Increasing the inference speed

Ch4osMy7h / FramenetParser

PyTorch code for "A Graph-Based Neural Model for End-to-End Frame Semantic Parsing" (EMNLP2021)

17 stars 5 forks source link

Increasing the inference speed #2

Open ajaysurya1221 opened 2 years ago

ajaysurya1221 commented 2 years ago

Hi @Ch4osMy7h,

I am doing my research on linguistics and I'm trying to speed up FrameNet. It took me 30 minutes to generate frames for a document with a string length of 84,193. I am trying to do the same for 15,00,000 documents and it's way beyond my lifespan. Do you have any idea how to speed up the inference?