Associated Professor@UCLA-CS. Research NLP, AI creativity, controllable generation, model evaluation, computational journalism, event. (she/her/hers)
Los Angeles, CA
- Tweets 835
- Following 558
- Followers 7,709
- Likes 3,186
Pinned Tweet
Thrilled, grateful, and humbled to have won 3 outstanding paper awards at #EMNLP2024!!! Not even in my wildest dreams. Immense thanks to my amazing students and collaborators!
All three works are on evaluating LLM’ abilities in creative narrative generation. 🧵👇
Big congrats 🎉🎊🍾 to UCLA PLUS lab @VioletNPeng and collaborators won 3! !outstanding paper awards at #EMNLP2024 👏👏👏
33
21
509
66,631
Two weeks ago, I heard back officially from NSF that my CAREER proposal was award.
Today, I heard back from the Vice Chancellor's office that I've been granted tenure! 🎉
I guess now I can finally say that I have a career and I go tenure! (1/n)
63
10
463
36,590
One of my most exciting results lately! We identify experts in MoE models for properties like safety and faithfulness, and steer them to improve/hurt model faithfulness and safety. Most shockingly, with stearMoE, we can jailbreak 100% safety guardrails for open models. Details 👇
🚨 You can bypass ALL safety guardrails of GPT-OSS-120B 🚨❗🤯
How? By detecting behavior-associated experts and switching them on/off.
📄 Steering MoE LLMs via Expert (De)Activation
🔗 arxiv.org/abs/2509.09660
🧵👇
5
36
261
30,594
Interested in knowing whether/how the open-domain NLG is biased? First, you need to go beyond sentiment analysis. Come to Emily’s talk at 4:00pm at #emnlp2019 session 7C AWE 203-205 to learn more.
10
58
216
I’m grateful for the enormous support from the community! It’s an honor to serve, and I’m excited to work hard alongside all of you to make a meaningful impact!
🎉🎉🎉 Congratulations to new #NAACL board members:
Steven Bethard @stevenbethard (Treasurer)
Nanyun (Violet) Peng @VioletNPeng
Helena Gómez-Adorno @helen_py
naacl.org/posts/2024-12-21-N…
Thank you all for voting!
12
10
160
17,661
For this special @aclmeeting, we created special 24 handbooks for 24 timezones bit.ly/3fbi1XL with links to paper PDF and presentation website for each paper. Should be very handy; please check out the one for your time zone! #NLProc Credit to @mingyu_ma
4
59
134
It’s a wrap! I hope y’all enjoyed #EMNLP25 as much as I did! Big shoutout to the photography team! All my photos in this post are taking from their websites. To all attendees: you should check out if you haven’t already!!
8
4
127
8,042
This Thanksgiving, I’m deeply grateful for opportunity to run for NAACL Board Member alongside so many incredible candidates who share a passion for making a difference!
I hope to earn your support as we work together to shape the future of our community! #NLProc #Gratitude
3
19
115
22,364
I’m organizing a workshop on Creative AI Across Modalities #CreativeAI at #AAAI23 in DC, great lineup of speakers touching creative generation for language, image, video, audio, and more. If you’re around and are interested, please come by!
7
22
103
23,619
Excited to kick start #EMNLP25 with my awesome co-chairs @c_christodoulop Carolyn Rose, @Tanmoy_Chak in Suzhou for the 30th anniversary! Especially happy to have hosted one of my favorite researchers @hengjinlp on a super inspiring keynote talk with a full house!
@VioletNPeng served as the Program Co-Chair for #EMNLP25, one of the largest NLP conferences, which received over 8,000 submissions and drew 6,000 participants.
1
9
102
20,995
Excited to share our #Neurips2022 paper on controllable text generation with theoretical guarantees. We decompose a sequence-level oracle into token-level guidance to steer the generation to consider future constraints. Impressive results for incorporating OOVs.
