Manual Training

What is Manual Training in Docsumo?
Manual Training allows users to directly control the training process of AI models by selecting data, adjusting parameters, and providing feedback. Learn more here.

How do I start Manual Training in Docsumo?
Manual Training can be started by accessing the training dashboard and selecting the manual training option. Step-by-step instructions are available here.

What types of documents are best suited for Manual Training?
Documents that require high accuracy or have complex layouts are best suited for Manual Training, where user input can guide the training process. More details are provided here.

How do I select the data for Manual Training?
Data can be selected based on document types, formats, and specific fields that need extraction. Selecting relevant data is crucial for effective training. Learn how to select data here.

Can I adjust the learning parameters during Manual Training?
Yes, you can adjust parameters like learning rate, batch size, and epochs to optimize the training process according to your needs. Adjustment tips are available here.

How do I monitor the progress of Manual Training?
Progress can be monitored through the training dashboard, which provides real-time updates on the model’s performance and learning curve. Monitoring tools are outlined here.

What are the benefits of using Manual Training?
Benefits include greater control over the training process, the ability to handle complex documents, and more accurate data extraction. Learn more about the benefits here.

Can I combine Manual Training with Automatic Training?
Yes, combining both methods allows for the flexibility of automatic improvements with the precision of manual adjustments. Learn how to combine them here.

What should I do if the model does not perform well during Manual Training?
If the model underperforms, consider adjusting the training data, refining the field mappings, or tweaking the learning parameters. Troubleshooting tips are provided here.

How do I provide feedback during Manual Training?
Feedback can be provided by reviewing extracted data and correcting errors, which helps the model learn and improve its accuracy. Feedback instructions are available here.

How many documents are needed for effective Manual Training?
Typically, a diverse set of 20-30 documents per type is recommended for effective Manual Training, but this can vary based on complexity. Document quantity guidelines are provided here.

How long does Manual Training take?
The duration depends on the complexity of the document and the amount of data being used. Manual Training can take from a few hours to several days. More details are provided here.

Can I stop or pause Manual Training?
Yes, you can stop or pause training at any time and resume later without losing progress. Instructions for pausing or stopping are available here.

What happens if I make a mistake during Manual Training?
Mistakes can be corrected by reprocessing the document, adjusting the field mappings, or reverting to a previous model version. Correction steps are provided here.

How do I ensure consistent training results?
Consistency can be achieved by using a standardized set of training documents and applying uniform parameters across training sessions. Consistency tips are outlined here.

Can I use Manual Training for multiple document types?
Yes, Manual Training can be applied to multiple document types, allowing you to train models for diverse needs. Learn how to manage multiple types here.

How do I evaluate the success of Manual Training?
Success can be evaluated by testing the model on a validation set and reviewing the accuracy of the extracted data. Evaluation steps are provided here.

What are the challenges of Manual Training?
Challenges include the time commitment, the need for precise data, and the complexity of managing the training process. Overcoming challenges is discussed here.

Can I use pre-labeled data for Manual Training?
Yes, pre-labeled data can be used to accelerate the training process and improve the accuracy of the model. Learn more about using labeled data here.

How do I handle errors during Manual Training?
Errors can be handled by reviewing error logs, adjusting training parameters, and refining the dataset. Error handling tips are provided here.