The data. Also tried phenotype-specific training.pic.twitter.com/33jiJ8CXyX
Prikaži ovu nit
Dmitriy Dligach is exploring various pre-training representations and models. The goal is to learn a universal text encoder (sounds simple, right?). @TMills #AMIA2019pic.twitter.com/WDc9Zg3dhq
-
-
-
Evaluating on i2b2 2008 and some internal substance misuse data.pic.twitter.com/yItaZpBDbx
Prikaži ovu nit -
Results on i2b2 2008. Model combining the dense (deep averaging network) plus sparse (bag of CUIs) works best.pic.twitter.com/JltkXMlhAo
Prikaži ovu nit
Kraj razgovora
Novi razgovor -
-
-
Outcome prediction from text/notes?
-
Outcome extraction might be a more accurate description. Not a trivial task!
- Još 4 druga odgovora
Novi razgovor -
Čini se da učitavanje traje već neko vrijeme.
Twitter je možda preopterećen ili ima kratkotrajnih poteškoća u radu. Pokušajte ponovno ili potražite dodatne informacije u odjeljku Status Twittera.