4 papers
TACO: Ternary Absolute-max Column-wise One-sparse Optimizer for LLM Fine-Tuning
Jichao Jiang, Cristian McGee, El Houcine Bergou +3
Full-parameter fine-tuning of large language models (LLMs) incurs substantial optimizer state memory overhead, limiting the model sizes that fit on modern GPUs. Existing approaches…
Trust the Direction, Search the Step: Zero-and-First-Order Methods for LLM Fine-Tuning
Cristian McGee, El Houcine Bergou, Aritra Dutta
Step-size selection remains a central challenge in large-scale neural network optimization; conservative steps slow convergence, while aggressive steps can destabilize it. We combi…
ART: Attention Run-time Termination for Efficient Large Language Model Decoding
Chen Qiu, Guozhong Li, Cristian McGee +2
Long-context decoding in Large Language Models (LLMs) is constrained by the cost of accessing and processing the Key-Value (KV) cache. Despite evidence that attention outputs depen…
Beyond LoRA: Is Sparsity-Induced Adaptation Better?
Elijah Cadenhead, Cristian McGee, Xin Li +2
Low-rank adaptation (LoRA) and its variants provide a memory- and compute-efficient alternative to full fine-tuning of pre-trained models. However, questions remain about the compa…