Happy to share our work on Controllable Text Generation with NeurAlly-Decomposed Oracle (NADO) accepted at #Neurips2022! Looks like our model is excited about NeurIPS as much as we do. See more in arxiv.org/abs/2205.14219
2
16
92
I’m super excited to share several fresh off the press works on controllable (creative) generation this Thursday at Stanford. It’ll be hybrid with zoom livestream. Please register if you’re interested!
This Thursday at 11 AM PT, we are beyond excited to host @VioletNPeng for our weekly NLP Seminar! Nanyun will be talking about Controllable Text Generation For Open-World Creativity. Registration form is here: forms.gle/GBeHq4nAiawN56or9
1
12
88
#NAACL2022 is #PlusLab @uclanlp's lucky conference! We have 7 papers (out of 9 we submitted) accepted to the main conference. As the cherry on top, @AlexanderSpangh's paper won an outstanding paper award!
#NAACL2022 is announcing best paper awards and outstanding papers: 2022.naacl.org/blog/best-pap…
Please check them out!
5
8
88
I’ve seen many questions about how to choose ARR tracks for submissions aim at the new tracks at #emnlp2025. We actually wrote a blogpost along with the 2nd CFP exactly to address this: 2025.emnlp.org/track-changes…
Please help us share it widely! Good luck with your emnlp submissions!
Happy to see #EMNLP2025 introducing new tracks on AI/LLM Agents, Code Models, Safety & Alignment, Reasoning, LLM Efficiency, and more. Big thanks to the organizers for making this happen! @emnlpmeeting #NLProc
Perfect venue for agentic research and language technologies. Submissions are due 5/19! 2025.emnlp.org/calls/main_co…
2
19
80
16,638
I'm at #AAAI19 presenting our work on storytelling. We proposed a hierarchically generation strategy plan-and-write arxiv.org/abs/1811.05701
Please play with our demo cwc-story.isi.edu Guaranteed fun with the interactive mode. :)
Friends at Honolulu: let’s catch up! #nlproc
1
11
76
We’re starting a new NLP seminar series! If you’re visiting the area and want to stop by, please let us know!
🚨 New NLP seminar series alert! 🚨
Check out UCLA NLP Seminar series featuring cutting-edge talks from top researchers in NLP and related areas.
Great lineup, timely topics, and open to all (zoom)! 🧠💬
📅 Schedule + details: uclanlp.github.io/nlp-semina…
2
7
77
8,631
If you’re interested in AI for humanity — computational journalism, creativity, narratives, etc. You should apply to work with Alex!
✨ Very overdue update:
I'll be starting as an Assistant Professor in CS at University of Minnesota, Twin Cities, Fall 2026. I will be recruiting PhD students!!
Please help me spread the word! [Thread] 1/n
1
7
75
9,511
Super proud of my PhD student @AlexanderSpangh who just got a Bloomberg Ph.D. Fellowship to work on computational journalism, natural language generation! Way to go Alex! Proudly co-advised with @emilio__ferrara.
4
3
71
Finally wrapped my first time serving as a program co-chair. Learned so much from my fellow co-chairs and felt bitter sweet to say bye to our Monday regular meetings.
Ensuring peer review quality and getting great ideas popularized is an ongoing mission. Next: #EMNLP25 Fight on!
Huge thanks to the #ICLR2025 Organizing Committee (including many who couldn't make it to the conference) 👏👏👏
2
1
72
5,420
Very proud and excited to share this work done at #AlexaAI @AmazonScience that composes original stories with visuals and sound effects for kids, based on user customized inputs. This pushes Alexa’s mission to be a companion for millions of customers 1/2 rotf.lol/yckn8cr3
3
7
68
🌟 Excited to share our latest work on controllable generation! 🚀 By integrating neurosymbolic techniques to instruction-tuned LLMs, we bring you the best of both worlds—precision and flexibility. Unlike purely prompt-based control, our approach offers precise controls such as keywords/phrase incorporation, word count control, text infilling etc. with strong guarantees. Check it out!
Proposing Ctrl-G, a neurosymbolic framework that enables arbitrary LLMs to follow logical constraints (length control, infilling …) with 100% guarantees.
