8 papers
Same Question, Different Answers: Evaluating LLM Reliability Beyond Accuracy
Kazem Faghih, Yize Cheng, Shoumik Saha +3
Large language models (LLMs) often achieve strong accuracy on benchmarks, yet it remains unclear how reliably they apply this knowledge when the same question is phrased in differe…
On The Application of Linear Attention in Multimodal Transformers
Armin Gerami, Seyedehanita Madani, Ramani Duraiswami
Multimodal Transformers serve as the backbone for state-of-the-art vision-language models, yet their quadratic attention complexity remains a critical barrier to scalability. In th…
Auditing Algorithmic Bias in Transformer-Based Trading
Armin Gerami, Ramani Duraiswami
Transformer models have become increasingly popular in financial applications, yet their potential risk making and biases remain under-explored. The purpose of this work is to audi…
Quantifying Document Impact in RAG-LLMs
Armin Gerami, Kazem Faghih, Ramani Duraiswami
Retrieval Augmented Generation (RAG) enhances Large Language Models (LLMs) by connecting them to external knowledge, improving accuracy and reducing outdated information. However,…
Transformer Based Linear Attention with Optimized GPU Kernel Implementation
Armin Gerami, Ramani Duraiswami
The original softmax-based attention mechanism (regular attention) in the extremely successful Transformer architecture computes attention between tokens, each embedded in a $D…
A Scalable MVDR Beamforming Algorithm That is Linear in the Number of Antennas
Sanjaya Herath, Armin Gerami, Kevin Wagner +2
The Minimum Variance Distortionless Response (MVDR) beamforming technique is widely applied in array systems to mitigate interference. However, applying MVDR to large arrays is com…