Bio (CV)

I am a research scientist at Facebook AI Research (FAIR). Previously, I was an instructor at Princeton and the IAS. In 2017, I completed a PhD in computer science at Stanford where I worked on machine learning and natural language processing co-advised by Chris Manning and Percy Liang. Before that I got a BASc at University of Toronto and worked on capsules with Geoffrey Hinton.

Contact: sidawxyz [at] gmail.com

Recently I worked on Code LLM: CWM, AST-FIM, SWE-RL, eval-arena, LEVER, Coder-reviewer, InCoder, MBR-exec.

and benchmarks: SWE-bench M, Spider 2.0, LiveCodeBench, SAFIM, CRUXEval, DS-1000.

Selected work

• Structure-Aware Fill-in-the-Middle Pretraining for Code
• Accessing higher dimensions does unsupervised translation with < 100MB of data + 90s method
• Naturalizing a programming language via interactive learning
• Data Noising as Smoothing in Neural Network Language Models
• Learning language games through interaction
• Estimating mixture models via mixtures of polynomials
• Fast dropout training and dropout as adapative regularization
• Baselines and bigrams (NBSVM) showed the power of bag of bigrams, adopted by fastText
• Capsules / transforming autoencoders

Publications

• Google Scholar
• List of old publications

Services

• Area chair for Neurips, ICLR, ICML

• Reviewer for TACL, ARR, Neurips, ICLR, EMNLP, JMLR, ICML

• Spring 2018: COS495 Natural language processing
• Fall 2015: Head TA for CS224 Natural language processing
• Winter 2013: TA for CS229T Statistical machine learning