# TR-BERT: Dynamic Token Reduction for Accelerating BERT Inference

arXiv: https://arxiv.org/abs/2105.11618 · HTML: https://mlsanity.com/paper/2105.11618/

## Key facts

- Cited by 40 papers in the map (data through 14 Oct 2026).
- Placed in Early exit and edge inference, one of 12 directions in Efficient LLM inference and serving.
- Builds on 43 of its 59 references that are in the map.
- First posted on arXiv 25 May 2021.

## Placed in

- Efficient LLM inference and serving / [Early exit and edge inference](https://mlsanity.com/inference/early-exit-and-edge-inference/)

## Builds on (most cited first, 30 of 43)

- [Attention Is All You Need](https://mlsanity.com/paper/1706.03762/) (2017)
- [BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding](https://mlsanity.com/paper/1810.04805/) (2018)
- [Adam: A Method for Stochastic Optimization](https://mlsanity.com/paper/1412.6980/) (2014)
- [RoBERTa: A Robustly Optimized BERT Pretraining Approach](https://mlsanity.com/paper/1907.11692/) (2019)
- [Distilling the Knowledge in a Neural Network](https://mlsanity.com/paper/1503.02531/) (2015)
- [DistilBERT, a distilled version of BERT: smaller, faster, cheaper and lighter](https://mlsanity.com/paper/1910.01108/) (2019)
- [SQuAD: 100,000+ Questions for Machine Comprehension of Text](https://mlsanity.com/paper/1606.05250/) (2016)
- [HotpotQA: A Dataset for Diverse, Explainable Multi-hop Question Answering](https://mlsanity.com/paper/1809.09600/) (2018)
- [Longformer: The Long-Document Transformer](https://mlsanity.com/paper/2004.05150/) (2020)
- [TriviaQA: A Large Scale Distantly Supervised Challenge Dataset for Reading Comprehension](https://mlsanity.com/paper/1705.03551/) (2017)
- [XLNet: Generalized Autoregressive Pretraining for Language Understanding](https://mlsanity.com/paper/1906.08237/) (2019)
- [A Baseline for Detecting Misclassified and Out-of-Distribution Examples in Neural Networks](https://mlsanity.com/paper/1610.02136/) (2016)
- [Character-level Convolutional Networks for Text Classification](https://mlsanity.com/paper/1509.01626/) (2015)
- [Big Bird: Transformers for Longer Sequences](https://mlsanity.com/paper/2007.14062/) (2020)
- [Know What You Don't Know: Unanswerable Questions for SQuAD](https://mlsanity.com/paper/1806.03822/) (2018)
- [Generating Long Sequences with Sparse Transformers](https://mlsanity.com/paper/1904.10509/) (2019)
- [TinyBERT: Distilling BERT for Natural Language Understanding](https://mlsanity.com/paper/1909.10351/) (2019)
- [RACE: Large-scale ReAding Comprehension Dataset From Examinations](https://mlsanity.com/paper/1704.04683/) (2017)
- [Are Sixteen Heads Really Better than One?](https://mlsanity.com/paper/1905.10650/) (2019)
- [A Primer in BERTology: What we know about how BERT works](https://mlsanity.com/paper/2002.12327/) (2020)
- [Universal Transformers](https://mlsanity.com/paper/1807.03819/) (2018)
- [Patient Knowledge Distillation for BERT Model Compression](https://mlsanity.com/paper/1908.09355/) (2019)
- [MobileBERT: a Compact Task-Agnostic BERT for Resource-Limited Devices](https://mlsanity.com/paper/2004.02984/) (2020)
- [Reducing Transformer Depth on Demand with Structured Dropout](https://mlsanity.com/paper/1909.11556/) (2019)
- [Q-BERT: Hessian Based Ultra Low Precision Quantization of BERT](https://mlsanity.com/paper/1909.05840/) (2019)
- [DeeBERT: Dynamic Early Exiting for Accelerating BERT Inference](https://mlsanity.com/paper/2004.12993/) (2020)
- [NewsQA: A Machine Comprehension Dataset](https://mlsanity.com/paper/1611.09830/) (2016)
- [The Lottery Ticket Hypothesis for Pre-trained BERT Networks](https://mlsanity.com/paper/2007.12223/) (2020)
- [DynaBERT: Dynamic BERT with Adaptive Width and Depth](https://mlsanity.com/paper/2004.04037/) (2020)
- [FastBERT: a Self-distilling BERT with Adaptive Inference Time](https://mlsanity.com/paper/2004.02178/) (2020)

