Amrit Singh Bedi
@amritsinghbedi.bsky.social
CS Faculty at UCF (SafeRR AI Lab), previous @UMD @ARL@IITK
Interested in Generative AI Alignment, RL, Optimization methods, Robotics, LLMs
Interested in Generative AI Alignment, RL, Optimization methods, Robotics, LLMs
A quick sneak peek into our approach and results:
December 10, 2024 at 1:35 PM
A quick sneak peek into our approach and results:
🔹Towards Eliminating Free-Riding in #Federated Learning via Truthful Mechanism Design (FACT)
arxiv.org/pdf/2405.13879
✅ No #Free #Riding
✅ Towards Guaranteed #Truthfulness
✅ Individually Rational
✅ Realistic Implementation
arxiv.org/pdf/2405.13879
✅ No #Free #Riding
✅ Towards Guaranteed #Truthfulness
✅ Individually Rational
✅ Realistic Implementation
December 8, 2024 at 2:47 PM
🔹Towards Eliminating Free-Riding in #Federated Learning via Truthful Mechanism Design (FACT)
arxiv.org/pdf/2405.13879
✅ No #Free #Riding
✅ Towards Guaranteed #Truthfulness
✅ Individually Rational
✅ Realistic Implementation
arxiv.org/pdf/2405.13879
✅ No #Free #Riding
✅ Towards Guaranteed #Truthfulness
✅ Individually Rational
✅ Realistic Implementation
🔹 A Principled Framework for Inference-Time Alignment of #LLMs
arxiv.org/abs/2405.20495
✅ No expensive #finetuning
✅ On-the-fly behavior adaptation
✅ Preserves #privacy of human preferences
✅ Potential to solve the #diversity alignment problem
arxiv.org/abs/2405.20495
✅ No expensive #finetuning
✅ On-the-fly behavior adaptation
✅ Preserves #privacy of human preferences
✅ Potential to solve the #diversity alignment problem
December 8, 2024 at 2:47 PM
🔹 A Principled Framework for Inference-Time Alignment of #LLMs
arxiv.org/abs/2405.20495
✅ No expensive #finetuning
✅ On-the-fly behavior adaptation
✅ Preserves #privacy of human preferences
✅ Potential to solve the #diversity alignment problem
arxiv.org/abs/2405.20495
✅ No expensive #finetuning
✅ On-the-fly behavior adaptation
✅ Preserves #privacy of human preferences
✅ Potential to solve the #diversity alignment problem