RT-DETR: a real-time DETR
Optional · paper guide 17 of 22 for DLCV 03A · One-Stage Detection, Read Through YOLOv4. Citation numbers refer to the reference list at the bottom of this page.
Navigate: this page and this track: Navigate: this page and this track
On this page
- YOLO v1: detection as a single regression
- YOLOv2 and YOLO9000: anchors, priors and joint training
- YOLOv3: three scales and Darknet-53
- SSD: multi-scale default boxes
- RetinaNet and the focal loss
- FCOS: per-pixel anchor-free detection
- CenterNet: objects as points
- YOLOR: explicit features and learned implicit representations
- YOLOX: anchor-free YOLO with SimOTA
- YOLOv6: a re-parameterized industrial detector
- YOLOv7: a trainable bag of freebies
- YOLOv9: programmable gradient information
- YOLOv10: NMS-free training with dual assignments
- YOLOv12: an attention-centric YOLO
- YOLOv13: hypergraph correlation
- YOLO26: a DFL-free end-to-end YOLO
- RT-DETR: a real-time DETR
- RT-DETRv2: a bag of freebies for RT-DETR
- D-FINE: boxes as refined distributions
- DEIM: dense one-to-one matching
- RF-DETR: architecture search on a DINOv2 backbone
- DEIMv2: DINOv3 features for real-time DETRs
Deep Learning for Computer Vision
Zhao et al.’s CVPR 2024 paper makes direct set prediction competitive with fast YOLO detectors by redesigning the expensive encoder and the initialization of decoder queries [21]. The model first finds useful image features, then refines a small set of candidate objects, rather than generating thousands of dense predictions and suppressing duplicates afterward.
Challenge
DETR [59] removes hand-designed NMS through one-to-one matching, but its original form is expensive and slow to train. Multi-scale features improve detection, especially across object sizes, while increasing the number of tokens. Flattening all high-resolution feature maps into a general-purpose attention encoder can spend much of the runtime on interactions that contribute little to the final boxes.
Meanwhile, a YOLO forward pass is only one part of its deployed detector. Thresholding and NMS depend on the image and confidence settings. RT-DETR therefore asks whether an end-to-end detector can match a fast dense detector when post-processing is counted. The claim is tied to its measured hardware and runtime; it is not a theorem that a Transformer is faster than a convolutional detector.
Method

Figure source: [21], arXiv:2304.08069v3, PDF page 5. Original PDF.
Backbone and shapes. The backbone produces three feature maps, commonly written , at strides 8, 16 and 32. Project their channel dimensions into a shared width . For a input, the spatial grids contain , and positions. These sizes explain the bottleneck: the finest level contributes most tokens, while the coarsest level carries the strongest semantic abstraction.
AIFI: within-scale interaction. Attention-based Intra-scale Feature Interaction applies self-attention only to the coarsest map. Schematically,
is the coarse feature sequence, its number of spatial locations, and its spatially restored output. Positional information is supplied to attention. Interacting semantically rich coarse tokens captures useful relationships while avoiding all-pairs attention over the much longer concatenated sequence. It does not mean high-resolution information is discarded; the other maps enter the next module.
CCFF: between-scale fusion. CNN-based Cross-scale Feature Fusion combines through top-down and bottom-up paths. Convolutions adjust channels and fuse adjacent resolutions; RepBlocks provide a training-friendly fusion module with a simpler deployable form. This deliberately assigns different jobs to different operations: attention handles coarse semantic interaction, and convolution handles efficient spatial and scale fusion. The result is a sequence of multi-scale image features for the decoder.
Query selection. The encoder predicts classes and boxes at candidate locations. Selecting queries by class confidence alone can choose features that recognize an object but localize it poorly. The paper’s uncertainty-minimal query selection couples classification supervision to localization quality, seeking features that do both well. A fixed number of high-ranking features initialize decoder content and box positions.
The term “uncertainty” here needs care. The paper motivates it using a discrepancy between classification and localization predictions; it is not a separately measured posterior variance that proves the system knows when it is wrong. The practical mechanism changes the training of the encoder predictions so that the scores used for selection reflect box quality. It is also not an instruction to use ground-truth IoU during deployment: ground truth is available only in training and evaluation.
