We don't use TF but are also releasing two neural parsers: https://github.com/elikip/bist-parser … 94.4UAS at 800words/sec or 93.8 at 1200. On a Macbook.
-
would be curious to know how to replicate it in keras, btw. it might be pushing the system a bit.
if all you're using are bi-LSTM and CNNs, it should not be an issue... are you using anything out of the ordinary?
-
-
it's biLSTMs, but a dynamic network structure on top of that.
Thanks. Twitter will use this to make your timeline better. UndoUndo
-
-
-
Complication is you have to make new input tensor each word, so constant data copying. https://github.com/elikip/bist-parser/blob/master/barchybrid/src/arc_hybrid.py#L106 …
Thanks. Twitter will use this to make your timeline better. UndoUndo
-
Loading seems to be taking a while.
Twitter may be over capacity or experiencing a momentary hiccup. Try again or visit Twitter Status for more information.