How To Train your Chatbot
Training sources
In order for you chatbot to be able to help your clients you need to train the chatbot with the knowledge about your business. You can do this by pointing the chatbot to your website, provide the chatbot with files (PDF, DOC, XLS, CSV, TXT) and build sets of questions and answers.
- Train the chatbot with your website
- Train the chatbot with files
- Train the chatbot with questions and answers
- Train the chatbot with text
You can train it with any combination of the sources. Each training material has different amount of training characters… & training links
Training limits - What is Training “Character”
A single character corresponds to one token of text. We don’t store entire websites or documents; instead, we focus exclusively on counting pure text. Typically, when images and other media are included, a page's actual text makes up only about 8% of the total size.
For example:
- To extract 400,000 characters, you’d need to scrape approximately a 5 MB page.
- To extract 11,000,000 characters, the required page size would be around 140 MB.
- For 15,000,000 characters, the necessary page size would be around 190 MB.
If these limits aren’t sufficient for your needs, you can purchase an add-on package: “+10 MB of training characters for all your chatbots.” This package increases each chatbot’s limit from, for instance, 15 MB to 25 MB, providing extra capacity for additional text.
Re-training
You can schedule automatic re-training for your website sources. The chatbot will visit the websites when scheduled and refresh the training data. Read more here:
Re-training