Ctrl-G beats GPT4 on the task of text editing by >30% higher satisfaction rate in human eval.
arxiv.org/abs/2406.13892
14
70
11,318
#pluslab members, mentors, alumni, and friends at @uclanlp had so much fun at #NAACL2022!
Looking forward to seeing everyone in the next conference!
1
4
67
The most ridiculous and funny way to delay a flight: it was “fueled too much to take off” and we stay still to “burn the fuel” for a good 45 mins…
What a good way to start the #EMNLP trip. 🤣
See everyone very soon in Singapore!
1
1
62
6,681
A special thank you to my full-spectrum collaborator, dearest colleague, best friend, and life partner @kaiwei_chang, and my 3-year-old girl for keeping my life happy and sane. Here's to more research adventures ahead! (n/n) #AcademicTwitter #Tenure
1
63
5,028
Attending #COLM for the first time. Heard too many great things about the conference to miss it!
I’ll stay until Thursday afternoon. Would be excited to chat about AI4FormalMath, Deep Research Agent, AI creativity, and Controllable Gen. Catch me at Palais des Congrès Montreal!
4
5
63
7,770
Happy to serve the community again! Looking forward to working with my awesome co-chairs! See you in Suzhou!
📢 📅 After a long process of soliciting & vetting bids, I'm excited that we've finally been able to reveal the location for #EMNLP2025 -- it'll be at the International Expo Centre, Suzhou, China from 5-9 November 2025. Looking forward to seeing you there!
@emnlpmeeting #NLProc
3
59
8,268
Excited to speak more about AI creativity at SSNLP today in Singapore ssnlp-website.github.io/ssnl…
Also look forward to hear what Qwen team has to say about their latest breakthrough!
Friends in Singapore: let’s catch up!
1
7
59
5,997
Dear friends at #NeurIPS2022, #PlusLab_UCLANLP will be presenting three papers at the main conference and one at the ENLSP workshop. This is a thread of some details about our presentations. My students and I are looking forward to seeing you at NOLA! (1/5)
1
13
54
Super excited to be at #IJCAI2022 to give the early career spotlight talk. I'll introduce some of my group #pluslab's work on creative generation and fairness in NLG. Come say hi if you're around!
p.s. first time visiting Vienna -- such a nice city!
#IJCAI2022 Early Career Spotlight Talks on 26 July:
🗣️Amos Azaria @arieluniversity
🗣️Hau Chan @UNLincoln @NebEngineering
🗣️Nanyun (Violet) Peng @VioletNPeng @UCLAComSci
🗣️Ferdinando Fioretto @nandofioretto @EngineeringSU
Chair: IJCAI-ECAI 2022 Chair Maria Gini @UCIrvine
2
3
52
I hope everyone is enjoying #EMNLP20! While it's fresh, I'd like to share several works from my group that will be presented in the next few days with details including links to papers, code, slideslive, twitter, etc. We look forward to meeting you in conf the discussions! 1/N
1
7
48
What a nice way to start #EMNLP2023!
I’m especially proud of our undergrad Stanley for his initiative and enthusiasm; @Wade_Yin9712 for the mentorship! I’m mostly just a cheerleader.
Thank the organizers of #Pan-DL for putting together this nice workshop and for the recognition!
😄Got my first award!
Our LEAF: Linguistically Enhanced Event Temporal Relation Framework got 🏆best paper award at Pan-DL workshop #EMNLP2023!
❤️Thanks Stanley Lim and @VioletNPeng! 1st author Stanley is a UCLA undergrad and it’s his 1st NLP proj! Excited to see his growth!
1
1
48
5,532
Just had a 1-hour meeting with such a diverse group of undergrad, graduate, and post-doctoral researchers in CS. Guess what topic we discussed? w/ @ewsheng @jieyuzhao11 @JiaosunT @sunipa17 @houyu0939 @ovalle_elia @mattiesansev @kaiwei_chang Jinn Kim
3
3
45
Working on story generation for years, these are some best generation results I’ve ever seen. Apparently far from perfect and takes several resampling, but checkout the story our system generated about us building our story generation system…🤣
To be presented at #EMNLP2022
Excited to share our EMNLP 2022 paper (w/ @VioletNPeng, @tydsh, Dan Klein): "Re3: Generating Longer Stories With Recursive Reprompting and Revision"!
