Papers › Rethinking and Improving Relative Position Encoding for Vision Transformer

Rethinking and Improving Relative Position Encoding for Vision Transformer

29 Jul 2021ICCV 2021 10arXiv:2107.14222archive 2025-07-28

Kan Wu, Houwen Peng, Minghao Chen, Jianlong Fu, Hongyang Chao

Relative position encoding (RPE) is important for transformer to capture sequence ordering of input tokens. General efficacy has been proven in natural language processing. However, in computer vision, its efficacy is not well studied and even remains controversial, e.g., whether relative position encoding can work equally well as absolute position? In order to clarify this, we first review existing relative position encoding methods and analyze their pros and cons when applied in vision transformers. We then propose new relative position encoding methods dedicated to 2D images, called image RPE (iRPE). Our methods consider directional relative distance modeling as well as the interactions between queries and relative position embeddings in self-attention mechanism. The proposed iRPE methods are simple and lightweight. They can be easily plugged into transformer blocks. Experiments demonstrate that solely due to the proposed encoding methods, DeiT and DETR obtain up to 1.5% (top-1 Acc) and 1.3% (mAP) stable improvements over their original versions on ImageNet and COCO respectively, without tuning any extra hyperparameters such as learning rate and weight decay. Our ablation and analysis also yield interesting findings, some of which run counter to previous understanding. Code and models are open-sourced at https://github.com/microsoft/Cream/tree/main/iRPE.

PaperPDFConference PDFCodeCode Syntology ran

In Syntology Open this paper in Syntology's Atlas, the map of the papers in Syntology's graph and their citations.

For agents, Syntology's MCP tool lists every function and class Syntology harvested from this paper and whether it ran (how to connect): get_harvested_code_for_paper(arxiv_id="2107.14222")

Code

Syntology Ran 7 of 9 code samples harvested from 1 repository linked to this paper; 2 have no recorded run. Of those that ran: 4 ran · our draft was wrong; 3 ran · fixture could not drive it.

By repository: official repository: 9 samples from 1 repository, 7 ran. The run record, sample by sample. “Ran” means executed on a synthesized input, not that the code is correct or reproduces the paper.

microsoft/cream officialmentioned in paperpytorch report

Repository list and official/mentioned flags are the archive's, frozen 2025-07-28. Reachability, where shown, is from one Syntology probe window (2026-09-16 to 2026-09-18); repositories not probed show nothing. GitHub stars are not tracked.

Code Syntology ran Syntology

9 samples harvested; 7 ran; 0 honoured the contract we drafted; 2 have no recorded run. Read from Syntology's graph 2026-09-24; that is when this build read the record, not when the samples ran.

4ran · our draft was wrong
3ran · fixture could not drive it
2unverified

Licence: 0 of the 9 samples are pointer only, meaning Syntology does not serve that copy's text. This page shows no code text for any sample; each one links to its file in the repository.

Harvested from microsoft/Cream. “Ran” means the sample executed on a synthesized input. It does not mean the output is correct, and nothing here reproduces the paper's results. “Honoured” and “violated” refer to a contract Syntology drafted from the code itself; “our draft was wrong” and “fixture could not drive it” are failures of Syntology's instrument, not of the code.

Each sample ends with its code_sha256, Syntology's identity for that exact code. An agent fetches the stored sample with Syntology's MCP tool get_code(code_sha256="…") (how to connect); click an identity to copy that call.

Repository labels, per sample. official repository: The archive marks this repository official for the paper. named in the paper: The archive records that the paper mentions this repository; it is not marked official. community (archive-listed): In the archive's code links for this paper, not marked official and not recorded as mentioned in the paper. found in paper text by Syntology: Syntology found this repository in the paper's own text; whether it is the authors' implementation is not asserted. community: Not in the archive's code links for this paper; a community repository Syntology harvested. Samples from a repository marked official are listed first. Licence labels name the repository's licence as recorded at harvest. “Pointer only” means Syntology does not serve that copy's text, for one of four reasons: no licence file was found; the licence was not identified; the licence is recorded as permissive but that copy's record is not marked cleared; or the licence is outside the permissive list Syntology serves text under (MIT, Apache-2.0, BSD and similar). Some licences outside that list permit redistribution, such as WTFPL, and GPL-3.0 under its conditions; they are simply not on the list. Hover a licence label for the reason. File links open the file on GitHub at the default branch, which may have changed since the harvest.