Decoder and matching. Object queries attend to image features and interact with one another, then produce class scores and box refinements at each decoder layer. One-to-one matching pairs predicted objects with ground-truth objects for the training loss. Unmatched predictions learn the background objective. A compact way to express the assignment is
is the number of ground-truth objects, an annotated class and box, a query prediction, and the matching cost combining classification and localization terms. An injective map assigns each target a distinct query; the Hungarian algorithm solves the finite assignment [60]. The cost used to choose assignments and the differentiable losses applied after assignment are related but distinct objects.
Why inference can omit NMS. Training rewards one query for each object and penalizes redundant foreground predictions through unmatched-query supervision. This encourages a set of unique detections. It does not mathematically prohibit duplicates. Inference still needs score selection, box decoding, coordinate restoration and transfer of outputs; “end to end” does not erase those costs.
Adjustable decoder depth. Auxiliary prediction heads supervise intermediate decoder outputs. A trained detector can return an earlier layer to trade some accuracy for latency without retraining. This is early output from an iterative refinement network, not a guarantee that every layer count has the same accuracy. A deployment implementation must actually stop computation at the chosen layer to obtain its speed benefit.
Results
The main table reports 53.1 AP / 108 FPS for RT-DETR-R50 and 54.3 AP / 74 FPS for RT-DETR-R101 on COCO validation at , using a T4 GPU and TensorRT FP16 with batch one [21]. Keep those conditions attached to the values. The paper compares against selected contemporary YOLO systems under its stated timing protocol, including post-processing; these numbers cannot be directly ranked against YOLOv4’s different GPU measurements.
Another important result is that useful accuracy remains when returning intermediate decoder outputs. For the six-layer-trained R50 model, Table 5 gives 53.1 AP at layer 6, 53.0 at layer 5, 52.7 at layer 4 and 52.4 at layer 3, with latency decreasing from 9.3 to 7.9 ms between the six- and three-layer outputs [21]. These are a measured trade-off for this trained model and runtime, not universal layer costs.
Ablations
Table 3 starts with a detector without the proposed encoder and adds different ways of processing scales. Its baseline A gives 43.0 AP / 7.2 ms. Within-scale interaction raises AP to 44.9 but latency to 11.1 ms; cross-scale interaction yields 45.6 / 13.3 ms. Decoupling the operations gives 46.4 / 12.2 ms. Restricting attention to the coarsest feature map gives 46.8 / 7.9 ms. The complete hybrid encoder reaches 47.9 / 9.3 ms [21]. This is a shorter ablation training setting, so its AP should not be substituted into the main table.
The lesson is stronger than “attention improves accuracy.” Where attention is applied determines its cost, and a hybrid can preserve useful interaction while avoiding expensive redundant processing. The final encoder is not the cheapest row, but it provides a better measured accuracy–latency compromise.
Table 4 compares query-selection training: 47.9 AP for vanilla selection and 48.7 AP for uncertainty-minimal selection in its ablation setting [21]. The associated score/IoU plots support the explanation that selection improves initialization quality. They do not show that selected queries never miss small objects or that uncertainty is calibrated in a deployment domain.
Conclusion
RT-DETR makes the NMS-free alternative concrete: control the encoder’s expensive token interactions, fuse scales efficiently, and initialize a compact decoder from useful image features. It gives a clear counterpart to YOLOv4’s dense-head recipe. The two systems address the same detection task through different prediction and assignment structures.
Problems
The paper explicitly identifies small-object detection as a weakness relative to some contemporary YOLO comparisons. Coarse semantic attention, finite query counts and downsampling do not remove the need to preserve tiny-object evidence. Dense scenes may also contain more annotated objects than a fixed query budget can represent.
Deployment support matters. Deformable sampling in the decoder may be harder to export efficiently than standard convolutions. A fast TensorRT result does not establish the same advantage on a CPU, mobile accelerator or unsupported operator set. RT-DETRv2 directly addresses part of this issue.
Finally, pretrained backbone data, training length, image resolution and precision all affect the comparison. Reproduce AP and latency with the same exported model artifact, rather than taking AP from a training framework and latency from a quantized graph whose outputs have not been checked. This lesson becomes especially clear in the later RF-DETR study.
