[01] Mechanistic Analysis of Alignment Algorithms in Language Models
Aarush Sinha, Ishan Garg, Veeraraju Elluru, Arth Singh, Kushal Garg
ls ./research
Research
Directions
- 01 / internals
- Reverse-engineering learned behaviors through circuits and representations.
- 02 / alignment
- Understanding and mitigating misalignment.
- 03 / safety
- Reducing anthropomorphic risks in machines.
Publications
[02] Mechanistically Interpreting Compression in Vision-Language Models
Veeraraju Elluru, Roberto Aguero, Arth Singh, Ajay Agarwal, Debojyoti Das, Hreetam Paul
[03] Lie Algebra-Based Semantic Flow for Incompleteness Detection
Manikandan Ravikiran, Veeraraju Elluru, Karrtik Iyer, Prasanna Pendse, Shayan Mohanty
[04] Bias-Aware Machine Unlearning
Sai Siddhartha Chary Aylapuram, Veeraraju Elluru, Shivang Agarwal
[05] Unsupervised Latent Representations for Cattle Image Segmentation
Veeraraju Elluru, Ahmed Sallam, Anderson Alves, Tiago Bresolin