Use o Document Manager

Esta página descreve como usar o Document Manager para rotular um novo conjunto de dados e treinar novamente um modelo de ML.

Access and configure Document Manager

Launch the created data labeling session in First Run Experience and go to the settings to configure the OCR.

Choose the OCR you intend to use in the OCR method dropdown menu. For UiPathDocumentOCR, paste the Document Understanding^TM license key (retrieve the Document Understanding API key from the Admin > License page) and then paste the OCR URL you generated when you deployed UiPathDocumentOCR. Check out Deploy UiPathDocumentOCR.

Configure the prelabelling with the models that you have deployed following the instructions describe in Deploy an out-of-the-box ML package. Paste the model public ML Skill endpoint and the Document Understanding license key, and then select Save.

For more details, please check the documentation here: Use a predefined schema.

Importe o documento

1. Select the Import button from a Document Manager Session.
2. Name the dataset and select Browse files to upload.
3. Selecione o documento que deseja carregar.
4. Select YES.

Create extraction fields

Select to create fields to be extracted.

Você pode criar até 40 campos.

For this validation exercise, you can create some common invoice fields such as date, name, invoice-no, and total. Please ensure to change the content type accordingly - date (date), name (string), invoice-no (string), and total (number).

Rotular documentos

Agora, você pode começar a rotular os documentos.

Select the Predict button on top to use the base invoice model to predict the labels for the defined fields, and correct it if the prediction is wrong.

Para alterar o rótulo, arraste o mouse sobre o campo e clique no atalho de teclado para rotulá-lo.

Use a seta na parte superior para mudar para o próximo documento até terminar a validação dos rótulos para todas as faturas carregadas.

Observação:

Since the Invoices base model has already performed really well and the sample invoice is simple without too much variation, the prediction accuracy is close to 100% in this case and you may not need to correct any labels.

Exportar documentos

1. Make sure to select the correct dataset in the dataset filtering and select the Export button .
2. Selecione Exportar.
3. Go to Datasets under the same AI Center project, you should observe the exported training dataset.

Treinamento de um modelo personalizado no AI Center

1. Go to Pipelines > Create new. Please select the evaluation run type, select the model package and the input dataset.
2. Selecione a subpasta em Exportar como o conjunto de dados de entrada.
3. Select Create to start the pipeline. It may take 1-2 hours for the pipeline to run on CPU machines.

Deploy the retrained ML model as an ML Skill

Go to ML Skills and create a new ML Skill.

Escolha o mesmo pacote de modelo de fatura criado anteriormente. Como retreinamos o modelo, agora há uma nova versão secundária de pacote (1 vs 0). Certifique-se de selecionar a mais recente.

Once the ML Skill is created, please go to Modify current deployment to make the ML skill public. Switch the toggle and select Confirm.

Copie o URL da habilidade de ML pública para uso posterior.

Congrats! You have now retrained an Invoice model with your own dataset and created the endpoint to access the model.