Dear Corpora list,
I am looking for applicants for two fully funded PhD/postdoc positions for 3 years starting ~September 2026.
Project description: LMLM proposes to build Linguistically Motivated Language Models (LMs). This involves adapting both the training regime and the LM architecture to incorporate different linguistic levels. The hypothesis is that this will lead to mitigation of shortcut learning and, in turn, more reliable LMs. The project will start on the lower levels, investigating how internal structures of words can be used in LMs, followed by words and the syntax level. There will be a strong focus on evaluation, incorporating both standard metrics and benchmarks as well as linguistically oriented and compositionality benchmarks. The successful candidate(s) will be part of the NLPnorth research group in the Data Science section of ITU, and are also encouraged to become affiliated with the AI Pioneer center. They will be supervised by me and a co-supervisor which is selected on a per-candidate basis. LMLM is funded by Carlsberg (https://www.carlsbergfondet.dk/en/what-we-have-funded/cf25-1861/).
More information is available on: https://candidate.hr-manager.net/ApplicationInit.aspx?cid=119&ProjectId=... , and feel free to contact me with any questions.
Best, Rob (https://robvanderg.github.io/)