Upcoming

2023

• Nov 10, 2023, 9:30am PST Invited talk: I will present RA-DIT at Cohere for AI Interactive Reading Group.
• Nov 3, 2023, 9-10am PST Invited talk: I will present RA-DIT at LLamaIdex Webinar.

Past

2023

• Oct 2, 2023 Exited to introduce "RA-DIT: Retrieval-Augmented Dual Instruction Tuning", a lightweight finetuning framework to efficiently and effectively retrofit any pre-trained LLM with retrieval capabilities.
• Jul 28, 2023 Panel Discussion at ICML 2023 Knowledge and Logical Reasoning (KLR) Workshop: Reasoning Capabilities of LLMs. Video
• Jul 28, 2023 Invited talk at ICML 2023 Knowledge and Logical Reasoning (KLR) Workshop: Knowledge and Skill Acquisition through LLM Pre-training and Instruction-tuning.
• May 25, 2023 Invited talk at the Stanford NLP Seminar: Large Language Models as Instructable Task Solvers: Lessons Learned and Future Possibilities. Slides

══ ⋆ I wandered off for a while and paused updating my announcement page. Coming back. ⋆ ══

2022

• Mar 28, 2022 Paper "Pretty Princess vs. Successful Leader: Gender Roles in Greeting Card Messages" (led by Jiao Sun) won the Best Paper Honorable Mention at CHI 2022.
• Feb 15, 2022 The XGLM models are now available in the Huggingface Transformers library, together with a demo of the 2.9B model on the XCOPA task.

2021

• Dec 21, 2021 Our team released the preprint "Few-shot Learning with Multilingual Language Models", together with a family of multilingual auto-regressive language models of up to 7.5B parameters, establishing new state-of-the-art in-context learning performance on language completion, NLI and machine translation over more than 20 languages .
• Dec 21, 2021 Our team released the preprint "Efficient Large Scale Language Modeling with Mixtures of Experts", together with a family of dense and Mixture-of-Expert (MoE) language models of up to 13B and 1.1T parameters.
• Nov 15, 2021 Paper "Pretty Princess vs. Successful Leader: Gender Roles in Greeting Card Messages" (led by Jiao Sun) is accepted to CHI 2022.
• Nov 11, 2021 Our paper "Testing Cross-Database Semantic Parsers Using Canonical Utterances" (led by Heather Lent during her internship at Salesforce Research) won the Best Paper Award at the Eval4NLP workshop at EMNLP 2021.
• Oct 21, 2021 Our team at Facebook AI Research is hiring research interns, please reach out if you are currently pursuing a PhD and is interested in applying.
• May 11, 2021 Our report on the NLC2CMD Challenge will appear in the PMLR post proceedings volume associated to the Competition Track @ NeurIPS2020.
• Apr 1, 2021 We released FetaQA (led by Linyong Nan), a free-form table question answering task complementary to SQL-based information seeking.
• Sep 05, 2021 Our paper "Testing Cross-Database Semantic Parsers Using Canonical Utterances" (led by Heather Lent during her internship at Salesforce Research) is accepted to the Eval4NLP workshop at EMNLP 2021.
• May 11, 2021 Our report on the NLC2CMD Challenge will appear in the PMLR post proceedings volume associated to the Competition Track @ NeurIPS2020.
• Apr 1, 2021 We released FetaQA (led by Linyong Nan), a free-form table question answering task complementary to SQL-based information seeking.
• Mar 10, 2021 Paper "Learning to Synthesize Data for Semantic Parsing" (led by Bailin Wang during his internship at Salesforce Research), is accepted to NAACL 2021.
• Mar 10, 2021 Paper "DART: Open-Domain Structured Data Record to Text Generation" (led by Linyong Nan), is accepted to NAACL 2021.
• Jan 12, 2021 Paper "GraPPa: Grammar-Augmented Pre-Training for Table Semantic Parsing" (led by Tao Yu), is accepted to ICLR 2021.
• Jan. 11, 2021 I'm joining Facebook AI, Menlo Park, CA as a research scientist.
• Jan. 6, 2021 We released Photon 2.0 at naturalsql.com.

