Prof. Anderson suggested a good update to share for all hands would be benchmarking our BERT model using a simpler classifier such as n-grams or similar, just so we have something reasonable to compare potential output to.
Ideas for benchmarking:
n-grams
what else?
if we want to include in all-hands, should have chosen model + run on chemprot data by Monday, 4/5
Prof. Anderson suggested a good update to share for all hands would be benchmarking our BERT model using a simpler classifier such as n-grams or similar, just so we have something reasonable to compare potential output to.
Ideas for benchmarking:
if we want to include in all-hands, should have chosen model + run on chemprot data by Monday, 4/5