Parameters

AI fundamentals

Parameters are a model's internal values, adjusted during training, that encode what it has learned. Their number gives the model's scale: a few million for a small specialised model, hundreds of billions for the largest LLMs. More parameters means more capability, but also more cost.

In practice at Gensai

Gensai arbitrates knowingly: a smaller, well-contextualised model often beats a poorly used giant.

Same theme

Artificial intelligence (AI)Algorithm Data Dataset Model (AI)Machine learning Deep learning Neural network Supervised learning Unsupervised learning Reinforcement learning Classification Regression Prediction Training Inference Hyperparameters Bias (AI)