2020

• Dec. 23, 2020 We released the code and an extended verision of our EMNLP Findings 2020 paper "Bridging Textual and Tabular Data for Cross-Domain Text-to-SQL Semantic Parsing".
• Dec. 12, 2020, 11:05-11:15am PST Invited talk: Tellina: A Deep Learning Powered End User Scripting Assistant. NLC2CMD challenge, NeurIPS 2020. Slides
• Dec. 11, 2020, 4:35-5:15pm PST Panel Discussion: Where and how can KRR benefit ML, and what should be explored? KR2ML Workshop, NeurIPS 2020. Video
• Dec. 11, 2020, 9:35-10:05am PST Invited talk: Bridging Textual and Tabular Representation: Is Attention All We Need? Workshop on Knowledge Representation & Reasoning Meets Machine Learning (KR2ML), NeurIPS 2020.
• Dec. 02, 2020 Invited talk: Reinforcement Learning for Knowledge Graph Reasoning. Knowledge Connexions Conference 2020. Video Slides
• Dec. 01, 2020 Guest lecture: Natural Language Interface to Databases for course Natural Language Processing, NYU. Thanks He He for hosting. Slides
• Nov. 19, 2020 Three papers presented at Intex-Sempar Workshop, EMNLP 2020.
• Nov. 16-Nov. 20, 2020 Virtually attending EMNLP 2020.
• Nov. 5, 2020 Great fun serving on the Salesforce Research Women in AI Panel Discussion with my fantastic female colleagues.
• Oct. 2, 2020 We released the preprint of GraPPa (led by Tao Yu), a pre-trained joint representation for text and tables with injected compositional inductive bias.
• Sep. 30, 2020 I'm serving as a system demonstration co-chair for NAACL 2021.
• Sep. 29, 2020 Paper "ColloQL: Robust Cross-Domain Text-to-SQL Over Search Queries" (led by Karthik Radhakrishnan) is accepted to INT-EX: Interactive and Executable Semantic Parsing at EMNLP 2020.
• Sep. 14, 2020 Our paper "Bridging Textual and Tabular Data for Cross-Domain Text-to-SQL Semantic Parsing." is accepted to Findings of EMNLP 2020.
• Aug 26, 2020 Our blog post, Talk to Your Data: One Model, Any Relational Database, is in the Hacker News Front Page.
• July 06, 2020 Excited to release the live demo of Photon, a deep-learning powered cross-domain natural language interface to data systems. Please join our ACL 2020 live Q&A sessions on July 7 at 10-11 PDT and 13-14 PDT.
• July 01, 2020 The CLAI team from IBM Research is organizing an exciting NLC2CMD Challenge at NeurIPS 2020. Tellina serves as the baseline system of the challenge and NL2Bash is one of the designated training sets.
• Jun 22, 2020 First call of paper is out for our EMNLP 2020 workshop, INT-EX: Interactive and Executable Semantic Parsing. Please consider submitting your work here.
• Jun 17, 2020 Paper "Double-Hard Debias: Tailoring Word Embeddings for Gender Bias Mitigation" is highlighted in The Batch by deeplearning.ai.
• Apr 3, 2020 System demonstration paper "Photon: A Robust Cross-Domain Text-to-SQL System" (co-led with our intern Jichuan Zeng) is accepted to ACL 2020.
• Apr 3, 2020 Paper "Double-Hard Debias: Tailoring Word Embeddings for Gender Bias Mitigation" (led by our intern Tianlu Wang) is accepted to ACL 2020.
• Feb 19, 2020 (Wed) 4:00pm-5:00pm Talk at Gates Commons, Paul G. Allen School of Computer Science & Engineering, University of Washington, The Age of Data Conversation: Talking to Your Relational Data.
• Feb. 18-20, 2020 Salesforce Research outreach at the University of Washington, Seattle.

2019

• Dec. 10-14, 2019 Traveling to NeurIPS 2019, Vancouver, BC, Canada.
• Nov 21, 2019 ZDNet covered our Dreamforce AI Research Keynote in a thorough article: "Salesforce teases its emerging AI capabilities".
• Nov 19, 2019 4:30pm-5:30pm I co-delivered the AI Resesarch Keynote at Dreamforce 2019: "The End of Data Entry and The Beginning of Data Conversations". Live Stream Video
•  Oct 30, 2019 Attending First Open Virtual Assistant Workshop hosted by Stanford OVAL.
• Oct 25, 2019 Our workshop proposal on INT-EX: Interactive and Executable Semantic Parsing is accepted as a one-day event at EMNLP 2020! Paper calls and other updates coming soon.
• Aug 12, 2019 Paper "CoSQL: A Conversational Text-to-SQL Challenge Towards Cross-Domain Natural Language Interfaces to Databases" (led by Tao Yu) is accepted to EMNLP 2019. Joint work with LILY and CroMa.
• Aug 12, 2019 Paper "Editing-based SQL Query Generation for Cross-Domain Context-Dependent Questions" (led by Rui Zhang) is accepted to EMNLP 2019. Joint work with LILY.
• May 13, 2019 Paper "SParC: Cross-Domain Semantic Parsing in Context" (led by Tao Yu) is accepted to ACL 2019. Joint collaboration with LILY, Yale's NLP lab.
• May 7, 2019 10:30am I'm giving a lightning tech talk -- "Towards Explainable and Scalable Knowledge Graph Inference" at the Salesforce Research booth in ICLR.
• May 5-9, 2019 Traveling to ICLR 2019, New Orleans, LA.
• Mar. 18, 2019 ZDNet covered our EMNLP'18 paper: "Salesforce Research: Knowledge graphs and machine learning to power Einstein".

2018

• Nov. 3, 2018 (Sat) Presenting "Multi-Hop Knowledge Graph Reasoning with Reward Shaping" (long talk) at EMNLP. Video Slides
• Oct. 27-Nov.5, 2018 Traveling to EMNLP 2018, Brussels, Belgium.
• Sep. 25-28, 2018 Traveling to Grace Hopper Conference (GHC) 2018, Houston, TX.
• Aug. 27, 2018 NL2Bash is in the Hacker News Front Page!
• Aug. 10, 2018 Our paper "Multi-Hop Knowledge Graph Reasoning with Reward Shaping" is accepted to EMNLP 2018. See you in Brussels!
• Jul. 31, 2018 Invited to Mining Scientific Knowledge Workshop: Extraction, Curation and Computation hosted by Chan Zuckerberg Initialtive.
• May 10, 2018 (Thu) NL2Bash will appear in the poster+demo session of LREC 2018, Miyazaki, Japan.
• Feb. 2, 2018 Presenting NL2Bash (long talk) in NL4SE Workshop, co-located with AAAI 2018.  Slides

2017

• Dec. 13, 2017 Our dataset paper "NL2Bash: A Corpus and Semantic Parser for Natural Language Interface to the Linux Operating System" is accepted to LREC 2018!
• Dec. 4-10, 2017 Traveling to NIPS 2017, Long Beach, CA.
• Oct. 16, 2017 Excited to join Salesforce AI Research (previously MetaMind) as a researcher on deep learning and natural language processing.
• Mar. 1, 2017 We released a technical report about Tellina, the natural language to command line translation tool we've been developing!