We *automatically* generate coherent stories, with consistent overarching plot, of *2000+ (!)* words, or even up to 7500!
(1/11)
2
3
47
We are hiring for all areas, all ranks! Send your application in. And if you want to know more about how awesome UCLA CS is, DM me! recruit.apo.ucla.edu/JPF0782…
1
39
My group built up some capabilities for biomedical discourse tagging and event extraction under an NIH grant. We annotated the #CORD19 dataset using our tools for the research community. More details, benchmark, and the annotated data tidy.ws/5fkizw.
3
8
40
Tired of complaining about *CL reviews on social media and conferences? Now’s your chance to make a real difference! Your feedback is invaluable in improving the process. Take a few minutes to share your thoughts—every response counts!
@ReviewAcl and the EMNLP PCs are seeking input on the current state of reviewing for *CL conferences. We would love to get your feedback on the current process and how it could be improved. To contribute your ideas and opinions, please follow this link! forms.office.com/r/P68uvwXYq…
6
36
3,510
@kaiwei_chang is running for VP-elect for SIGDAT with the vision to empowering young researchers, fostering a diverse community, strengthening industry-academia collaboration, and more. Your support and help disseminating the information would be much appreciated!
I am honored to be nominated by SIGDAT (the org that oversees EMNLP) to run for VP-elect with other awesome candidates who share the goal of improving our community. Please check your email to vote by 3/24.🗳️ See details: bit.ly/3ItRc0S
8
42
5,311
Excited about several innovative ideas we’re trying out around AI/ML conference reviewing.
Yes, we’re fearless!
For #ICLR2025, we're piloting a feedback agent that provides optional feedback to reviewers. The aim is to help make reviews more constructive and actionable for authors. blog.iclr.cc/2024/10/09/iclr…
4
39
8,052
Thank @UCLAengineering for covering my research! Shout out to @yufei_t who is now all over the place for creative generation: poetry, stories, puns, hyperboles, and soon lyrics! If you’re interested in these topics, please follow her! :)
Having a writer's block? Did you know AI can help? @UofCalifornia @UCLA professor Nanyun (Violet) Peng @VioletNPeng @UCLAComSci has created an #AI algorithm able to write a sonnet and more. samueli.ucla.edu/ucla-comput…
1
5
41
I want to extend my sincere thank you to all my students, mentors, collaborators, colleagues, my hero(ine) letter writers, and everyone who supported me on this journey. Nothing will be possible without the supports and help I got along the way! (2/n)
1
39
4,136
Finally got in for #NAACl2024! Look forward to seeing old and new friends!
Today from #PlusLab:
1. Oral presentation@2pm on hallucination evaluation (tinyurl.com/newpmkrw) in the Summarization session by @HaoyiQiu
2. Oral on event detection for pandemic prediction (tinyurl.com/y67d6w78) @2:36pm in the IE session by @tparekh97
3. Afternoon poster on creative problem solving (tinyurl.com/6m5cbvtw) starting at 2pm by @yufei_t with proud collaboration with AI2 @faeze_brh @Lianhuiq @Ronan_LeBras @YejinChoinka
Come catch up with us!
8
38
5,997
I'm excited to co-organizing the WordPlay workshop at #ACL2024. You still have 3 weeks to prepare your submission and papers accepted/undersubmission to other venues are welcome! Look forward to seeing you in Bangkok!!
Reminder that there's only a couple more weeks (May 31) until the deadline for the Wordplay: When Language Meets Games workshop at ACL in Bangkok!!
Submit all your papers on language agents, simulations, narrative, AI for games, and more!!
ALT ASCII art that says Wordplay when language meets games
6
37
6,710
Still managed to have some fun at #EMNLP2020. Look at us! 70% females! #PlusLab@EMNLP2020. #NLProc PC: @TuhinChakr
2
36
I like the findings model, but would appreciate if there are clearer guidelines for reviewers, ACs, and SACs about how to decide which papers to go for which track.
