Some questions about fine-tuning from fine-tuned model

when i try to fine-tuning from a fine-tuned model, i found it will created a new model ,and this model will cover my first fine-tuning`s example.
this situation is nomal or i used wrong method param ?
the old model is based on curie
my fine-tuned method param:
“training_file”: “file-sXSA8Rq3ooxX9r7rwz4zPMkn”,