Probing results can be useful for model developments.
This paper probes the behaviors of DNNs in generalizations. We find that the information of out-of-domain generalization encodes more “linearly” on the intermediate layers than other layers for VLCS and PACS, but not so for RotatedMNIST.
Introducing additional labels can help OOD generalization performance.
This is the course project report for Speech Language Processing in 2022 Winter.
This paper uses learning theory to recommend suitable dataset sizes to reliably reproduce probing findings.
This paper adopts psycholinguistic studies on neural language models to analyze where the meanings are carried.
Here are some books I found interesting in 2021.
- Capital and Ideology (Piketty)
- Keep Sharp: Build a Better Brain at Any Age (Gupta)
- The Nature of Expertise (Chi etal)
In this paper, we estimate the task-specific information in text-based classification datasets.
In this paper, we propose an unsupervised framework that infers the source text with prominent influence on the moral time course.
This paper studies the writing features and their indications towards conference / workshop appearances in AI venues.