Chia-Chien Hung (@cc_hung_) profile picture

Chia-Chien Hung

@cc_hung_

Senior Research Scientist @NECLabsEU (Human-Centric AI) | NLP Researcher @dwsunima

chiachienhung.github.io
Joined November 2021
  • Tweets14
  • Following116
  • Followers66
1Media
Photo by Chia-Chien Hung (@cc_hung_)
📢Looking for a Ph.D. position in Natural Language Processing/ AI?!📢 We offer a fully-funded position for three years with the possibility of choosing from a broad range of topics! 🥳 Reach out to me if you are interested or apply directly! uni-hamburg.de/stellenangebo…

Ausschreibung

uni-hamburg.de
5
26
1
65
11,830
We are excited to share that the Workshop on Human-Centered Large Language Modeling (HuCLLM), our 1st edition will take place @aclmeeting #ACL2024NLP in Bangkok, Thailand! More details can found on our website: hucllm-workshop.github.io/ Cfp to follow soon 🚀 #NLProc
Image from Workshop on Human-Centered Large Language Modeling (@HuCLLM)
8
1
56
10,403
Chia-Chien Hung retweeted
Kiril Gashteovski (@kgashteo) profile picture
Kiril Gashteovski@kgashteo8 Dec 2023
Come to our poster “Linking Surface Facts to Large Scale Knowledge Graphs” at the East Foyer, 4B #emnlp2023
Image from Kiril Gashteovski (@kgashteo)
2
9
528
Happy to present our Efficient Task-Agnostic Domain Adaptation work @#ACL2023, which was done during amazing PhD Sabbatical @Bosch_AI 🙌Thanks my co-authors: Lukas Lange @jannikstroetgen 📖aclanthology.org/2023.findin… 👋In-person poster: RepL4NLP workshop (HarbourB) today@11-12 (1/2)
TADA: Efficient Task-Agnostic Domain Adaptation for Transformers

TADA: Efficient Task-Agnostic Domain Adaptation for Transformers

Chia-Chien Hung, Lukas Lange, Jannik Strötgen. Findings of the Association for Computational Linguistics: ACL 2023. 2023.

aclanthology.org
1
9
1,098
We proposed TADA, a modular and efficient task-agnostic domain adaptation method, demonstrating the efficacy in 4 tasks across 14 domains in both single- and multi-domain settings. Additional analysis highlighted the advantages of TADA in few-shot transfer setup. (2/2)
1
246
Chia-Chien Hung retweeted
MilaNLP (@MilaNLProc) profile picture
MilaNLP@MilaNLProc5 May 2023
aclanthology.org/2023.findin… Demographic factors shape language and improve #NLProc tasks, but can modern language models incorporate them? It's complicated. The performance gains we see often stem from other factors. Demographic specialization remains unsolved.
Can Demographic Factors Improve Text Classification? Revisiting Demographic Adaptation in the Age...

Can Demographic Factors Improve Text Classification? Revisiting Demographic Adaptation in the Age...

Chia-Chien Hung, Anne Lauscher, Dirk Hovy, Simone Paolo Ponzetto, Goran Glavaš. Findings of the Association for Computational Linguistics: EACL 2023. 2023.

aclanthology.org
1
2
3
852
Happy to share that I presented our work about Demographic Adaptation at Insights Workshop today at EACL2023. 🙌Thanks to my co-authors: @anne_lauscher @dirk_hovy @gg42554 @spponzius 📖Check here: aclanthology.org/2023.findin… (1/2)
Can Demographic Factors Improve Text Classification? Revisiting Demographic Adaptation in the Age...

Can Demographic Factors Improve Text Classification? Revisiting Demographic Adaptation in the Age...

Chia-Chien Hung, Anne Lauscher, Dirk Hovy, Simone Paolo Ponzetto, Goran Glavaš. Findings of the Association for Computational Linguistics: EACL 2023. 2023.

aclanthology.org
1
2
16
1,870
Specialization methods proven effective for other types of knowledge (e.g., dialog, geographic), fail to adequately instill demographic knowledge into PLMs. The results indicate that demographic specialization still an “open problem” in the age of Transformers. (2/2)
2
180
Happy to be at #NAACL2022 presenting our work! We introduce Multi2WOZ, a robust multilingual multi-domain task-oriented dialog (TOD) dataset, spanning four typologically diverse languages (DE, RU, AR, ZH). @anne_lauscher @licwu @spponzius @gg42554 aclanthology.org/2022.naacl-… (1/2)
Image from Chia-Chien Hung (@cc_hung_)
2
6
16
This work is aimed at bridging the gap of systematic investigation of cross-lingual transfer approaches for TOD. We show that our novel conversational specialization framework for the target language leads to exceptional sample efficiency in downstream few-shot transfer. (2/2)
3
Chia-Chien Hung retweeted
Philipp Mayr (@Philipp_Mayr) profile picture
Philipp Mayr@Philipp_Mayr10 Jun 2022
The training data of our shared task "SV-Ident 2022 -- Survey Variable Identification in Social Science Publications" @SDProc @coling2022 is released: vadis-project.github.io/sv-i… Get the data: github.com/vadis-project/sv-… #coling2022 @spponzius contact: [email protected]
This Post is from an account that no longer exists. Learn more
5
3
(2/2) Our works @#acl2022🛸 We present an efficient framework for specialising task-oriented dialog systems to domains -- with only 50% training data we reach the performance of a non-specialised model! #findings @spponzius @gg42554 @dwsunima @MilaNLProc arxiv.org/abs/2110.08395
Image from Anne Lauscher (she/her) (@anne_lauscher)
5
4
1
19
Load more