Matteo Grella

icon indicating a website link https://scholar.google.com/citations?user=n7x6nV0AAAAJ&hl=en icon indicating an email [email protected]

icon company Head of Research and AI @ Crisis24 (part of GardaWorld) icon location Italy

Results 42 comments of Matteo Grella

Add support for array type

Hey @sbinet do you see any cons on merging #9 as it is?

Add support for pickling object

@sbinet would you be interested to help on this?

Failed to serialize models/vblagoje/bart_lfqa/spago_model.bin

Thank you, I am on it

Failed to serialize models/vblagoje/bart_lfqa/spago_model.bin

Hello @ior308 can you update to the latest version and give it a try? Thank you!

WordPieceTokenizer in Chinese can not use.

Thanks @sunhailin-Leo we understood the issue and we are going to fix it asap

cybertron's moka-ai/m3e-base model is inconsistent with Python

Thanks, I’m checking. Back to you asap.

Slow speed in obtaining results.

What is your OS and architecture (amd64 vs arm64)?

Slow speed in obtaining results.

Hello @akaler727 would you mind to update to the latest version and give it a try?

Size Paragraph

Hi @devalexandre yes we are constrained by the positional encoder max 512 token (tokens /= words). Please update to the latest version and give it a try, it should log...

Adding support for Distilbert

Thank you! What differs DistilBERT from BERT?