Re: runs reviewing
29 views
Skip to first unread message
Seif Mechti
Jun 14, 2020, 7:01:21 AM6/14/20
to Matti Wiegmann, Martin Potthast, pan-works...@googlegroups.com
Hello Matti
When we can see the results of the celebrity profiling task...
Thank you
Le ven. 12 juin 2020 à 9:13 PM, Seif Mechti <mecht...@gmail.com> a écrit :
Hello Matti ...We validate the run ...Thank you ...Le mar. 9 juin 2020 à 3:24 PM, Matti Wiegmann <matti.w...@uni-weimar.de> a écrit :Hey Seif,You can enter one run to the competition. If you have more then one valid evaluation, please let me know which one. Naturally, we can't show you the results beforehand.There are currently 3 Teams participating in Celebrity Profiling.You can continue working until the end of the week.Best,MattiOn Tue, Jun 9, 2020 at 2:13 PM Seif Mechti <mecht...@gmail.com> wrote:Thank you Matti for your supportIf we continue due to competitive purpose ...we can ameliorate results?If you unblind the results ..we can't continue?How many teams in the competition please ?For the results and the overview paper ,if possible , The name of our Team is isTU-KSA20The new deadline is 12 /06?Thank you in advanceLe lun. 8 juin 2020 à 9:55 AM, Matti Wiegmann <matti.w...@uni-weimar.de> a écrit :Yes, the last runs worked, Congratulations!Just tell me if you've finished your submission, then I'll accept them as entries to the competition and unblind the results.Best,MattiOn Mon, Jun 8, 2020 at 9:40 AM Seif Mechti <mecht...@gmail.com> wrote:We fixed fhe problem ...Thank you ....i think now fhe run is ok ...Le sam. 6 juin 2020 à 1:58 PM, Matti Wiegmann <matti.w...@uni-weimar.de> a écrit :Hey,I can't unblind the full logs, since you expose the data. I can post you the uncritical parts here:Stdout:python3 -W ignore full_predict.py /media/test-datasets/celebrity-profiling/pan20-celebrity-profiling-test-dataset-2020-02-28 /tmp/mechti20/2020-06-05-23-05-23/output followers_celebrities ['/media/test-datasets/celebrity-profiling/pan20-celebrity-profiling-test-dataset-2020-02-28', '/tmp/mechti20/2020-06-05-23-05-23/output', 'followers_celebrities'] $inputDataset : /media/test-datasets/celebrity-profiling/pan20-celebrity-profiling-test-dataset-2020-02-28 $outputDir : /tmp/mechti20/2020-06-05-23-05-23/output $outputFile : /tmp/mechti20/2020-06-05-23-05-23/output/labels.ndjson $modelDir : followers_celebrities $inputDataset : /media/test-datasets/celebrity-profiling/pan20-celebrity-profiling-test-dataset-2020-02-28/follower-feeds.ndjson $inputDataset : /media/test-datasets/celebrity-profiling/pan20-celebrity-profiling-test-dataset-2020-02-28/follower-feeds.ndjson write ./splits/follower-feeds_1.ndjson write ./splits/follower-feeds_2.ndjson write ./splits/follower-feeds_3.ndjson write ./splits/follower-feeds_4.ndjson write ./splits/follower-feeds_5.ndjson write ./splits/follower-feeds_6.ndjson write ./splits/follower-feeds_7.ndjson write ./splits/follower-feeds_8.ndjson write ./splits/follower-feeds_9.ndjson write ./splits/follower-feeds_10.ndjson write ./splits/follower-feeds_11.ndjson write ./splits/follower-feeds_12.ndjson write ./splits/follower-feeds_13.ndjson write ./splits/follower-feeds_14.ndjson write ./splits/follower-feeds_15.ndjson write ./splits/follower-feeds_16.ndjson write ./splits/follower-feeds_17.ndjson write ./splits/follower-feeds_18.ndjson write ./splits/follower-feeds_19.ndjson write ./splits/follower-feeds_20.ndjson clean/follower-feeds_18.csv /home/mechti20/pan-2020/splits/follower-feeds_18.ndjsonid text 0 10526 [[RT @DominicKing_DM: Footnote to the game. Li...<continues>cleaning... 67.86444044113159 /home/mechti20/pan-2020/clean/follower-feeds_8.csv id text 0 10326 yes greatest economy history thank corey stand...<continues>4 10210 2lac diyas #bringbackdrgpradhan swagat firse 1...This was the last line. There is nothing in the Stderr.On Sat, Jun 6, 2020 at 2:45 PM Seif Mechti <mecht...@gmail.com> wrote:Execution logs....sorryLe sam. 6 juin 2020 à 1:44 PM, Seif Mechti <mecht...@gmail.com> a écrit :Dear Matti ...we are using the same software ..for the train and for test ...please can we obtain the logs errors to fix the problemThank you in advance...Le sam. 6 juin 2020 à 12:58 PM, Matti Wiegmann <matti.w...@uni-weimar.de> a écrit :Hey Seif,Apparently your software still does not produce a labels.ndjson in the output dir. Is there any major difference to the working software you ran on the training dataset that could explain this?Best,MattiOn Fri, Jun 5, 2020 at 11:25 PM Seif Mechti <mecht...@gmail.com> wrote:Thank you Dr MattiWe have fixed the errors ...please we need a new reviews on softwares...Thank you ...I have participared in pan 2013 2014 and 2015...Le ven. 5 juin 2020 à 4:22 PM, Matti Wiegmann <matti.w...@uni-weimar.de> a écrit :Hey Seif,Yes, the test-datasets are secret and the results are hidden until the end of the competition. We will publish the results of one evaluator run as the entry to the competition.However, runs on the test-data are equivalent to runs on the training data, so if your software work on the training datasets, it also works the same on the test data. Exceptions are noted on the task pages on pan.webis.de. For celebrity-profiling, for example, there is no celebrity-feeds.ndjson in the test data but only the follower-feeds.ndjson.Best,MattiOn Fri, Jun 5, 2020 at 5:07 PM Seif Mechti <mecht...@gmail.com> wrote:For test corpus, is it normal that we have no result to download and some columns are marked hidden...Le ven. 5 juin 2020 à 11:08 AM, Matti Wiegmann <matti.w...@uni-weimar.de> a écrit :On Fri, Jun 5, 2020 at 11:24 AM Seif Mechti <mecht...@gmail.com> wrote:Dear organizersusername in Tira: Mechti20we are partcipating in the celebrity profiling taskwe want to review our runs?thank you in advance--CordialementSeif MECHTIMIRACL LABORATORYFaculty of Economic Sciences and management of SfaxANLP Research Group
Site web : http://sites.google.com/site/anlprg
Email : mecht...@gmail.com
Matti Wiegmann
Jun 14, 2020, 7:28:01 AM6/14/20
to Seif Mechti, Martin Potthast, pan-works...@googlegroups.com
Hey Seif,
I've unblinded the submissions of all participants by now. You can view them on the Leaderboard on tia.io.
Best,
Matti
Reply all
Reply to author
Forward
0 new messages