## Built on by (most cited first, 30 of 40)

- [LLM-Pruner: On the Structural Pruning of Large Language Models](https://mlsanity.com/paper/2305.11627/) (2023)
- [MoEfication: Transformer Feed-forward Layers are Mixtures of Experts](https://mlsanity.com/paper/2110.01786/) (2021)
- [Learned Token Pruning for Transformers](https://mlsanity.com/paper/2107.00910/) (2021)
- [Fit and Prune: Fast and Training-free Visual Token Pruning for Multi-modal Large Language Models](https://mlsanity.com/paper/2409.10197/) (2024)
- [Towards Efficient Generative Large Language Model Serving: A Survey from Algorithms to Systems](https://mlsanity.com/paper/2312.15234/) (2023)
- [Model Compression and Efficient Inference for Large Language Models: A Survey](https://mlsanity.com/paper/2402.09748/) (2024)
- [Zero-TPrune: Zero-Shot Token Pruning through Leveraging of the Attention Graph in Pre-Trained Transformers](https://mlsanity.com/paper/2305.17328/) (2023)
- [Token Reduction Should Go Beyond Efficiency in Generative Models -- From Vision, Language to Multimodality](https://mlsanity.com/paper/2505.18227/) (2025)
- [Packed Levitated Marker for Entity and Relation Extraction](https://mlsanity.com/paper/2109.06067/) (2021)
- [A Survey on Model Compression and Acceleration for Pretrained Language Models](https://mlsanity.com/paper/2202.07105/) (2022)
- [AIM: Adaptive Inference of Multi-Modal LLMs via Token Merging and Pruning](https://mlsanity.com/paper/2412.03248/) (2024)
- [When Large Vision-Language Model Meets Large Remote Sensing Imagery: Coarse-to-Fine Text-Guided Token Pruning](https://mlsanity.com/paper/2503.07588/) (2025)
- [A Survey on Dynamic Neural Networks for Natural Language Processing](https://mlsanity.com/paper/2202.07101/) (2022)
- [Transkimmer: Transformer Learns to Layer-wise Skim](https://mlsanity.com/paper/2205.07324/) (2022)
- [Efficiently Controlling Multiple Risks with Pareto Testing](https://mlsanity.com/paper/2210.07913/) (2022)
- [AdapLeR: Speeding up Inference by Adaptive Length Reduction](https://mlsanity.com/paper/2203.08991/) (2022)
- [Bridging Local Details and Global Context in Text-Attributed Graphs](https://mlsanity.com/paper/2406.12608/) (2024)
- [Pyramid-BERT: Reducing Complexity via Successive Core-set based Token Selection](https://mlsanity.com/paper/2203.14380/) (2022)
- [ConsistentEE: A Consistent and Hardness-Guided Early Exiting Method for Accelerating Language Models Inference](https://mlsanity.com/paper/2312.11882/) (2023)
- [Sparse Attention Acceleration with Synergistic In-Memory Pruning and On-Chip Recomputation](https://mlsanity.com/paper/2209.00606/) (2022)
- [Rethinking Chain-of-Thought Reasoning for Videos](https://mlsanity.com/paper/2512.09616/) (2025)
- [Accelerating Attention through Gradient-Based Learned Runtime Pruning](https://mlsanity.com/paper/2204.03227/) (2022)
- [Language Model Guided Interpretable Video Action Reasoning](https://mlsanity.com/paper/2404.01591/) (2024)
- [SmartTrim: Adaptive Tokens and Attention Pruning for Efficient Vision-Language Models](https://mlsanity.com/paper/2305.15033/) (2023)
- [Constraint-aware and Ranking-distilled Token Pruning for Efficient Transformer Inference](https://mlsanity.com/paper/2306.14393/) (2023)
- [CItruS: Chunked Instruction-aware State Eviction for Long Sequence Modeling](https://mlsanity.com/paper/2406.12018/) (2024)
- [AttMEMO : Accelerating Transformers with Memoization on Big Memory Systems](https://mlsanity.com/paper/2301.09262/) (2023)
- [SemToken: Semantic-Aware Tokenization for Efficient Long-Context Language Modeling](https://mlsanity.com/paper/2508.15190/) (2025)
- [SinkPruner: Sink-Free Visual Token Pruning for Multimodal Large Language Models](https://mlsanity.com/paper/2609.01004/) (2026)
- [Infor-Coef: Information Bottleneck-based Dynamic Token Downsampling for Compact and Efficient language model](https://mlsanity.com/paper/2305.12458/) (2023)

## Abstract (arXiv)

Existing pre-trained language models (PLMs) are often computationally expensive in inference, making them impractical in various resource-limited real-world applications. To address this issue, we propose a dynamic token reduction approach to accelerate PLMs' inference, named TR-BERT, which could flexibly adapt the layer number of each token in inference to avoid redundant calculation. Specially, TR-BERT formulates the token reduction process as a multi-step token selection problem and automatically learns the selection strategy via reinforcement learning. The experimental results on several downstream NLP tasks show that TR-BERT is able to speed up BERT by 2-5 times to satisfy various performance demands. Moreover, TR-BERT can also achieve better performance with less computation in a suite of long-text tasks since its token-level layer number adaption greatly accelerates the self-attention operation in PLMs. The source code and experiment details of this paper can be obtained from https://github.com/thunlp/TR-BERT.