3
35
Tired of prompt-based control for NLG without any guarantee? GeLaTo 🍨 presented an alternative framework for control with tractable probabilistic models (TPM) that gives you 100% adherence to the constraints. Just accepted to #ICML2023 oral. @HonghuaZhang2 @meihuadang @guyvdb
Reliable control of large language models is a crucial problem. We propose GeLaTo (Generating Language with Tractable Constraints), a neuro-symbolic framework that allows LLMs to generate texts following logical/lexical constraints with 100% guarantee. See arxiv.org/abs/2304.07438
1
2
34
6,119
Remember AMR?🤭 Turns out, with its clear, structured representation of the semantic information in text, we can targetedly generate challenging negative examples with disguising hallucination to train automatic evaluators to spot them! #NAACL24
🔥 Unlocking the power of Abstract Meaning Representations, AMRFact generates coherent, factually inconsistent summaries with high error-type coverage to improve the factuality evaluation on abstractive summarization!
📣 Check out our new #NAACL2024🇲🇽work: arxiv.org/abs/2311.09521
6
32
4,531
Where do people purchase GPU servers nowadays? The vendor I have been using has a hefty 1-year lead time for A100/H100 servers. Any recommendations on faster options or reliable suppliers would be greatly appreciated! #TechCommunity #GPU
10
4
33
16,929
Glad to serve as one of the publicity chairs! Please follow this official account (and us) for more updates from #NAACL2022!
Hello 👋 from the NAACL 2022 Publicity Chairs (@VioletNPeng @sameer_ @ewsheng)! We will be using this account to keep everyone updated on conference-related details. Please follow + retweet!
2
32
Thanks for the mention. Shout out for awesome
@ewsheng
who did the great work! A small disclaimer: the examples are selected in the interest of space. But we showed distributional biases (see figure). More details see the paper: arxiv.org/abs/1909.01326
9
28
Please come meet and talk to our awesome students! We look forward to establishing precious research friendships!
Excited for #EMNLP2023? #UCLANLP & members @cfyang58 @Wade_Yin9712 @FanYin63689862 @kamath_amita @DiWu0162 @P_N_Kung @telinwu @sunjiao123sun_ @HaoyiQiu @yufei_t @yu_bryan_zhou will present a range of work on NLG, fairness, instruction tuning, and multi-modality. 🧵👇 (1/18)
2
31
4,551
Proud of this work where we show event detection can generalize from one epidemic to another by identifying epidemic-related event types (e.g. symptoms) even if the actual mentions (e.g. of symptoms) are distinctive. So training on COVID, we can generalize to Monkeypox! #NAACL24
🦠How to avoid a future COVID pandemic? 🔍How to detect epidemics early?
Excited to share our latest work SPEED - an Event Detection framework to extract epidemic-related events from Tweets and can provide breakthrough early warnings for the unseen epidemic of Monkeypox! 🤯😱
6
31
5,116
Excited about this upcoming visit. Look forward to meet UniBocconi faculty and students!
DMI Talk Series - Fall 2021
"Controllable Natural Language Generation for Fairness and Creativity" by @VioletNPeng (University of California, Los Angeles).
📅 When: September 8th, 18:00-19:00 CEST
🗒️ Details: dmi.unibocconi.eu/wps/wcm/co…
®️ Register: unibocconi-it.zoom.us/meetin…
4
30
Very interesting workshop with such a cool-looking website! Guess what? I’ll be speaking on this workshop about creative language generation. Please consider submitting your works and/or attending the workshop!
Interested in the intersection of AI, language, and computer games?
The NeurIPS Wordplay Workshop deadline is 9 October 2020.
