(fixed wrong link)
Show this thread
New code example on http://keras.io , from
@NandanApoorv: using @huggingface Transformers for text extraction on the SQuAD question-answering dataset. Trained on TPU, which brings epoch time from ~1 hour on a K80 to only 5min.https://keras.io/examples/nlp/text_extraction_with_bert/ …
-
-
Thanks. Twitter will use this to make your timeline better. UndoUndo
-
-
-
This is interesting
@fchollet. Since it's on TPU, it can be deployed on a number of edge and iot devices. Thanks . -
I think he referred that it can be trained on a TPU, my guess is that the Bert shown is very computational expensive to run on a edge device e.g: coral edge tpu
End of conversation
New conversation -
-
-
Thanks. Twitter will use this to make your timeline better. UndoUndo
-
-
-
Beautiful written code
Thanks. Twitter will use this to make your timeline better. UndoUndo
-
-
-
This is really good work! It’d be great if we can have some numbers (f1 score) in the evaluation step for reproducibility
-
oor full ah theditu iruka pola compare panna - Show replies
New conversation -
-
-
The way code is written
! Beautiful, concise and clean!Thanks. Twitter will use this to make your timeline better. UndoUndo
-
-
-
This is so clean and concise. +1
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.