Container: Context Aggregation Networks

Authors: peng gao, Jiasen Lu, hongsheng Li, Roozbeh Mottaghi, Aniruddha Kembhavi

NeurIPS 2021 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental Our CONTAINER architecture achieves 82.7 % Top-1 accuracy on Image Net using 22M parameters, +2.8 improvement compared with Dei T-Small, and can converge to 79.9 % Top-1 accuracy in just 200 epochs.
Researcher Affiliation Collaboration Peng Gao1,2, Jiasen Lu4, Hongsheng Li2, Roozbeh Mottaghi3,4, Aniruddha Kembhavi3,4 1 Shanghai AI Laboratory 2 CUHK-Sense Time Joint Lab, CUHK 3 University of Washington 4 PRIOR @ Allen Institute for AI
Pseudocode No The paper does not contain structured pseudocode or algorithm blocks.
Open Source Code Yes Code is released at https://github.com/allenai/container.
Open Datasets Yes Our CONTAINER architecture achieves 82.7 % Top-1 accuracy on Image Net using 22M parameters... Table 1: Image Net [12] Top-1 accuracy comparison... Table 4: Comparing the CONTAINER-LIGHT backbone with several previous methods at the tasks of object detection and instance segmentation using the Mask-RCNN and Retina Net networks. on the COCO dataset [38].
Dataset Splits Yes We train Dei T [54] and CONTAINER-LIGHT for 100 epochs at the self supervised task of visual representation learning using the DINO framework [6]. Table 3: Image Net Top-1 Acc for CONTAINER-LIGHT and Dei T-S with varying training sizes. Table 4 compares several backbones applied to the Retina Net detector [37] on the COCO dataset [38].
Hardware Specification No The paper does not provide specific hardware details such as exact GPU/CPU models, processor types, or memory amounts used for running its experiments. It only mentions general terms like 'huge GPU memory'.
Software Dependencies No The paper does not provide specific ancillary software details with version numbers (e.g., 'PyTorch 1.9', 'TensorFlow 2.x') needed to replicate the experiment.
Experiment Setup No The paper states: 'Please see the appendix for details of the models, training and setup.' This indicates that specific experimental setup details, such as hyperparameters or system-level training settings, are not provided in the main text.