I have been doing the image generation with the leonardo AI. And I have listed variety of the problems that we have with the dataset and the models. And I have learned that the LLMs that we train and work with are lot different these days. We just don't get how the app building works with if we are not exposed with the LLM.
So how does the LLM works? How does the models make what we train them to do? This is where you have to teach them how to do the LLM and also we have to learn to make the most out of such models when we are planning on building the apps.
First step is creating the model. Like say you want to count the cats in an image. You have to create a model that recognizes the cats in an image. And then another part of the code would be making sure to count the cat appearing in the model.
And then second step would be making sure that you would be training the model. Because it would take some iterations for the LLM to work and then it would produce the results for you. This is what we should be making the result wise update there. I have built some really good apps using the trained models there.
Open source models are also a good option for you to learn from. Learn from the people who already made the LLM and the dataset before you. And then learn how to train the models this way you would learn what to do and what not to do with those models. Training such models takes forever and also would be a good thing for app development.
You can also use the existing models through the API which is what you would learn. I think API based AI models can be costly but they can be a good learning point. I have not yet reached this level because it would be costing me a good amount and that is what I am not aiming for here as well. That's what I am planning to do better.
Learning AI and machine learning is ongoing process. As you learn more you would code more and also learn to take advantage of existing tools. Something I have to learn and also show you here on Hive. I'll be covering this each week one small tutorial at a time.
