Transformer-Lite: High-efficiency Deployment of Large Language Models on Mobile Phone GPUs

TL;DR

Transformer-Lite introduces four optimization techniques to enable high-efficiency deployment of large language models on mobile phone GPUs, significantly improving inference speed and user experience.

Contribution

The paper presents a novel mobile inference engine, Transformer-Lite, with four key optimizations for fast LLM inference on device GPUs, compatible with Qualcomm and MTK processors.

Findings

Achieved 10x speedup over CPU-based FastLLM.

Attained 2-3x faster decoding speeds compared to GPU-based MLC-LLM.

Supported LLMs from 2B to 14B parameters with high efficiency.

Abstract

The Large Language Model (LLM) is widely employed for tasks such as intelligent assistants, text summarization, translation, and multi-modality on mobile phones. However, the current methods for on-device LLM deployment maintain slow inference speed, which causes poor user experience. To facilitate high-efficiency LLM deployment on device GPUs, we propose four optimization techniques: (a) a symbolic expression-based approach to support dynamic shape model inference; (b) operator optimizations and execution priority setting to enhance inference speed and reduce phone lagging; (c) an FP4 quantization method termed M0E4 to reduce dequantization overhead; (d) a sub-tensor-based technique to eliminate the need for copying KV cache after LLM inference. Furthermore, we implement these methods in our mobile inference engine, Transformer-Lite, which is compatible with both Qualcomm and MTK processors. We evaluated Transformer-Lite's performance using LLMs with varied architectures and parameters ranging from 2B to 14B. Specifically, we achieved prefill and decoding speeds of 121 token/s and 14 token/s for ChatGLM2 6B, and 330 token/s and 30 token/s for smaller Gemma 2B, respectively. Compared with CPU-based FastLLM and GPU-based MLC-LLM, our engine attains over 10x speedup for the prefill speed and 2~3x speedup for the decoding speed.