Monday, October 3, 2016: 9:54am – 2:06pm (4.2 hours):
Most weeks my coursework is due later in the week. This week the coursework was due in the beginning of the week, and I was not feeling great on Monday. My hours at lab this week will reflect both irregularities.
I wrote the code for running Tool.jar on the files to create posEx.txt and negEx.txt. Automating this process by running a few things in parallel seemed to cause some pretty extreme memory problems though. My final commit for the day included the phrase: “DO NOT RUN runthis.sh RIGHT NOW! HAS SOME REALLY BAD ERRORS in the parallelization step for Tool.jar.” Needless to say, I still had my work cut out.
Since I missed the majority of hackathon I will work on that project later in the week.
Tuesday, October 4, 2016:
I was feeling better on Tuesday, but spent the time I usually attend Professor’s class checking my code. I had a code review for Search Informatics and an assignment due.
Wednesday, October 5, 2016: 9:42am – 5:36pm (7.9 hours):
Kaushik and I started the pipeline, encountering an error when we actually tried to run RDN-boost on it. It turned out that generatePrimaryShares.py was adding the complete file name (including the .txt extension) to the NEWposEx.txt file it generated. When that was resolved we combined the posEx.txt created by Tool.jar for docs500 and around 700 examples that we created from the previously missed examples.
The training step alone took around 5 hours. While waiting, I worked on runthis.sh, met Professor’s guest with the rest of the lab, and played a game of cricket since Professor’s stumps arrived in the mail.
Testing should be quicker, but I ran out of disk space on the server. The log file generated during training (NEWtrainOutput.txt) was around 273mb, I’ll need to decide whether this is worth keeping. I emailed Homa and Patrick with our updates, I’ll pick up here on Friday.
Thursday, October 6, 2016: 12:30pm – 2:00pm (1.5 hours)
Mayukh passed his qualifiers and Professor was running in between meetings, so reading group was pushed a bit beyond what I could attend today. The team met to discuss what Mayukh learned (since others were hopefully taking qualifiers soon).
Friday, October 7, 2016: 8:10am – 3:00pm (6.8 hours)
Busy day in the lab, did more updates to runthis.sh.