Server error when using MultiLingualTextClassification

Help AI Center
1

Hi,

After adding a MultiLingualTextClassification package & training it using a small dataset with sample data, when I try to use the ML skill from my UiPath Studio I always get the following error message:

{
"code": "InternalServerError",
"message": "Connection error, and we cannot find the requested files in the cached path. Please try again or make sure your Internet connection is on.",
"stacktrace": " File \"/microservice/main.py\", line 49, in predict\n self.model.predict(df)\n File \"/home/aifabric/.local/lib/python3.8/site-packages/torch/autograd/grad_mode.py\", line 27, in decorate_context\n return func(*args, **kwargs)\n File \"<frozen document_understanding.classification.model>\", line 508, in predict\n File \"<frozen document_understanding.classification.dataset>\", line 206, in prepare_df_for_predict\n File \"<frozen document_understanding.classification.dataset>\", line 172, in prepare_df_for_bert\n File \"/home/aifabric/.local/lib/python3.8/site-packages/transformers/tokenization_utils_base.py\", line 1743, in from_pretrained\n resolved_vocab_files[file_id] = cached_path(\n File \"/home/aifabric/.local/lib/python3.8/site-packages/transformers/file_utils.py\", line 1078, in cached_path\n output_path = get_from_cache(\n File \"/home/aifabric/.local/lib/python3.8/site-packages/transformers/file_utils.py\", line 1264, in get_from_cache\n raise ValueError(\nValueError: Connection error, and we cannot find the requested files in the cached path. Please try again or make sure your Internet connection is on.",
"trace_id": null,
"reason": null
}

I have tried deleting the package, re-adding & retraining it but the issue remains the same.

Anyone knows what is going wrong? Am I using wrong data to train it? I find it hard to find a correct example of how to structure the training data.

2

Was the model trained successfully before?

3

Hi Jeremy,

Yes, I completed a Training pipeline successfully before deploying.