_rp_2d_cross_cols microsoft/Cream/iRPE/DETR-with-iRPE/models/rpe_attention/irpe.py official repository ran · fixture could not drive it MIT (permissive) · affb9ad6b732b9e4 · report
_rp_2d_cross_rows microsoft/Cream/iRPE/DETR-with-iRPE/models/rpe_attention/irpe.py official repository ran · fixture could not drive it MIT (permissive) · 434e634a4741440f · report
_rp_2d_euclidean microsoft/Cream/iRPE/DETR-with-iRPE/models/rpe_attention/irpe.py official repository ran · our draft was wrong MIT (permissive) · 37c3e0e9c4bac792 · report
_rp_2d_product microsoft/Cream/iRPE/DETR-with-iRPE/models/rpe_attention/irpe.py official repository ran · fixture could not drive it MIT (permissive) · cd9ab56e1ede5e05 · report
_rp_2d_quant microsoft/Cream/iRPE/DETR-with-iRPE/models/rpe_attention/irpe.py official repository ran · our draft was wrong MIT (permissive) · f1ae97ecf9f3f3e1 · report
get_bucket_ids_2d_without_skip microsoft/Cream/iRPE/DETR-with-iRPE/models/rpe_attention/irpe.py official repository ran · our draft was wrong MIT (permissive) · 8a9e6d9d71d3c2f8 · report
piecewise_index microsoft/Cream/iRPE/DETR-with-iRPE/models/rpe_attention/irpe.py official repository ran · our draft was wrong MIT (permissive) · 5632af974aa9bde4 · report
get_bucket_ids_2d microsoft/Cream/iRPE/DETR-with-iRPE/models/rpe_attention/irpe.py official repository unverified MIT (permissive) · c6d5f9784d683be8 · report
iRPE microsoft/Cream/iRPE/DETR-with-iRPE/models/rpe_attention/irpe.py official repository unverified MIT (permissive) · 5bc74c794e5f05ef · report

Tasks

Image ClassificationObject Detection

1 archive task tag without a task page not shown.

Results from the paper archive 2025-07-28

TaskDatasetModelMetricValueRank at snapshotLeaderboardReport
Image Classification ImageNet DeiT-B with iRPE-K GFLOPs 35.368 #542 of 1060 Archive leaderboard report
Image Classification ImageNet DeiT-B with iRPE-K Number of params 87M #542 of 1060 Archive leaderboard report
Image Classification ImageNet DeiT-B with iRPE-K Top 1 Accuracy 82.4% #542 of 1060 Archive leaderboard report
Image Classification ImageNet DeiT-S with iRPE-QKV GFLOPs 9.770 #642 of 1060 Archive leaderboard report
Image Classification ImageNet DeiT-S with iRPE-QKV Top 1 Accuracy 81.4% #642 of 1060 Archive leaderboard report
Image Classification ImageNet DeiT-S with iRPE-QK GFLOPs 9.412 #660 of 1060 Archive leaderboard report
Image Classification ImageNet DeiT-S with iRPE-QK Top 1 Accuracy 81.1% #660 of 1060 Archive leaderboard report
Image Classification ImageNet DeiT-S with iRPE-K GFLOPs 9.318 #675 of 1060 Archive leaderboard report
Image Classification ImageNet DeiT-S with iRPE-K Number of params 22M #675 of 1060 Archive leaderboard report
Image Classification ImageNet DeiT-S with iRPE-K Top 1 Accuracy 80.9% #675 of 1060 Archive leaderboard report
Image Classification ImageNet DeiT-Ti with iRPE-K GFLOPs 2.568 #985 of 1060 Archive leaderboard report
Image Classification ImageNet DeiT-Ti with iRPE-K Number of params 6M #985 of 1060 Archive leaderboard report
Image Classification ImageNet DeiT-Ti with iRPE-K Top 1 Accuracy 73.7% #985 of 1060 Archive leaderboard report
Object Detection COCO minival DETR-ResNet50 with iRPE-K (300 epochs) box AP 42.3 #157 of 220 Archive leaderboard report
Object Detection COCO minival DETR-ResNet50 with iRPE-K (150 epochs) box AP 40.8 #175 of 220 Archive leaderboard report

Ranks are positions in the archive's leaderboards as they stood at the 2025-07-28 snapshot. Results published since then are not among these rows, so a rank here is not a current standing.

Methods

Absolute Position EncodingsAdamAttentionConvolutionDeiTDense ConnectionsDetrDropoutFeedforward NetworkLabel SmoothingLayer NormalizationLinear LayerMulti-Head AttentionPosition-Wise Feed-Forward LayerResidual ConnectionSoftmaxTransformer

Report a problem or propose a change · a person checks every report against the paper or source before anything changes; decisions are listed on /corrections