WRPN: Wide Reduced-Precision Networks

Authors: Asit Mishra, Eriko Nurvitadhi, Jeffrey J Cook, Debbie Marr

ICLR 2018 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental We report results and show that WRPN scheme is better than previously reported accuracies on ILSVRC-12 dataset while being computationally less expensive compared to previously reported reduced-precision networks.
Researcher Affiliation Industry Asit Mishra, Eriko Nurvitadhi, Jeffrey J Cook & Debbie Marr Accelerator Architecture Lab Intel Labs {asit.k.mishra,eriko.nurvitadhi,jeffrey.j.cook,debbie.marr}@intel.com
Pseudocode No The paper describes its quantization scheme using mathematical formulas (Equation 2) but does not provide structured pseudocode or an algorithm block.
Open Source Code No The paper mentions using 'tensorpack' and provides its GitHub link (https://github.com/ppwwyyxx/tensorpack), but this is a third-party framework used by the authors, not their own source code for the WRPN methodology.
Open Datasets Yes We report results on Alex Net (Krizhevsky et al., 2012), batch-normalized Inception (Ioffe & Szegedy, 2015), and Res Net-34 (He et al., 2015) on ILSVRC12 (Russakovsky et al., 2015) dataset. (Russakovsky et al., 2015) refers to 'Image Net Large Scale Visual Recognition Challenge.'
Dataset Splits Yes We use Tensor Flow (Abadi et al., 2015) and tensorpack for all our evaluations and use ILSVRC-12 train and val dataset for analysis.
Hardware Specification Yes For GPU, we evaluate WRPN on Nvidia Titan X Pascal and for FPGA we use Intel Arria-10. Our RTL design targets Arria-10 1150 FPGA. For our ASIC study, we synthesize the PE design using Intel 14 nm process technology to obtain area and energy estimates.
Software Dependencies No The paper states, 'We use Tensor Flow (Abadi et al., 2015) and tensorpack for all our evaluations', but it does not specify version numbers for these software dependencies.
Experiment Setup No The paper states, 'all the runs have same hyper-parameters and training is carried out for the same number of epochs as baseline network' and 'We used the same hyper-parameters and learning rate schedule as the baseline network.' However, it does not explicitly list the specific values of these hyperparameters or the learning rate schedule.