This is a chatbot I made using a backoff model. I took s AI course. So thanks to for teaching me. This time I created all of code myself. You can now also change the creativity. I made the training data using myself and chatgpt. Chatgpt expanded the training data I made so as to create similar responses. I added some things since the pre-alpha version: 1. Very basic memory. It keeps the previous things you and the chatbot said in the input. 2. Spell check 3. Punctuation can throw off the chatbot. Hello. and Hello? are 2 very different things. Now I have made it so when the chatbot sees any punctuation, even in the training data, it will all be a period. And the chatbot still writes punctuation. 4. The training data contains small conversations between the chatbot and the user. A problem is that it can't tell where the conversation ends and where another begins. It can either pick things to say from unrelated sentences or pick the nothingness at the end of each conversation in the training data. I have fixed it so that it now does not say nothing and always picks characters from the same conversation in the training data instead of from the end of one conversation to the beginning of the next one.
