Hi @deewuok,
I responded to your question about the input file in the dedicated thread here. Hope it helps!
About the traincommand:
To start with a tip: you can run the command with the -h flag ( for help) to quickly see all the available options. If you do that, you'll see that, indeed, there's no --output option. The location of the output model is the optional positional argument that should be listed first. So your command should be:
python -m prodigy train ./tmp_model --ner food_data en_vectors_web_lg --eval-split 0.2 --paths.init_tok2vec=.\tok2vec_cd8_model289.bin
Also note that the config overrides ( paths.init_tok2vec) in your case, should appear at the end.