
Computational linguist, machine learning scientist, cellist, random generator at @AmazonScience
Berlin, Germany
Joined February 2009
- Tweets183
- Following535
- Followers387
- Likes974
Ke Tran retweeted
1/4 #ACL2024 Excited to share our new paper on the impact of fine-tuning on the qualitative advantages of LLMs in machine translation! 🤖 Our work highlights the importance of preserving LLM capabilities during fine-tuning.
arxiv.org/abs/2405.20089
Ke Tran retweeted
Does Vocabulary Selection cause human perceived translation quality degradations not visible in BLEU? Yes! Find out more in our #NAACL2022 paper:
arxiv.org/abs/2205.06618
Code: github.com/awslabs/sockeye/t…
joint work with @EvaHasler @sonytrenous @ketran @hifelix84 @unattributed
Ke Tran retweeted
I'm recruiting for 3 Phd + 2 post-doc positions in #NLProc on multilingual neural machine translation at the University of Amsterdam. Apply by Feb 13.
PhD positions: vacatures.uva.nl/UvA/job/PhD…
Post-doc positions: vacatures.uva.nl/UvA/job/Pos…
For questions DM or email me.
@AmsterdamNLP
Ke Tran retweeted
I'm looking for a bright and enthusiastic student that will join me and the @GroNlp group, to design more Interpretable Neural MT models! 4-year salaried PhD position in beautiful Groningen (northern Netherlands), deadline 22 March rug.nl/about-ug/work-with-us… #NLProc @univgroningen
Ke Tran retweeted
I am extremely excited and proud to share with you the ELLIS PhD program. To students around the world: please apply. This will hopefully become one of the most competitive PhD programs in the world. ellis.eu/en/news/ellis-phd-p…
Ke Tran retweeted
We're looking for a talented PhD student to work on the Responsible Processing of Text Data, at the intersection of #NLP and #privacy, supervised by @turkmenf @brtvrh and myself at @univgroningen. Details here: cs.rug.nl/~bisazza/misc/RUG-… Application deadline: April 1st!
Ke Tran retweeted
About a week left to apply to join my group as a postdoc on NMT probabll.github.io/vacancies… Required: PhD (MT/NLP), experience developing NN models. Plus: unsup learning, fluent in a under-resourced language (eg tr/gu/sw/kk). Drop me a message if you have questions.
We have a page! probabll.github.io And we are hiring a postdoc to develop low-resource NMT uva.nl/shared-content/uva/en… apply before 3 May! Check what we’ve been up to and apply to join us at Probabll!
Ke Tran retweeted
My colleague Christof Monz @c_monz is looking for a PhD student to work on NLP for Responsible Data Science in his team. This is a fully funded position. Starting date: ASAP. Deadline: March 28, 2019. See ivi.uva.nl/shared/uva/en/vac…
Ke Tran retweeted
New work with the Pyro team (arxiv.org/abs/1902.03210) on efficient batched discrete variable inference through tensor variable elimination.
Practically generalizes 'einsum' with batching, semirings (log-space, max, etc), and marginals. Powers Pyro's discrete directed inference.
Ke Tran retweeted
"Tensor Considered Harmful" (nlp.seas.harvard.edu/NamedTe…).
A polemic against numpy / pytorch and a proposal for a named tensor (github.com/harvardnlp/namedt…). (New year's goal, be more troublesome.)
Ke Tran retweeted
"A Tutorial on Deep Latent Variable Models of Natural Language" arxiv.org/abs/1812.06834 including notes and derivations. EMNLP slides and code are also available nlp.seas.harvard.edu/latent-…
Ke Tran retweeted
Kyunghyun Cho Visits ILPS, with talk on November 29, 3pm, Amsterdam Science Park, ilps.science.uva.nl/kyunghyu… via @ILPS_Amsterdam
Ke Tran retweeted
New post on my #EMNLP2018 Highlights: Inductive bias, cross-lingual learning, and more
blog.aylien.com/emnlp-2018-h…
Ke Tran retweeted
We have a hot new dataset for 🔥NLI with commonsense🔥! --- 113k multiple choice questions adversarially constructed to make your favorite LM/NLI/QA models struggle 😎 Stay tuned for the leaderboard going online in two weeks! #nlproc #emnlp2018 @uwnlp
Announcing SWAG, a new natural language inference dataset, to appear at #emnlp2018. We present a general framework for collecting adversarial QA pairs at scale, minimizing bias. With @ybisk, @royschwartz02, @yejinchoinka.
rowanzellers.com/swag/
arxiv.org/abs/1808.05326
Ke Tran retweeted
Standard ML cycle:
1. Group 1 releases a dataset with massive biases, doesn't test baselines.
2. Groups 2 to 199 publish models with increasingly great results on dataset.
3. Group 200 shows that those models were probably just fitting the biases.
Seems good for everyone's CV.
New paper with @dkaushik96 is up: ***How Much Reading Does Reading Comprehension Require? A Critical Investigation of Popular Benchmarks*** We establish sensible RC baselines, finding question- and passage-only models perform surprisingly well.
arxiv.org/abs/1808.04926
Ke Tran retweeted
I am really proud of a grad in our group, but not for the reasons most commonly shared on Twitter... [1/7]
Ke Tran retweeted
Has AI surpassed humans at translation? Not even close!
I've been working with Sharon Zhou on her super accessible @skynet_today piece on the shortcomings of NMT. It's primarily aimed at a non-tech audience, so share with your non-tech friends & family!
skynettoday.com/editorials/s…
Ke Tran retweeted
Access denied: @mdr forced to skip SIGIR with many co-authored papers, only because he travelled to Iran in November & visa granting process takes more than half a year. All that & a plea for open science! Read this blog post. nitter.cf/BenCarterette/status/1…



























