Data is at the core of your AI block process. Once you have identified your automation need, the next step is to prepare the data. Levity is very flexible with how it allows data uploads.
Labeled or unlabeled?
It is possible to upload both labeled and unlabeled data. At some point, training data needs to get a label but this can be done at multiple points. If you're not sure, just start with your raw data and work from there.
Format & upload
π·Β Image classifier
One by one:Β Click to upload: Common image file formats (JPG, PNG, GIF) Batch:Β CSV file upload: Single file containing one column with URLs pointing to image location
π PDF classifier
One by one:Β Click to upload: Files in PDF format only.Batch:Β CSV file upload: Single file containing one column with URLs pointing to PDF location
π¬ Text classifier
CSV file upload: Single file containing two columns of data - one column with the text to be classified, one column with the label that the model should learn. Header row is optional.
Quantity of training data
For all: More is better!
Images:
At least 20 examples per class. You can get some reasonable results in many cases with 100+ examples.
PDFs:
At least 100 examples per class
Text:
At least 100 examples per class
Questions?
Weβre available to provide advice, answer questions, and talk through setup options whenever youβre ready. Click here to read more about getting in touch.