References
- C.-Y. Wang, A. Bochkovskiy and H.-Y. M. Liao, “YOLOv7: Trainable bag-of-freebies sets new state-of-the-art for real-time object detectors,” in Proc. IEEE/CVF CVPR, pp. 7464–7475, 2023. doi
- C.-Y. Wang, I-H. Yeh and H.-Y. M. Liao, “YOLOv9: Learning what you want to learn using programmable gradient information,” in Proc. ECCV, 2024. doi
- G. Jocher, “Ultralytics YOLOv5,” software (first released 2020; v7.0 released 22 November 2022). release · documentation
- G. Jocher, A. Chaurasia and J. Qiu, “Ultralytics YOLOv8,” software, version 8.0.0, 2023 (no paper). docs · code
- G. Jocher and J. Qiu, “Ultralytics YOLO11,” software, version 11.0.0, 2024 (no paper). docs · code
- J. Lu, J. Jia, J. Yawl and C. Zhang, “YOLOv14: Adaptive real-time object detection for diverse imaging conditions,” arXiv:2608.04720, 2026 (withdrawn by the authors, 20 August 2026). arXiv
- J. Redmon, S. Divvala, R. Girshick and A. Farhadi, “You only look once: Unified, real-time object detection,” in Proc. IEEE CVPR, pp. 779–788, 2016. doi · arXiv
- J. Redmon and A. Farhadi, “YOLO9000: Better, faster, stronger,” in Proc. IEEE CVPR, pp. 6517–6525, 2017. doi · arXiv
- J. Redmon and A. Farhadi, “YOLOv3: An incremental improvement,” tech. report, arXiv:1804.02767, 2018. arXiv
- W. Liu, D. Anguelov, D. Erhan, C. Szegedy, S. Reed, C. Fu, A. C. Berg, “SSD: Single Shot MultiBox Detector,” Lecture Notes in Computer Science / Computer Vision – ECCV 2016, 2016. doi · manuscript
- T.-Y. Lin, P. Goyal, R. Girshick, K. He and P. Dollár, “Focal loss for dense object detection,” in Proc. IEEE ICCV, pp. 2999–3007, 2017. doi
- Z. Tian, C. Shen, H. Chen, T. He, “FCOS: Fully Convolutional One-Stage Object Detection,” 2019 IEEE/CVF International Conference on Computer Vision (ICCV), 2019. doi · manuscript
- X. Zhou, D. Wang, P. Krähenbühl, “Objects as Points,” arXiv preprint, 2019. source
- C. Wang, I. Yeh, H. M. Liao, “You Only Learn One Representation: Unified Network for Multiple Tasks,” arXiv preprint, 2021. source
- Z. Ge, S. Liu, F. Wang, Z. Li and J. Sun, “YOLOX: Exceeding YOLO series in 2021,” arXiv:2107.08430, 2021. arXiv
- C. Li, L. Li, H. Jiang, K. Weng, Y. Geng et al., “YOLOv6: A single-stage object detection framework for industrial applications,” tech. report, arXiv:2209.02976, 2022. arXiv
- A. Wang, H. Chen, L. Liu, K. Chen, Z. Lin, J. Han and G. Ding, “YOLOv10: Real-time end-to-end object detection,” in Proc. NeurIPS, 2024. arXiv
- Y. Tian, Q. Ye and D. Doermann, “YOLOv12: Attention-centric real-time object detectors,” in Advances in Neural Information Processing Systems 38 (NeurIPS), 2025. proceedings · arXiv
- M. Lei, S. Li, Y. Wu, H. Hu, Y. Zhou et al., “YOLOv13: Real-time object detection with hypergraph-enhanced adaptive visual perception,” arXiv:2506.17733, 2025. arXiv
- G. Jocher, J. Qiu, M. Liu, S. Lyu, F. C. Akyon and M. E. Kalfaoglu, “Ultralytics YOLO26: Unified real-time end-to-end vision models,” arXiv:2606.03748, 2026 (software released January 2026). arXiv · docs
- Y. Zhao, W. Lv, S. Xu, J. Wei, G. Wang, Q. Dang, Y. Liu, J. Chen, “DETRs Beat YOLOs on Real-time Object Detection,” Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2024. source
- W. Lv, Y. Zhao, Q. Chang, K. Huang, G. Wang and Y. Liu, “RT-DETRv2: Improved baseline with bag-of-freebies for real-time detection transformer,” arXiv:2407.17140, 2024. arXiv