Workshop held 11 or 12 December 2020
wordplay-workshop.github.io/…
4
29
Really happy for @Sarikgha who got her first full conference paper accepted by #AAAI2020! @RealAAAI I’m especially proud of seeing her gradually growing into a confident researcher! #womenintech
Also another successful collaboration w/ @aram_galstyan and Ralph @USC_ISI
Pleased to share that our work with title “Predictive Engagement: An Efficient Metric for Automatic Evaluation of Open-Domain Dialogue Systems” has been accepted at AAA20 @RealAAAI. It’s a joint work with @VioletNPeng @aram_galstyan and Ralph Weischedel.
arxiv.org/abs/1911.01456
1
2
30
Will be giving a tutorial on KB construction and reasoning with @xiangrenUSC @WilliamWangNLP tomorrow morning at NAACL! Stay tuned!
Heading to @NAACLHLT and will teach a tutorial on scalable KB constriction & reasoning (focusing on learning with limited supervision; low resources; reasoning) on Fri morning w/ @VioletNPeng @WilliamWangNLP. Look forward to meeting folks! #NAACL2018 #nlproc #IE #KBC
1
5
30
As a graduate student, I used to use the softconf trick to check *CL conference decisions before the emails are sent out. Now for the first time, I refreshed openreview to got my Neurips papers decisions. Maybe the field hasn’t really changed too much after all…🤣
30
We got popular requests to volunteer to serve as an AC at #ICLR2025.
Here’s the nomination link so you can volunteer yourself or others!
We are now accepting nominations for reviewers and ACs at ICLR 2025.
To nominate yourself or someone else, please complete this form: forms.gle/VKmG1DJgWzKTY9WJ8
1
3
28
6,947
Congratulations Dr. Sheng @ewsheng for successfully defending her thesis! I enjoyed every moment working with and learning from Emily. can’t wait to see the amazing things she’ll do getting out of the door! My first Phd; proudly co-adviced with @natarajan_prem #AcademicMama
I’m a doctor now! A huge thank you to all those who have supported me on this long journey, especially my advisors, @natarajan_prem and @VioletNPeng, as well as @kaiwei_chang, Shri, and Yan 🎉. Hopefully this work inspires future work in and beyond fairness in NLG :)
2
26
Thanks for the shout out! My student @ewsheng did all the ground work. More to come!
2
21
Highly recommended everyone to apply!
Apply to our PhD program! We accept multiple students every year in speech and NLP in @JHUCompSci @JHUECE @JhuCogsci
You can learn more about the application process, and get helpful tips in general, from our FAQ:
clsp.jhu.edu/apply-for-phd/p…
1
24
Thanks for having me! It was a great experience visiting the Stanford NLP group!
Today we’re delighted to have Nanyun @VioletNPeng from @nlp_usc talking to us about low-resource settings—domains like biomedical text as well as languages #NLProc
22
Adding to the stack of numerous QA datasets… but this one is special. We look into event semantic relation reasoning, including causal, conditional, counterfactual, subevent, and coreference, which are shown to be challenging for models. Pick up your hammer 🔨and give it a shot!
(1/5) Introducing our #EMNLP21 paper “ESTER: A Machine Reading Comprehension Dataset for Event Semantic Relation Reasoning.” We invite everyone interested in event-centric reasoning to test your models on ESTER and submit results to our leaderboard: github.com/PlusLabNLP/ESTER
2
24
Since we’re giving more paper awards, maybe we can consider revive the non-parallel sessions for orals and posters and only give orals for the award winning papers?
23
1,135
Sure, should’ve attached the paper to begin with. Here you go: arxiv.org/pdf/1909.01326.pdf
1
2
21
Bias in NLG is more subtle to define yet it’s especially important since many NLG systems *directly interact* with users. @ewsheng did a great job surveying papers about biases in NLG with summarization of tends and challenges. Come chat with us if you’re interested in the area!
Come chat with us about societal biases in NLG at the #acl2021nlp poster session tomorrow! (Tues @ 5pm PT)
4
21
Another four great talks conclude this year’s #SoCalNLP. @gh_marjan @nnakashole @AnimaAnandkumar @LukeZettlemoyer
Also had a great outdoor poster session (that’s what SoCal is all about). See everyone next year!
1
21
Glad to contribute some to the collection of @nlp_usc at @emnlp2019. Some milestones to share...
