Natural language understanding / machine translation / book reading
New York
Joined December 2011
- Tweets869
- Following262
- Followers239
- Likes409
Raphael Shu retweeted
Introducing OpenAgents Workspace.
One workspace where all your AI agents work together, shared chat, shared files, shared browsers.
Connect OpenClaw, Claude Code, Codex and more. Watch them collaborate, and work while you sleep.
Open source on GitHub.
Please give us a star:
github.com/openagents-org/op…
#OpenSource #AIAgents #MultiAgent
Raphael Shu retweeted
My agent just shipped a feature, and another announced it on X and LinkedIn.
I didn't write a word.
OpenAgents Workspace automates the full loop.
One workspace. Two agents. See how coding and marketing agents collaborate with each other.
Try it→ github.com/openagents-org/op…
#AIAgents #OpenSource #MultiAgent
Raphael Shu retweeted
Introducing OpenAgents — the open-source framework for building AI Agent Networks.
From isolated bots ➝ collaborative communities of countless numbers of agents that work, learn & grow together.
The future is an internet of agents. Star us on Github for OpenAgents Day 1 badge.
Raphael Shu retweeted
Happy to share our new #BLEURT-based reward optimization using a ranking-based contrastive margin loss for #NMT. We reconfirm the limitation of BLEU from our reward optimization results and human evaluation.
Congrats to
@raphaelshu, Kangmin Yoo.
arxiv: arxiv.org/abs/2104.07541
Raphael Shu retweeted
New work: "Iterative Refinement in the Continuous Space
for Non-Autoregressive Neural Machine Translation" w/ @raphaelshu @kchonyc
arxiv.org/pdf/2009.07177.pdf
We propose an inference procedure for non-autoregressive NMT that refines translations *purely in the continuous space.*
Raphael Shu retweeted
Happy to announce our most recent paper: BLEU might be Guilty but References are not Innocent - arxiv.org/abs/2004.06063 - We show that it is possible to calculate reliable automatic scores (even with BLEU) for high quality MT output by using a novel reference generation method.
Raphael Shu retweeted
Animated attention by @kchonyc which took his good 3hrs @facebookai @PyTorch #torched #deeplearning #nlp #attention #seq2seq #kyunghyuncho #nyu #cds
Raphael Shu retweeted
“On the Discrepancy between Density Estimation and Sequence Generation”
arxiv.org/pdf/2002.07233.pdf
Seq2seq models are optimized w.r.t log-likelihood. We investigate the correlation btw. LL and generation quality on machine translation.
w/ @dustinvtran, @orf_bnw, @kchonyc. (1/n)
Raphael Shu retweeted
Some representative references on each group with more exciting approaches missed and more to come soon!
Raphael Shu retweeted
The "Deep Learning Toolbox" has greatly expanded in the last decade thanks to our wonderful research community. Also, important progress has been made to make our community more inclusive and less toxic. Still, there's LOTS to do, and I plan to keep focusing on advancing both.
Raphael Shu retweeted
At the very last day of this decade our paper w Omar, Seokho and @kchonyc on molecular geometry prediction with generative graph neural nets has been published to @SciReports. Check it out here nature.com/articles/s41598-0…. Happy New Year everyone!
Raphael Shu retweeted
Decisions released 🎉 Congratulations to accepted papers; to those who we could not accommodate, we wish you success in your ongoing research. See our blog for the first of our reflections. See you soon in Ethiopia. #ICLR2020 #OurHatata 🇪🇹🌍link.medium.com/ysHgHxoLy2
Raphael Shu retweeted
What does a pruned deep neural network "forget"?
Very excited to share our recent work w Aaron Courville, Yann Dauphin and @DreFrome
weightpruningdamage.github.i…
Raphael Shu retweeted
Sam Bowman @sleepinyourhat is giving an invited talk at @uwnlp! Check out the live video at youtu.be/8BX3gafxfMI
Raphael Shu retweeted
desperate plea for help from AAAI program chairs -- everyone please stop trying to seeing your reviews in CMT so we can fix the problem -- CMT can't handle the load at all and we can't get in anymore ourselves! @RealAAAI
Raphael Shu retweeted
Talk I gave explaining strategies to supercharge your PyTorch code (16-bit, multi+single node parallelization, dataloaders, accumulated grads, and the @PyTorchLightnin flags to turn these features on. m.youtube.com/watch?v=TM_jRr…
@MILAMontreal @AmiiThinks @NYUDataScience @stanfordnlp
Raphael Shu retweeted
The #DataScience @UDelaware community got treated to 2 outstanding talks in the last 2 days! @Prof_DavidBader on “Massive Scale Anslytics” and @kchonyc on “A Generalized Framework of Sequence Generation” – at Delaware, USA
Raphael Shu retweeted
"The traditional declarative programming model of building a graph and executing it via a tf.Session is discouraged" the end of an era github.com/tensorflow/tensor…
