AI Quiz
Check which claims the examples actually support.
How it works
Before running this reference answer, decide whether a low training loss proves generalisation, whether attention alone is a complete transformer, whether a RAG source guarantees a faithful answer, and whether a browser chart counts as model evaluation. Each false claim confuses a useful component with evidence for a broader capability.
Run the code
Try in playground loads this complete standard-library Python example without executing it. Choose Run when ready. Local review uses Python 3.12.14; the existing browser editor uses Pyodide 0.27.7. These small fictional examples need no account, external data, model-provider key or database service.
Example
answers = {
"training loss proves generalisation":False,
"attention head is a complete transformer":False,
"retrieval guarantees faithful generation":False,
"a chart replaces held-out evaluation":False,
}
for statement,answer in answers.items():
print(statement+": "+str(answer))
Expected result
training loss proves generalisation: False attention head is a complete transformer: False retrieval guarantees faithful generation: False a chart replaces held-out evaluation: False
For each statement, name the missing test or component. Training needs independent evaluation; a transformer needs more architecture and learning; RAG needs generation checks and permissions; a plot needs a sound sampling and metric design. These distinctions prevent exaggerated capability claims.