11 citations · 17 across the 5 of their papers we have counts for
5 papers
If You Can't Use Them, Recycle Them: Optimizing Merging at Scale Mitigates Performance Tradeoffs
Muhammad Khalifa, Yi-Chern Tan, Arash Ahmadian +6
Model merging has shown great promise at combining expert models, but the benefit of merging is unclear when merging "generalist" models trained on many tasks. We explore merging i…
The Multilingual Alignment Prism: Aligning Global and Local Preferences to Reduce Harm
Aakanksha, Arash Ahmadian, Beyza Ermis +4
A key concern with the concept of "alignment" is the implicit question of "alignment to what?". AI systems are increasingly used across the world, yet safety alignment is often foc…
Back to Basics: Revisiting REINFORCE Style Optimization for Learning from Human Feedback in LLMs
Arash Ahmadian, Chris Cremer, Matthias Gallé +5
AI alignment in the shape of Reinforcement Learning from Human Feedback (RLHF) is increasingly treated as a crucial ingredient for high performance large language models. Proximal…
Pushing Mixture of Experts to the Limit: Extremely Parameter Efficient MoE for Instruction Tuning
Ted Zadouri, Ahmet Üstün, Arash Ahmadian +3
The Mixture of Experts (MoE) is a widely known neural architecture where an ensemble of specialized sub-models optimizes overall performance with a constant computational cost. How…
Intriguing Properties of Quantization at Scale
Arash Ahmadian, Saurabh Dash, Hongyu Chen +5
Emergent properties have been widely adopted as a term to describe behavior not present in smaller models but observed in larger models. Recent work suggests that the trade-off inc…