- Y. Peng, H. Li, P. Wu, Y. Zhang, X. Sun and F. Wu, “D-FINE: Redefine regression task in DETRs as fine-grained distribution refinement,” in Proc. ICLR, 2025. arXiv · code
- S. Huang, Z. Lu, X. Cun, Y. Yu, X. Zhou, X. Shen, “DEIM: DETR with Improved Matching for Fast Convergence,” Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2025. source
- I. Robinson, P. Robicheaux, M. Popov, D. Ramanan and N. Peri, “RF-DETR: Neural architecture search for real-time detection transformers,” in Proc. ICLR, 2026. arXiv
- S. Huang, Y. Hou, L. Liu, X. Yu and X. Shen, “Real-time object detection meets DINOv3,” arXiv:2509.20787, 2025 (revised 2026). arXiv
- A. Bochkovskiy, C.-Y. Wang and H.-Y. M. Liao, “YOLOv4: Optimal speed and accuracy of object detection,” arXiv:2004.10934, 2020. arXiv
- M. Tan, R. Pang, Q. V. Le, “EfficientDet: Scalable and Efficient Object Detection,” 2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2020. doi · manuscript
- NVIDIA, “TensorRT documentation,” official documentation. link
- J. Deng, W. Dong, R. Socher, L. Li, Kai Li, Li Fei-Fei, “ImageNet: A large-scale hierarchical image database,” 2009 IEEE Conference on Computer Vision and Pattern Recognition, 2009. source
- K. Simonyan, A. Zisserman, “Very Deep Convolutional Networks for Large-Scale Image Recognition,” arXiv preprint, 2014. source
- K. He, X. Zhang, S. Ren, J. Sun, “Deep Residual Learning for Image Recognition,” arXiv preprint, 2015. source
- T.-Y. Lin, P. Dollár, R. Girshick, K. He, B. Hariharan and S. Belongie, “Feature pyramid networks for object detection,” in Proc. IEEE CVPR, 2017. doi
- S. Liu, L. Qi, H. Qin, J. Shi and J. Jia, “Path aggregation network for instance segmentation,” in Proc. IEEE/CVF CVPR, pp. 8759–8768, 2018. doi
- K. He, X. Zhang, S. Ren, J. Sun, “Spatial Pyramid Pooling in Deep Convolutional Networks for Visual Recognition,” Lecture Notes in Computer Science / Computer Vision – ECCV 2014, 2014. doi · manuscript
- C.-Y. Wang, H.-Y. M. Liao, I-H. Yeh, Y.-H. Wu, P.-Y. Chen and J.-W. Hsieh, “CSPNet: A new backbone that can enhance learning capability of CNN,” CVPR Workshops, 2020. arXiv · doi
- A. Bochkovskiy and contributors, “Darknet: YOLOv4 configuration and detection layer,” official implementation, accessed 2 November 2026. configuration · assignment and decoding
- M. Tan, Q. V. Le, “EfficientNet: Rethinking Model Scaling for Convolutional Neural Networks,” ICML, 2019. source · proceedings
- G. Huang, Z. Liu, L. Van Der Maaten, K. Q. Weinberger, “Densely Connected Convolutional Networks,” 2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2017. doi · manuscript
- S. Xie, R. Girshick, P. Dollar, Z. Tu, K. He, “Aggregated Residual Transformations for Deep Neural Networks,” 2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2017. doi · manuscript
- D. Misra, “Mish: A Self Regularized Non-Monotonic Activation Function,” BMVC, 2020. source
- P. Ramachandran, B. Zoph, Q. V. Le, “Searching for Activation Functions,” arXiv preprint, 2017. source
- J. Yu, Y. Jiang, Z. Wang, Z. Cao, T. Huang, “UnitBox: An Advanced Object Detection Network,” ACM Multimedia, 2016. source
- H. Rezatofighi, N. Tsoi, J. Gwak, A. Sadeghian, I. Reid, S. Savarese, “Generalized Intersection over Union: A Metric and A Loss for Bounding Box Regression,” arXiv preprint, 2019. source