USC NLP has 13 papers accepted at @emnlp2019 by students of @VioletNPeng @jonathanmay @xiangrenNLP. Details coming soon. We will see you at Hong Kong! #NLProc @USCViterbi @USC_ISI @CSatUSC
2
22
If you’re at #NAACL2021, the #pluslab at #uclanlp will be presenting quite some papers covering the topics of language generation; fairness and social NLP; event extraction. More details about the papers and the schedule in the quoted tweet 👇. We look forward to meeting you!
Dear friends, UCLANLP🐻(@kaiwei_chang’s lab & @VioletNPeng’s #PLUSlab) will present the following papers/demo at #NAACL2021
Check out the detailed paper information at 👇web.cs.ucla.edu/~kwchang/blo…
Look forward to meeting you all at the conference! (1/14) #NLProc
2
22
A super fun and exiting collaboration with @MohitIyyer’s group on story generation! We have a unique dataset with interesting metadata about challenges and goals that the characters face. Submit your generative model to our platform and have real users evaluate them! #CreativeAI
5
22
I am grateful for every moment I've spent at USC ISI, where the institute and my colleagues provided me tremendous help and support to start my career. I look forward to continuing the collaborations and keeping a spot in the beautiful MDR! 2/2
2
20
Speaking at the #Neurips2020 #Wordplay workshop about my group’s work @seraphinagt @TuhinChakr on creative generation. Thank you for having me!
The fourth invited talk is by @VioletNPeng on Creative Language Generation: Stories, Sarcasms, and Similes.
6
21
#AAAI20 friends, come see @Sarikgha presenting our work on evaluating open domain dialog sys! We showed that the engagement of a conversation could be predicted early on with individual utterance pairs, and yield high correlation with human! @aram_galstyan @USC_ISI
Super excited to present our latest work on the automatic evaluation of open-domain dialogue systems by leveraging predictive engagement at @RealAAAI. If your are interested to know more please come and visit our poster #NLP6421 tomorrow evening (02/10)!
3
20
I really like the gathertown feature for paper presentation. Reminds me a lot of the physical conferences.
Also, miss the real conferences soooo much. This is PlusLab at #EMNLP2019 Hong Kong.
19
Glad to be part of the exciting event: Amazon and UCLA just launched a center for humanity and AI today! Look forward to pushing the frontier of computational humanity!
Thrilled to participate in the launch event of the Science Hub at UCLA. Looking forward to the many advances that this partnership will generate! @JayathiMurthy @UCLAengineering @AmazonScience @VioletNPeng @kaiwei_chang
2
18
Hi, Friends at @aclmeeting, my group and collaborators are in-person in Dublin to present several papers at #ACL2022.
Here's one in collaboration with colleagues in software engineering to test and improve model robustness with Sibylvariant Transformations
Augmentations inject more diversity into your data, but nearly all are constrained to preserve the source semantics. Turns out that you can knowably transform the data and label together and this can significantly improve model performance!
arxiv.org/abs/2205.05137
#acl2022nlp
1
3
19
Alex is an extraordinary researcher with a truly unmatched combination of journalism, CS, and music expertise. His work in computational journalism and narratives stand out as innovative and with profound positive impact. Any department should consider hire him! #CSJobs
✨✨✨Hello everyone, I’m on the faculty job market this year.✨✨✨ I’m completing my PhD at USC, where I study agentic planning in creative contexts.
But before I get deeper into my research, I really want to tell you a little bit about myself :)
19
4,903
GATE is a light-weighted approach to efficiently fuse graph structures into transformers. We show its use case on cross-lingual transfer; maybe you can use it for your applications. Come to our poster #AAAI2021. Room: BC-D1-R2, Feb 4, 08:45-10:30AM, 04:45-06:30 PM PST. #UCLANLP
How to transfer a relation/event extraction model across different languages by leveraging parse structures? 🧐 Come to our Graph Attention Transformer Encoder (GATE) paper (bit.ly/39OwIzH) at #AAAI2021 @ahmadwasi @VioletNPeng #UCLANLP (1/3)
2
17
My hypothesis: it’s simply because autoregressive training needs less engineering decisions (what will you encode, what will you decode, how much to mask out, etc.) especially when your data/tasks are more diverse. Also, it’s easier to generate sentences.
