4 papers · 1 filter
CODA: Rewriting Transformer Blocks as GEMM-Epilogue Programs
Han Guo, Jack Zhang, Arjun Menon +4
Transformer training systems are built around dense linear algebra, yet a nontrivial fraction of end-to-end time is spent on surrounding memory-bound operators. Normalization, acti…
Speculative Speculative Decoding
Tanishq Kumar, Tri Dao, Avner May
Autoregressive decoding is bottlenecked by its sequential nature. Speculative decoding has become a standard way to accelerate inference by using a fast draft model to predict upco…
BitDelta: Your Fine-Tune May Only Be Worth One Bit
James Liu, Guangxuan Xiao, Kai Li +4
Large Language Models (LLMs) are typically trained in two phases: pre-training on large internet-scale datasets, and fine-tuning for downstream tasks. Given the higher computationa…
Medusa: Simple LLM Inference Acceleration Framework with Multiple Decoding Heads
Tianle Cai, Yuhong Li, Zhengyang Geng +4
Large Language Models (LLMs) employ auto-regressive decoding that requires sequential computation, with each step reliant on the previous one's output. This creates a bottleneck as…