Posts by Collection


Semantic Code Search

For the course Machine Learning for Programming at Cambridge, I proposed a neural re-ranking model to improve results of a robust baseline for semantic code search, a retrieval task where a code snippet must be found given a particular natural language query.

Re-ranking for Machine Reading

For the Machine Learning for Language Processing course at Cambridge, I explored some extensions to our (then under-review) ACL 2020 paper “Machine Reading of Historical Events.” Apart from an empirical study of different attention mechanisms, I also explored how to cast this as a ranking problem. While improvements were modest in the setting we cared about, I found that this approach gave substantial improvements in a setting were we have a chronological ordering of events, but year annotations only for some of them.

Reinforcement Learning

For the Reinforcement Learning course at Edinburgh, I applied different RL techniques to a particular robot football scenario, spanning a range of settings (discrete vs continuous states, single vs multi-agent).

Compilers and Emulators

For the Compiling Techniques course at Edinburgh, I built a C to MIPS compiler using Java. Somewhat related, I also implemented a simplified MIPS emulator.


Machine Reading of Historical Events Paper Code

by Or Honovich*, Lucas Torroba Hennigen*, Omri Abend, Shay B. Cohen

in Proceedings of the 58th Annual Meeting of the Association for Computational Linguistics, July 2020

We introduce the historical event ordering (HEO) task, where a series of short textual descriptions of historical events, potentially alongside some additional information, are ordered chronologically. We compile two datasets for this task, and compare the performance of two models in it.

Intrinsic Probing Through Dimension Selection Paper Code

by Lucas Torroba Hennigen, Adina Williams, Ryan Cotterell

in Proceedings of the 2020 Conference on Empirical Methods in Natural Language Processing, November 2020

We introduce a novel framework for intrinsic probing that leverages a decomposable multivariate Gaussian probe. We run experiments on 36 languages from the Universal Dependencies treebanks, and find that fastText concentrates its linguistic structure more than BERT.



Teaching experience 1

Undergraduate course, University 1, Department, 2014

This is a description of a teaching experience. You can use markdown like any other post.

Teaching experience 2

Workshop, University 1, Department, 2015

This is a description of a teaching experience. You can use markdown like any other post.