মিডিয়া
- টুইট
- টুইট এবং উত্তর
- মিডিয়া, বর্তমান পৃষ্ঠা।
-
NeuralCoref v3.0 is out
!
- up to 100x faster than v2.0 (thanks Cython)
- Integrated in spaCy models and pipeline
+
=
- Based on the fast neural net model by @stanfordnlp, trained in@PyTorch Check it out: https://github.com/huggingface/neuralcoref … Cc@spacy_iopic.twitter.com/z8OjrzEOAU
-
Lot of great works on Word/Sentence Embeddings in 2017/early-2018! ELMo by
@mattthemathman at@allenai_org, InferSent by@alex_conneau et al., learning by multi-tasking from MILA/@MSRMontreal...@SanhEstPasMoi and I wrote a post summarizing the current baselines, SOTA and trendspic.twitter.com/QM7Q1dialb
-
"Proof by intimidation" are way too common nowadays! But my favorites are proofs by "circular arguments"pic.twitter.com/edIfHslTvc
-
You can tune the acceptance probability threshold T to trade-off computation for statistical accuracypic.twitter.com/4fkCv31NRB
এই থ্রেডটি দেখান -
Interesting combination of variational inference and rejection sampling by
@ermonste &@adityagrover_. Rejection sampling is not dead! And a nice blog post https://goo.gl/qfkpd1 pic.twitter.com/uoqILvGVAM
এই থ্রেডটি দেখান -
The "Virtual Stunman" of @jasonpeng0, Sergey Levine
@Mvandepanne and@pabbeel is a really nice piece of reinforcement learning combined with imitation learning. Interesting insights! And the blog post is great, check it out! http://bair.berkeley.edu/blog/2018/04/10/virtual-stuntman/ …pic.twitter.com/0S6GZ1djYk -
We take our neural nets too seriously. I see them more like this than any terminator-like robot.pic.twitter.com/SzKOEIvhOF
-
Neural net training in one slide. If you sometimes need to explain it to a general audience.pic.twitter.com/0oo5HMWtZc
এই থ্রেডটি দেখান -
To introduce the work we'll be presenting at ICLR, I tried to draft the most visual & intuitive introduction to Meta-Learning I could come up with. Check it out if you like watching cute animated neurons sending information and gradients to each other! https://medium.com/huggingface/from-zero-to-research-an-introduction-to-meta-learning-8e16e677f78a …pic.twitter.com/ML4T4SA47B
এই থ্রেডটি দেখান -
@demishassabis story at#aiforhumanity is a nice one: reading a lot about theoretical physics as a kid he found out that even the greatest physicists struggled with problems like the great unification theory. Decided to switch to AI to build something that could augment humanspic.twitter.com/ywE1AL05g4
-
@mounir and@LatanyaSweeney at#aiforhumanity on the problem of black box algorithmic newsfeeds/ads. Dreaming of a "why" button that'd tell you "I showed you this ad because you clicked on this earlier and a third-party told me you bought this from them"pic.twitter.com/mdJlsx4AT7
এই থ্রেডটি দেখান -
@mathbabedotorg at#aiforhumanity on the huge importance of understanding failure cases and bias in the algorithms we use at large scale, not only in gov but also big private comp. Deeply resonate with the latest article of@fchollet I read yesterdaypic.twitter.com/SdnR71ZNin
-
#aiforhumanity Justine Cassell from CMU on crafting human-compatible AI and how conversational AI could actually help improve human-human interactions (think of a positive/compassionate AI that gives you confidence in yourself / power to reconnect to others when you are down)pic.twitter.com/DsdusmM1NV
-
Really proud of the way the French gov is embracing and questioning the AI revolution! At
#aiforhumanity today in Paris with a crazy lineup, half a day of debats between@drfeifei@ylecun@demishassabis Berkeley's Stuart Russell@noricoco@mathbabedotorg@sama@VillaniCedricpic.twitter.com/2fwPIavO8X
-
Yeepee
Our first submited paper was accepted @ICLR18 (workshop track)! More about it soon!pic.twitter.com/Y0DWuEzuIM
-
What is more interesting that a SOTA Deep Learning model? A good paper studying the limits of a SOTA model! https://github.com/thomwolf/awesome-limits-of-dl4nlp …pic.twitter.com/uOCJgWKz2V
এই থ্রেডটি দেখান -
Join the movement to beat smog.
#CleanAirTogether http://thndr.me/eZ58Le pic.twitter.com/k1Z9MXhPOt
-
Improving latent variable models for dialog generation by Julian Serban, A. Ororbia et al.
@emnlp2017#emnlp2017 Super interesting !pic.twitter.com/prMqPUKMiM
-
Great talk by
@kchonyc on going down to character (bytes ?) level in NMT at#emnlp2017. Philosophical questions inside !pic.twitter.com/sbSptxhNCT
লোড হতে বেশ কিছুক্ষণ সময় নিচ্ছে।
টুইটার তার ক্ষমতার বাইরে চলে গেছে বা কোনো সাময়িক সমস্যার সম্মুখীন হয়েছে আবার চেষ্টা করুন বা আরও তথ্যের জন্য টুইটারের স্থিতি দেখুন।