15
1,199
Come join us!
We're hiring in NLP! @USCViterbi Computer Science has multiple openings in AI: NLP, ML, Vision, AI+X, etc. Ping me if you are interested and pls forward to people who will be on the market! Happy to chat at @emnlp2019 too. Pls help RT! cs.usc.edu/about/open-facult…
15
Finally got to metaphors! We contribute an automatically curated metaphor dataset, along with a fine approach to generate novel metaphors. Another fun collaboration with @TuhinChakr and Smara, and more to come!
🔥Metaphors are not to be trifled with🔥 Excited to share #NAACL2021 preprint titled “MERMAID: Metaphor Generation with Symbolism and Discriminative Decoding”
arxiv.org/pdf/2103.06779.pdf . Joint work with my figurative NLG constants @VioletNPeng and Smaranda Muresan. #NLProc
4
16
Everyone knows the importance of data, and thus synthetic data. But how to generate them so they help recognizing novel concept? Contractive features are what you need! 👇
#GPT4o image generation brings synthetic visual data quality to the next level. 🖼️
🤔Is synthetic visual data finally ready to be used for improving VLMs?
🚀 We show success with CoDA, using contrastive visual data augmentation to help teach VLMs novel and confusing concepts.
2
16
4,432
Come work with us! More topics including creative generation, BioNLP, event-centric NLU! #UCLA_NLP #SoCalNLP @CS_UCLA
DM me if you're interested in a postdoc opportunity at UCLA NLP to work on exciting topics including fairness/robustness in NLP, machine common sense, cross-lingual transfer, language generation, and more! @VioletNPeng
2
15
Done with the morning session. Tons of interesting discussions and questions. Thank the organizers @LuWangCS @feiliu_nlp Giuseppe_Carenini Jackie_Chi_Kit_Cheung for having me!
@VioletNPeng emphasizing on how creative generation can be viewed as an inverse process of summarization at Workshop on the frontiers of Summarization #nlproc #emnlp2019
1
14
Great team work across several groups and especially my creative AI ladies @anjalisaa @Ale_Cervone Shereen.
Stay tuned for the actual launch and looking forward to more on #creativeAI!! 2/2
3
14
I have long suspect that GPT-3 memorizes some classics. My colleague once prompted GPT-3 to reproduce Harry Potter and then got weird access error for a while.
This is a smart approach!
The IP implication is worrisome…
New paper out today, asking: What books has ChatGPT/GPT-4 *memorized*? A LOT. Harry Potter, Pride & Prejudice, 1984, LotR, Hunger Games, GoT, 50 Shades of Grey, Dune. Memorization is linked to web popularity--lots of old classics + new sci-fi/fantasy. 1/6 arxiv.org/abs/2305.00118
1
1
14
4,409
I always have too much fun working on creative generation. But for this one, “I had so much fun even my tailbone puts on a happy face!” 🤣 (copyright belongs to our model HypoGen). We can also break the pattern and have more creative generation via paraphrasing. Check it out! 👇
Is generating hyperboles easy? Our machine says yes!
Check our new #EMNLP2021 Findings paper "HypoGen: Hyperbole Generation with Commonsense and Counterfactual Knowledge" with Arvind and @VioletNPeng!🧾arxiv.org/pdf/2109.05097.pdf
Code and data coming soon!
ALT Written by our HypoGen model given "Generating hyperboles is easy" as input: Generating hyperboles is so easy, even a computer gets a job!
1
13
Will the Prize in Physiology or Medicine be the next?
Time for me to aspire one in Literature!
BREAKING NEWS
The Royal Swedish Academy of Sciences has decided to award the 2024 #NobelPrize in Chemistry with one half to David Baker “for computational protein design” and the other half jointly to Demis Hassabis and John M. Jumper “for protein structure prediction.”
2
14
2,198