Base model
A pretrained checkpoint you fine-tune on top of.
Glossary
AI and fine-tuning terms, in plain language
A quick reference for the words you'll run into across Foundry — no PhD required.
Core terms
Fine-tuning
Adapting a base model to your data and task.
Dataset
The examples (often JSONL) used to train a model.
Training job
A run that produces a new named model.
Named model
Your fine-tuned model destination you can call.
Workspace
A job-shaped UI opened after training completes.
Credits
Metered units spent on fine-tunes and chat.
Inference
Running a model to generate a response.
More terms
Epoch
One full pass over the training dataset.
Loss
A number that drops as the model learns.
Token
A chunk of text the model reads and writes.
Prompt
The input you give a model at inference time.
System prompt
Instructions that steer a model's behavior.
Glossary injection
Feeding term mappings to guide translation.
Want the full docs?
Every term links back to detailed documentation.