TOPIC AI
Parameters learnable weights in the model
IN 10 SECONDS
The internal variables (weights and biases) of a neural network that are adjusted during training to capture language patterns and mathematical relationships.
GOTCHA More parameters generally improve model performance, but increase memory footprint and response latency requirements exponentially.
HOW PARAMETERS DRIVE PREDICTION
01 Input layer receives token embeddings vector inputs.
02 Weight multiply multiplies inputs by parameter weights across attention layers.
03 Softmax peak generates output probabilities, selecting the next token.
POKE IT YOURSELF
ollama show llama3 --parameters — view parameter counts and sizes
Drill this topic →
~60 sec read