- Z. Zheng, P. Wang, W. Liu, J. Li, R. Ye and D. Ren, “Distance-IoU loss: Faster and better learning for bounding box regression,” in Proc. AAAI, pp. 12993–13000, 2020. doi
- H. Zhang, M. Cisse, Y. N. Dauphin and D. Lopez-Paz, “mixup: Beyond empirical risk minimization,” in Proc. ICLR, 2018. arXiv
- S. Yun, D. Han, S. Chun, S. J. Oh, Y. Yoo, J. Choe, “CutMix: Regularization Strategy to Train Strong Classifiers With Localizable Features,” 2019 IEEE/CVF International Conference on Computer Vision (ICCV), 2019. doi · manuscript
- Nitish Srivastava, Geoffrey Hinton, Alex Krizhevsky, Ilya Sutskever, Ruslan Salakhutdinov, “Dropout: A Simple Way to Prevent Neural Networks from Overfitting,” Journal of Machine Learning Research, 2014. source
- G. Ghiasi, T. Lin, Q. V. Le, “DropBlock: A regularization method for convolutional networks,” NeurIPS, 2018. source · proceedings
- C. Szegedy, V. Vanhoucke, S. Ioffe, J. Shlens, Z. Wojna, “Rethinking the Inception Architecture for Computer Vision,” 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2016. doi · manuscript
- S. Ioffe and C. Szegedy, “Batch normalization: Accelerating deep network training by reducing internal covariate shift,” in Proc. ICML, PMLR vol. 37, pp. 448–456, 2015. link
- Z. Yao, Y. Cao, S. Zheng, G. Huang, S. Lin, “Cross-Iteration Batch Normalization,” 2021 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2021. doi · manuscript
- S. Woo, J. Park, J. Lee, I. S. Kweon, “CBAM: Convolutional Block Attention Module,” Lecture Notes in Computer Science / Computer Vision – ECCV 2018, 2018. doi · manuscript
- I. Loshchilov, F. Hutter, “SGDR: Stochastic Gradient Descent with Warm Restarts,” ICLR, 2017. source
- T.-Y. Lin, M. Maire, S. Belongie, J. Hays, P. Perona, D. Ramanan, P. Dollár and C. L. Zitnick, “Microsoft COCO: Common objects in context,” in Proc. ECCV, pp. 740–755, 2014. doi
- COCO Consortium, “COCO detection evaluation,” official evaluation description. link
- ONNX, “Open Neural Network Exchange,” official project page. link
- B. Jacob, S. Kligys, B. Chen, M. Zhu, M. Tang, A. Howard, H. Adam and D. Kalenichenko, “Quantization and training of neural networks for efficient integer-arithmetic-only inference,” in Proc. IEEE/CVF CVPR, pp. 2704–2713, 2018. doi
- N. Carion, F. Massa, G. Synnaeve, N. Usunier, A. Kirillov and S. Zagoruyko, “End-to-end object detection with transformers,” in Proc. ECCV, 2020. doi
- H. W. Kuhn, “The Hungarian method for the assignment problem,” Naval Research Logistics Quarterly, vol. 2, pp. 83–97, 1955. doi
- J. Terven, D.-M. Córdova-Esparza and J.-A. Romero-González, “A comprehensive review of YOLO architectures in computer vision: From YOLOv1 to YOLOv8 and YOLO-NAS,” Machine Learning and Knowledge Extraction, vol. 5, no. 4, pp. 1680–1716, 2023. doi · arXiv
- N. Jegham, C. Y. Koh, M. Abdelatti and A. Hendawi, “YOLO evolution: A comprehensive benchmark and architectural review of YOLOv12, YOLO11, and their previous versions,” arXiv:2411.00201, 2024 (revised 2025). arXiv
- R. Sapkota and M. Karkee, “Ultralytics YOLO evolution: An overview of YOLO27, YOLO26, YOLO11, YOLOv8, and YOLOv5 object detectors for computer vision and pattern recognition,” arXiv:2510.09653, 2025 (revised 2026). arXiv
- T. Shehzadi, K. A. Hashmi, M. Liwicki, D. Stricker and M. Z. Afzal, “Object detection with transformers: A review,” Sensors, vol. 25, no. 19, 2025. doi
- Ultralytics, “YOLO27 (coming soon),” official documentation page, accessed 2 November 2026. docs
