8000 GitHub - erfanloghmani/DeconfoundLM: Official repository of the paper "Aligning Language Models with Observational Data: Opportunities and Risks from a Causal Perspective". The project explores methods to fine-tune large language models (LLMs) using observational data, tackling challenges like spurious correlations and confounding.
[go: up one dir, main page]
More Web Proxy on the site http://driver.im/
Skip to content

Official repository of the paper "Aligning Language Models with Observational Data: Opportunities and Risks from a Causal Perspective". The project explores methods to fine-tune large language models (LLMs) using observational data, tackling challenges like spurious correlations and confounding.

Notifications You must be signed in to change notification settings

erfanloghmani/DeconfoundLM

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

2 Commits
 
 

Repository files navigation

Aligning Language Models with Observational Data

Official repository of the paper "Aligning Language Models with Observational Data: Opportunities and Risks from a Causal Perspective". The project explores methods to fine-tune large language models (LLMs) using observational data, tackling challenges like spurious correlations and confounding. We propose DECONFOUNDLM, a novel approach to mitigate these issues by correcting for known confounders in the reward signals.

Code to be released soon.

You can find the project webpage at deconfoundlm.github.io.

About

Official repository of the paper "Aligning Language Models with Observational Data: Opportunities and Risks from a Causal Perspective". The project explores methods to fine-tune large language models (LLMs) using observational data, tackling challenges like spurious correlations and confounding.

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published
0