• S
    Of wdl hugectr (#47) · b4ab8201
    ShawnXuan 提交于
    * add dockerfile
    
    * chmod build.sh
    
    * fix
    
    * add scripts for 1 node test
    
    * add multi node test script
    
    * chmod +x
    
    * add launch.sh
    
    * add hugectr_conf.json template
    
    * mv sh and json to scripts
    
    * modify
    
    * modify
    
    * add generate hugectr conf json and gpu memory usage
    
    * extract info from losg
    
    * add README.md
    
    * add comments
    
    * ouput files for report
    
    * add log file name
    
    * update readme
    
    * update hugectr readme
    
    * update hugectr readme
    
    * correct README.md
    
    * add hugectr report
    
    * update tools scripts (#79)
    
    * update
    
    * rm useless files
    
    * add multi nodes test
    
    * ignore unfinished log files
    
    * Fix hugerctr tool (#80)
    
    * ignore unfinished log files
    
    * modify report file name
    
    * update run_all.sh
    
    * update leading dim
    
    * modify scripts
    
    * rm useless script
    
    * rm useless test cases
    
    * update README
    
    * add 4 nodes draft report
    
    * oneflow wdl scripts
    
    * updata oneflow wdl scripts
    
    * add wdl report
    
    * extract scripts
    
    * update git commits
    
    * docker files
    
    * fix table in hugectr report
    
    * re organize WDL folder
    
    * add readme
    
    * draft for summary report of wdl
    
    * add imgs for wdl
    
    * draft report of wdl finished
    
    * add figures of latency and mem usage
    
    * refine picture align
    
    * refine pictures of hugectr
    
    * refine pictures of oneflow wdl
    
    * add fixed test pictures
    
    * remove loss/auc pictures (no need for pefroming tests)
    rename to keep accordingly
    deleted:    ../../../../HugeCTR/reports/hugectr_test_8v100_report.md
    deleted:    ../../../../HugeCTR/reports/imgs/300k_iters_loss_auc.png
    deleted:    ../../../../HugeCTR/reports/imgs/500_iters_loss_auc.png
    renamed:    wdl_report_1027.md -> oneflow_wdl_test_4x8v100_report.md
    
    * refine report(translate)
    
    * refine W&D report overview
    
    * reconstruct wdl tests in OneFolow and HugeCTR
    
    * upload gnuplot imgs and update readme
    
    * fix image and rename
    
    * add: dlperf_Wide_and_Deep_test_report_v1_ch.md
    
    * add: dlperf_wide_and_deep_test_report_v1_cn.md
    
    * refine wdl cn report
    
    * refine reports of wdl
    
    * refine cn typos
    
    * fix links error
    
    * fix imgs url of reports
    
    * add report links
    
    * add blank line
    
    * update conclusion of report
    
    * refine
    
    * fix error link
    
    * fix images
    
    * refine
    
    * add performance data of hugectr
    
    * refine
    
    * refine
    
    * refine
    
    * format
    
    * add description of special cases
    
    * remove useless comments
    
    * refine pip links
    
    * add instructions on run scripts, add hugectr user guide links
    
    * fix typos
    
    * update doker of hugectr
    
    * fix typos
    
    * fix words
    
    * fix words
    
    * fix typos: OneFolow -> OneFlow
    
    * refine en conclusion
    
    * refine README.md
    
    * refine README.md
    
    * refine readme.md
    
    * refine dlperf_wide_and_deep_test_report_v1.md
    
    * fix typos, refine conclusion
    Co-authored-by: NOuYang Yu <xuanjiuye@gmail.com>
    Co-authored-by: Ndoombeaker <later@usopp.net>
    Co-authored-by: MarDino's avatarMARD1NO <359521840@qq.com>
    Co-authored-by: NBBuf <1182563586@qq.com>
    Co-authored-by: NFlowingsun007 <flowingsun007@163.com>
    Co-authored-by: NLiang Depeng <liangdepeng@gmail.com>
    b4ab8201