Error when running step (9) in fine_tuning function!

[abdelrahmaan]

Get this error
RuntimeError: Expected tensor for argument #1 'indices' to have one of the following scalar types: Long, Int; but got torch.cuda.FloatTensor instead (while checking arguments for embedding)

[elmadany]

Hi @abdelrahmaan

Please provide the full error.

[abdelrahmaan]

Hi @elmadany
Here you are.

[INFO] step (9) start fine_tuning
Epoch: 0%| | 0/5 [00:00<?, ?it/s]

RuntimeError Traceback (most recent call last)
in ()
----> 1 report_df = fine_tuning(config)

10 frames
/usr/local/lib/python3.7/dist-packages/torch/nn/functional.py in embedding(input, weight, padding_idx, max_norm, norm_type, scale_grad_by_freq, sparse)
2042 # remove once script supports set_grad_enabled
2043 no_grad_embedding_renorm(weight, input, max_norm, norm_type)
-> 2044 return torch.embedding(weight, input, padding_idx, scale_grad_by_freq, sparse)
2045
2046

RuntimeError: Expected tensor for argument #1 'indices' to have one of the following scalar types: Long, Int; but got torch.cuda.FloatTensor instead (while checking arguments for embedding)

[elmadany]

This error seems due to the torch version and the training code not being updated.
I recommend using the Huggingface trainer. These an examples,
https://github.com/huggingface/transformers/tree/master/notebooks
First, select the trainer code based on your task. Then use "UBC-NLP/MARBERT" as the model name.
I hope these will help you

[nboudad]

I get the same error!!
@abdelrahmaan did you find a solution?

[abdelrahmaan]

I did MARABERT by k-train
Try to use it.

[nboudad]

Thanks @abdelrahmaan
For me, I fixed the issue using:
!pip install GPUtil pytorch_pretrained_bert transformers==4.12.2 sentencepiece==0.1.96

[Laratomeh]

Hello @elmadany
Can you please just tell me how to print the classification report in this code? Actually, it is urgent manner
https://colab.research.google.com/drive/1M0ls7EPUi1dwqIDh6HNfJ5y826XvcgGX?usp=sharing
because I want to know the f1 score for each label