Bénédicte Colnet brilliantly defended her PhD on generalization of causal effects across populations on June 29th 2023. The question she tackled is the following: to empirically evaluate a causal effect, randomized controlled trials (RCTs) give excellent control of confounding effects (internal validity), but they often are performed on a population…
First paper with the soda affiliation: Imputing out-of-vocabulary embeddings with LOVE makes language models robust with little cost (https://arxiv.org/abs/2203.07860) Making language models robust to unknown words (eg typos): a bit of contrastive learning can extend language models without retraining them! The idea of LOVE (Learning Out-of-Vocabulary Embeddings) is to map unknown…
The Soda team, with its research project, is born. We have been working on the project, and assembling the team for a while, so we are really excited to be going live!
