Weights
The full set of learned parameters of a model, saved as files. Having the weights means being able to run the model.
Also: model weights, checkpoint
Once a model is trained, its weights are stored as large files: each parameter as a number, often in 16-bit format, so a 70-billion-parameter model is about 140 gigabytes. Whoever holds the weights can run, study and fine-tune the model on their own hardware. Labs that keep weights private offer the model only through an app or an API. A snapshot saved partway through training is called a checkpoint, and some open projects publish hundreds of them so researchers can study how learning unfolds.