April 7th, 2006

study, words, language

partially fixed...

well, B. and I found a part of the problem from earlier. Now our numbers are in the low twenties instead of the high thirties -- but they're still in the wrong distribution.


We're currently running it all again with much fewer bells-and-whistles ("only one whistle") to see if we can at least get the system to run at baseline level. Running below baseline is a sign of serious bug.

this stinks, since D. is leaving town Sunday very very (very!) early in the morning and I'd really like to have a little time to spend with her.

ah well. The process of building this architecture has been helpful, and if we miss this deadline we may be able to publish some of these results in a different conference. Still, it'd be nice to get it in print sooner.

... but first, results.
  • Current Mood