Closed tomcobley closed 1 year ago
Hi,
Sorry, the TEXT_MODEL_PATH
seems a redundant parameter in this script. The params of OntoModel are randomly initialized in the pretraining by default when GO_ENCODER_CLS
is set to embedding
, and you could modify this path to load the param of PubMedBERT
if you set the GO_ENCODER_CLS
to bert
.
Hi!
I am trying to pretrain the model using the datasets and pretrained models mentioned in the README and the
run_pretrain.sh
script.However, I have run into a few problems which seem to be due to my choice of
TEXT_MODEL_PATH
inrun_pretrain.sh
on this line.What should this model be?
I assumed this meant the PubMedBERT model mentioned in the README, but execution fails if I use this - am I missing something?
Many thanks in advance!