Skip to content

LeMei/CausalNlp-ReadingGroup

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

74 Commits
 
 
 
 
 
 

Repository files navigation

Six groups of causality in LLM and VLM

  1. Evaluation of causal ability of LLM and VLM by prompt (in-context learning, ….), factual knowledge

  2. Improve the model/framework performance on causal task (four levels: causality discovery, association, intervention, counteractuals)

  3. Spurious relation elimination between features and prediction by causality inference on downstream tasks (mainly interventions, Do(), backdoor and frontdoor adjustment), aiming to improve model performance with consideration of causal inference

  4. Domain adaptation of task with causal learning (related to #3) and the relationship between causality and generalization (exist or not) and why (probing task)

  5. Bias elimination from Dataset or modality by causal inference

  6. Metrics and benchmark for causal ability evaluation

image image

Probing work, assessment in LLM

Datasets

Application in NLP task

Sentiment analysis

Fact Checking

Named Entity Recognition

Reasoning

Sequence Modeling

Image Caption

Reinforcement Learning

image

Causal inference and LLM

About

repository for Causal&NLP works

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published