Parameters
The numerical values inside an AI model that are adjusted during training to encode what it has learned.
Parameters are the numbers inside a neural network that get tuned during training. When you hear "a 70 billion parameter model", it means the model has 70 billion individual numerical values that were adjusted across millions of training examples until the model's outputs became accurate. More parameters generally means more capacity to learn complex patterns.
Think of a mixing board in a recording studio — hundreds of knobs and sliders, each controlling a small aspect of the sound. Training an AI is like having an automated system turn all those knobs simultaneously, billions of times, until the output sounds exactly right. The final positions of all those knobs are the parameters.
More parameters doesn't automatically mean a better or smarter model. A smaller model trained on better data with better techniques can outperform a much larger one. Parameter count is a measure of capacity, not quality — and larger models are significantly more expensive to run.