The SpanTransformer a language model calls after
generate_text / generate_object, and when a stream_text
stream ends. The one in context at the call wins over the one in
context when the model was made.
The SpanTransformer a language model calls after
generate_text / generate_object, and when a stream_text
stream ends. The one in context at the call wins over the one in
context when the model was made.