boston, MA
Joined June 2012
Photo by Sam Stites @sam@types.pl (@SamStites)Photo by Sam Stites @sam@types.pl (@SamStites)Photo by Sam Stites @sam@types.pl (@SamStites)Photo by Sam Stites @sam@types.pl (@SamStites)Photo by Sam Stites @sam@types.pl (@SamStites)Photo by Sam Stites @sam@types.pl (@SamStites)Photo by Sam Stites @sam@types.pl (@SamStites)Photo by Sam Stites @sam@types.pl (@SamStites)Photo by Sam Stites @sam@types.pl (@SamStites)
Sam Stites @sam@types.pl retweeted
(1/5) Codex and other LLMs for code are usually benchmarked on a single PL (usually Python). We wanted to know: how do they fare on other languages? Can they generalize across PLs?
1
14
3
50
Excited to present my work with Hao Wu (@Hao_Wu_), Babak Esmaeili (@bob_smiley_), and Jan-Willem van de Meent (@jwvdm) on Nested Variational Inference at #NeurIPS2021 today. Come and join us in poster session 1 at spot C0.
New on ArXiv: Nested Variational Inference arxiv.org/abs/2106.11302 Work by Heiko Zimmermann (@zmheiko), Hao Wu (@Hao_Wu_), Babak Esmaeili (@bob_smiley_), and myself. (this is an extended version of our work at AABI this year; openreview.net/forum?id=tvxL…) [1/]
8
25
Sam Stites @sam@types.pl retweeted
New working paper: Learning Proposals for Probabilistic Programs with Inference Combinators ArXiv: arxiv.org/abs/2103.00668 Joint work, several years in the making, by Sam Stites* (@SamStites), Heiko Zimmermann* (@zmheiko), Hao Wu (@Hao_Wu_), and Eli Sennesh (@EliSennesh) [1/]
1
35
2
118
Sam Stites @sam@types.pl retweeted
PROBPROG 2021 will take place online on Wed Oct 20 – Fri Oct 22 this year. The deadline for extended abstracts and syndicated submissions is Thu May 6 AOE. See probprog.cc for information and submission instructions!
13
29
Sam Stites @sam@types.pl retweeted
I’ve got T5 working in @hasktorch. – at Montréal, Québec
3
3
23
Sam Stites @sam@types.pl retweeted
we are already drowning in unmanageable complexity of too much software that is ridden with bugs and security flaws, and soon this will be dwarfed by oodles of bug and security flaw ridden synthesized code that fewer and fewer people will be able to maintain and understand.
GPT3 writing code. A compiler from natural language to code. People don't understand — this will change absolutely everything. We're decoupling human horsepower from code production. The intellectual equivalent of the discovery of the engine. player.vimeo.com/video/42681…
9
14
1
91
Sam Stites @sam@types.pl retweeted
I'm very happy to close the decade with a new result: Bayesian updating is lenses - squinting not necessary I believe this establishes the importance of lenses much more strongly than before. In categorical cybernetics they will be absolutely the central concept
6
10
1
100
Sam Stites @sam@types.pl retweeted
Freedom and democracy is worth fighting for. PyTorch vs TensorFlow or SGD vs Adam... probably not.
12
68
6
575
Sam Stites @sam@types.pl retweeted
#CallForSubmissions #CallforPapers PROBPROG 2020 is seeking Extended Abstracts & Syndicated Submissions by *Jan 10 2020* for our April 2020 international conference dedicated to probabilistic programming! probprog.cc/#call
Sam Stites @sam@types.pl retweeted
Pyro 1.0 is released: stable APIs, a jit-compatible Predictive helper, new normalizing flows, parallel-scan GPs and state space models, more-automatic AutoGuides, an OED tutorial, FoldedDistribution, and PyroModule[-] to make your nn.Modules Bayesian github.com/pyro-ppl/pyro/rel…
1
79
3
328
Sam Stites @sam@types.pl retweeted
1/ New work by Alican (@alicanb_) and Babak (@BabakEsmaeili10): "Evaluating Combinatorial Generalization in Variational Autoencoders" (arxiv.org/abs/1911.04594) In this paper we ask the question: "To what extent do VAEs generalize to unseen combinations of features?"(thread)
1
39
2
109
Sam Stites @sam@types.pl retweeted
1/ New on arXiv: "Amortized Population Gibbs Samplers with Neural Sufficient Statistics" arxiv.org/abs/1911.01382. Work by: Hao Wu (@Hao_Wu_), Heiko Zimmermann (@zmheiko), Eli Sennesh (@EliSennesh), and Tuan Anh Le (@tuananhle7). (thread below)
1
13
1
35
Sam Stites @sam@types.pl retweeted
Are you a researcher or practitioner at the intersection of formal methods and numerical/high-performance computing? We'd love to receive your abstract or paper and meet you at the workshop! Extended FHPNC'19 submission deadline: Saturday, May 25. icfp19.sigplan.org/home/FHPN…
5
1
Sam Stites @sam@types.pl retweeted
New #nlproc blog post: "Design Patterns for Production NLP Systems" deliprao.com/archives/294 – at San Francisco, CA
2
61
174
Sam Stites @sam@types.pl retweeted
Numerical Programming in Functional Languages icfp18.sigplan.org/track/npf… submission deadline in the 3 days. Please re-tweet.
Sam Stites @sam@types.pl retweeted
That $35 that scientific journals charge you to read a paper goes 100% to the publisher, 0% to the authors. If you just email us to ask for our papers, we are allowed to send them to you for free, and we will be genuinely delighted to do so. xitter.cf/mantia/status/10135597…
What‘s something that seems obvious within your profession, but the general public seems to misunderstand?
616
40,404
2,392
99,338
Sam Stites @sam@types.pl retweeted
Fun fact: I use this Dr. Seuss style rhyme to memoize how Hutton-style parsers are defined ^^ #haskell
1
4
9
Sam Stites @sam@types.pl retweeted
GHC 8.2.1 is compiling my examples 4x faster than 8.0.2. Heaviest example went from 43-45s to 9-11s. Even without -0O! Kudos! #haskell
4
65
3
149
Sam Stites @sam@types.pl retweeted
First steps of freedom!! 😄 instagram.com/p/BUMgk0_BPbs/ #ChelseaIsFree
1,539
15,596
1,837
49,051