ETHZ++ NLP Reading Group

Time: Friday 10:30-12:00 Hybrid

Reading List 2023

Date Presenter Title Slides
August 30 Lukas The Prosody Playbook
August 2 Theodor Acquiring Linguistic Knowledge from Multimodal Input (Thesis)
July 19 Luca Divergence functions for NLG (Thesis)
June 30 (@14:30) Clemente Towards Variable Free Semantics link
June 14 Clemente On the Intersection of Context-Free and Regular Languages
June 14 Everyone EMNLP Abstract Swap
June 7 Everyone Research introductions practice
May 24 Tom The Evolution of Automatic Metrics, From String Matching to Black-Box LLM link
May 17 Andreas A hitchhiker’s Guide to Earley’s Algorithm
May 10 Clara Modeling Uncertainty in Neural Networks link
May 3 Mario On the Alignment of Production Variability in Humans and Text Generator link
April 26 Alex RLHF and GPT
April 19 Kevin Concepts and Interpretability
April 5 Mitja Communicative Feedback in Language Acquisition link
March 29 Ethan & Nik Discussions on GPT-4
March 22 Ethan How to Write Reviews and Responses
March 15 Nik The Ordered Matrix Dirichlet for State-Space Models link
March 8 Tianyu Taggers: Modeling Text Structures in Linear Time and Space
March 1 Leonie Everything is a Construction: New Goals for Syntactic and Semantic Probing link
February 24 Eleftheria Learning Persona Representations of Film Characters
February 15 Zhijing & Yuxin Aligning the Representation Spaces of MultiBERTS
February 8 Anej
Vilém
RNNs and Weighted Finite-State Automata
Poor Man’s Quality Estimation: Predicting Reference-Based MT Metrics Without the Reference

link
February 1 Everyone Research update

Reading List 2022

Date Presenter Topic Title Slides
November 4 Alex Pragmatics / RSA Pragmatic Language Interpretation as Probabilistic Inference
Pragmatically Informative Text Generation
October 28 Eleanor / Eric Malmi Generation / Decoding Text Generation with Text-Editing Models
October 21 Afra Generation / Decoding Diffusion-LM Improves Controllable Text Generation
October 14 Clara Generation / Decoding Structured Denoising Diffusion Models in Discrete State-Spaces
October 7 Vésteinn Generation / Decoding CoNT: Contrastive Neural Text Generation