Kamel Alrashedy

icon indicating a website link https://kamel773.github.io/

icon location Atlanta, US icon location Ph.D. student at Georgia Tech. Area of research: NLP and Deep Learning

Results 8 comments of Kamel Alrashedy

Instructions to load and use CuBERT pretrained and fine-tuned models

Hi @DNGros , I tried to run cubert on TensorFlow 2, and I am getting the error below. TensorFlow 2 do not support the contrib. Should I run cubert on...

Instructions to load and use CuBERT pretrained and fine-tuned models

Hi @ElchaabiMohamed Yeah, it was a bit complicated to fine tune Cubert. To solve this issue, you should run CuBERT in TensorFlow 1.14. I would recommend to have a look...

Instructions to load and use CuBERT pretrained and fine-tuned models

Hi @ElchaabiMohamed I solved in such a way switching `import tensorflow.compat.v1 as tf` to `import tensorflow as tf`

Cubert: How to get the embedding source code

Hi @alexkarargyris I modify the source code to get the embedding from CuBERT. It was a bit hard to work on TF 1.14. Kamel

ORQA/REALM - Invalid argument: indices[0] = 10311577 is not in

@ioannist I have the same issues. Have you solved it?

ORQA/REALM - Invalid argument: indices[0] = 10311577 is not in

cc @kentonl I am working on fine tuning ORQA in my own dataset, and I have the same issues. We appreciate your help in solving this issues. Kind regards,

Update run_classifier.py

@googlebot I signed it! Any update on my pull request. The [issue](https://github.com/google-research/google-research/issues/571#issuecomment-1160769111)

Update run_classifier.py

Hi [maniatis](https://github.com/maniatis), I opened this pull request to fix a bug in the source code. I also modified the `run_classifier.py` to extract the embedding from CuBERT model and I would...