So i trained a model with jsonl in this format
{"input_text": "bio: TXT", "output_text": "LABEL"}
The model works well in the test, especially if the prompt includes the context and the delimitation of the labels. But all my batch requests have internal errors.
What is the structure and format I should use to get a jsonl with the label prediction? The input is just one column with lines of plain/text that should be classified into labels, also how do I include the instruction in the format?
Please I will gladly appreciate any help, especially if someone can give me the appropriate jsonl format to get my output.