FiRST: Finetuning Router-Selective Transformers for Input-Adaptive Latency Reduction

• URL: http://arxiv.org/abs/2410.12513v1
• Date: Wed, 16 Oct 2024 12:45:35 GMT
• Title: FiRST: Finetuning Router-Selective Transformers for Input-Adaptive Latency Reduction
• Authors: Akriti Jain, Saransh Sharma, Koyel Mukherjee, Soumyabrata Pal,
• Abstract summary: FIRST is an algorithm that reduces inference latency by using layer-specific routers to select a subset of transformer layers adaptively for each input sequence. Our approach reveals that input adaptivity is critical - indeed, different task-specific middle layers play a crucial role in evolving hidden representations depending on task.
• Score: 11.146015814220858
• License:
• Abstract: Auto-regressive Large Language Models (LLMs) demonstrate remarkable performance across domanins such as vision and language processing. However, due to sequential processing through a stack of transformer layers, autoregressive decoding faces significant computation/latency challenges, particularly in resource constrained environments like mobile and edge devices. Existing approaches in literature that aim to improve latency via skipping layers have two distinct flavors - 1) Early exit 2) Input-agnostic heuristics where tokens exit at pre-determined layers irrespective of input sequence. Both the above strategies have limitations - the former cannot be applied to handle KV Caching necessary for speed-ups in modern framework and the latter does not capture the variation in layer importance across tasks or more generally, across input sequences. To address both limitations, we propose FIRST, an algorithm that reduces inference latency by using layer-specific routers to select a subset of transformer layers adaptively for each input sequence - the prompt (during prefill stage) decides which layers will be skipped during decoding. FIRST preserves compatibility with KV caching enabling faster inference while being quality-aware. FIRST is model-agnostic and can be easily enabled on any pre-trained LLM. We further improve performance by incorporating LoRA adapters for fine-tuning on external datasets, enhancing task-specific accuracy while maintaining latency benefits. Our approach reveals that input adaptivity is critical - indeed, different task-specific middle layers play a crucial role in evolving hidden representations depending on task. Extensive experiments show that FIRST significantly reduces latency while retaining competitive performance (as compared to baselines), making our approach an efficient solution for LLM deployment in low-resource environments.

Related papers

• Progressive Mixed-Precision Decoding for Efficient LLM Inference [49.05448842542558]
  We introduce Progressive Mixed-Precision Decoding (PMPD) to address the memory-boundedness of decoding. PMPD achieves 1.4$-$12.2$times$ speedup in matrix-vector multiplications over fp16 models. Our approach delivers a throughput gain of 3.8$-$8.0$times$ over fp16 models and up to 1.54$times$ over uniform quantization approaches.
  arXiv  Detail & Related papers  (2024-10-17T11:46:33Z)
• TidalDecode: Fast and Accurate LLM Decoding with Position Persistent Sparse Attention [7.4088392854630625]
  Large language models (LLMs) have driven significant advancements across diverse NLP tasks. This paper introduces TidalDecode, a system for fast and accurate LLM decoding through position persistent sparse attention.
  arXiv  Detail & Related papers  (2024-10-07T14:30:27Z)
• Dual-Path Adversarial Lifting for Domain Shift Correction in Online Test-time Adaptation [59.18151483767509]
  We introduce a dual-path token lifting for domain shift correction in test time adaptation. We then perform dual-path lifting with interleaved token prediction and update between the path of domain shift tokens and the path of class tokens. Experimental results on the benchmark datasets demonstrate that our proposed method significantly improves the online fully test-time domain adaptation performance.
  arXiv  Detail & Related papers  (2024-08-26T02:33:47Z)
• SHERL: Synthesizing High Accuracy and Efficient Memory for Resource-Limited Transfer Learning [63.93193829913252]
  We propose an innovative METL strategy called SHERL for resource-limited scenarios. In the early route, intermediate outputs are consolidated via an anti-redundancy operation. In the late route, utilizing minimal late pre-trained layers could alleviate the peak demand on memory overhead.
  arXiv  Detail & Related papers  (2024-07-10T10:22:35Z)
• Optimizing a Transformer-based network for a deep learning seismic processing workflow [0.0]
  StorSeismic is a recently introduced model based on the Transformer to adapt to various seismic processing tasks. We observe faster pretraining and competitive results on the fine-tuning tasks and, additionally, fewer parameters to train compared to the vanilla model.
  arXiv  Detail & Related papers  (2023-08-09T07:11:42Z)
• Dynamic Perceiver for Efficient Visual Recognition [87.08210214417309]
  We propose Dynamic Perceiver (Dyn-Perceiver) to decouple the feature extraction procedure and the early classification task. A feature branch serves to extract image features, while a classification branch processes a latent code assigned for classification tasks. Early exits are placed exclusively within the classification branch, thus eliminating the need for linear separability in low-level features.
  arXiv  Detail & Related papers  (2023-06-20T03:00:22Z)
• LAIT: Efficient Multi-Segment Encoding in Transformers with Layer-Adjustable Interaction [31.895986544484206]
  We introduce Layer- Interactions in Transformers (LAIT) Within LAIT, segmented inputs are first encoded independently, and then jointly. We find LAIT able to reduce 30-50% of the attention FLOPs on many tasks, while preserving high accuracy.
  arXiv  Detail & Related papers  (2023-05-31T06:09:59Z)
• Latency Adjustable Transformer Encoder for Language Understanding [0.8287206589886879]
  This paper proposes an efficient Transformer architecture that adjusts the inference computational cost adaptively with a desired inference latency speedup. The proposed method detects less important hidden sequence elements (word-vectors) and eliminates them in each encoder layer using a proposed Attention Context Contribution (ACC) metric. The proposed method mathematically and experimentally improves the inference latency of BERT_base and GPT-2 by up to 4.8 and 3.72 times with less than 0.75% accuracy drop and passable perplexity on average.
  arXiv  Detail & Related papers  (2022-01-10T13:04:39Z)
• Video Super-Resolution Transformer [85.11270760456826]
  Video super-resolution (VSR), with the aim to restore a high-resolution video from its corresponding low-resolution version, is a spatial-temporal sequence prediction problem. Recently, Transformer has been gaining popularity due to its parallel computing ability for sequence-to-sequence modeling. In this paper, we present a spatial-temporal convolutional self-attention layer with a theoretical understanding to exploit the locality information.
  arXiv  Detail & Related papers  (2021-06-12T20:00:32Z)
• Cluster-Former: Clustering-based Sparse Transformer for Long-Range Dependency Encoding [90.77031668988661]
  Cluster-Former is a novel clustering-based sparse Transformer to perform attention across chunked sequences. The proposed framework is pivoted on two unique types of Transformer layer: Sliding-Window Layer and Cluster-Former Layer. Experiments show that Cluster-Former achieves state-of-the-art performance on several major QA benchmarks.
  arXiv  Detail & Related papers  (2020-09-13T22:09:30Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.