- M. Oquab, T. Darcet, T. Moutakanni, H. Vo, M. Szafraniec et al., “DINOv2: Learning robust visual features without supervision,” arXiv:2304.07193, 2023. arXiv
- O. Siméoni, H. V. Vo, M. Seitzer, F. Baldassarre, M. Oquab et al., “DINOv3,” arXiv:2508.10104, 2025. arXiv
- X. Ding, X. Zhang, N. Ma, J. Han, G. Ding and J. Sun, “RepVGG: Making VGG-style ConvNets great again,” in Proc. IEEE/CVF CVPR, pp. 13728–13737, 2021. doi
- R. Girshick, “Fast R-CNN,” 2015 IEEE International Conference on Computer Vision (ICCV), 2015. doi · manuscript
- S. Ren, K. He, R. Girshick, J. Sun, “Faster R-CNN: Towards Real-Time Object Detection with Region Proposal Networks,” arXiv preprint, 2015. source
- M. Everingham, L. Van Gool, C. K. I. Williams, J. Winn, A. Zisserman, “The Pascal Visual Object Classes (VOC) Challenge,” International Journal of Computer Vision, 2009. source
- A. Shrivastava, A. Gupta, R. Girshick, “Training Region-Based Object Detectors with Online Hard Example Mining,” 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2016. doi · manuscript
- H. Zhang, Y. Wang, F. Dayoub and N. Sünderhauf, “VarifocalNet: An IoU-aware dense object detector,” in Proc. IEEE/CVF CVPR, pp. 8510–8519, 2021. doi
- YOLOX contributors, “YOLOX detection head,” official implementation, accessed 2 November 2026. code
- K. Duan, S. Bai, L. Xie, H. Qi, Q. Huang, Q. Tian, “CenterNet: Keypoint Triplets for Object Detection,” arXiv preprint, 2019. source
- Z. Ge, S. Liu, Z. Li, O. Yoshie and J. Sun, “OTA: Optimal transport assignment for object detection,” in Proc. IEEE/CVF CVPR, pp. 303–312, 2021. doi
- C. Li, L. Li, Y. Geng, et al., “YOLOv6 v3.0: A Full-Scale Reloading,” arXiv preprint, 2023. source
- C. Feng, Y. Zhong, Y. Gao, M. R. Scott, W. Huang, “TOOD: Task-Aligned One-Stage Object Detection,” Proceedings of the IEEE/CVF International Conference on Computer Vision, 2021. source
- Z. Gevorgyan, “SIoU Loss: More Powerful Learning for Bounding Box Regression,” arXiv preprint, 2022. source
- X. Li, W. Wang, L. Wu, S. Chen, X. Hu, J. Li, J. Tang and J. Yang, “Generalized focal loss: Learning qualified and distributed bounding boxes for dense object detection,” in Advances in Neural Information Processing Systems 33 (NeurIPS), 2020. proceedings
- X. Ding, H. Chen, X. Zhang, K. Huang, J. Han, G. Ding, “Re-parameterizing Your Optimizers rather than Architectures,” ICLR, 2023. source
- S. Zhang, C. Chi, Y. Yao, Z. Lei, S. Z. Li, “Bridging the Gap Between Anchor-Based and Anchor-Free Detection via Adaptive Training Sample Selection,” 2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2020. doi · manuscript
- C. Wang, H. M. Liao, I. Yeh, “Designing Network Design Strategies Through Gradient Path Analysis,” arXiv preprint, 2022. source
- T. Dao, D. Fu, S. Ermon, A. Rudra, C. Ré, “FlashAttention: Fast and Memory-Efficient Exact Attention with IO-Awareness,” Advances in Neural Information Processing Systems 35, 2022. doi · manuscript
- D. P. Kingma, J. Ba, “Adam: A Method for Stochastic Optimization,” ICLR, 2015. source
- K. Jordan, “Muon: An optimizer for hidden layers in neural networks,” technical blog, 2024. source
- S. Shao, Z. Li, T. Zhang, C. Peng, G. Yu, X. Zhang, J. Li, J. Sun, “Objects365: A Large-Scale, High-Quality Dataset for Object Detection,” Proceedings of the IEEE/CVF International Conference on Computer Vision, 2019. source