csv format examples , error ✘ Invalid task format for view ID 'ner_manual'

Running the following commands

prodigy ner.manual ner_data2 blank:en ${xseed}/prodigy/test/data2.csv --label DISEASE,DRUG
prodigy ner.manual ner_data1 blank:en ${xseed}/prodigy/test/data1.csv --label DISEASE,DRUG
prodigy ner.manual ner_data0 blank:en ${xseed}/prodigy/test/data0.csv --label DISEASE,DRUG

The following example csv file fails

cat data2.csv
This is a sentence.,POSITIVE,0
This is another sentence.,NEGATIVE,0.1

when the browser starts up.

✘ Invalid task format for view ID 'ner_manual'

meta   value is not a valid dict

{'text': 'This is a sentence.', 'label': 'POSITIVE', 'meta': '0', '_input_hash': 1061759252, '_task_hash': -1161079259, '_is_binary': False, 'tokens': [{'text': 'This', 'start': 0, 'end': 4, 'id': 0, 'ws': True}, {'text': 'is', 'start': 5, 'end': 7, 'id': 1, 'ws': True}, {'text': 'a', 'start': 8, 'end': 9, 'id': 2, 'ws': True}, {'text': 'sentence', 'start': 10, 'end': 18, 'id': 3, 'ws': False}, {'text': '.', 'start': 18, 'end': 19, 'id': 4, 'ws': False}]}

NO ISSSUES with data1.csv

cat data1.csv
This is a sentence.,POSITIVE,
This is another sentence.,NEGATIVE,

NO ISSSUES with data0.csv

This is a sentence.,,
This is another sentence.,,

What are valid values for Label and Meta? I am doing NER.

Could you put some examples of valid csv files here
prodigy-recipes/example-datasets at master · explosion/prodigy-recipes (github.com)

Hi and sorry, it looks like we introduced an incompatibility in the CSV loder in v1.11. We'll fix this for the next release. The "meta" in the final data should definitely always be a dict, but because CSV files can't easily represent nested data like this, Prodigy should support a meta column that then becomes {"meta": {"meta": "whatever"}}. We'll fix this for the next release!

In the meantime, you could leave out the Meta if you don't need it (you can just remove the column all together and it won't matter). Alternatively, you can also provide JSON instead.

Thanks for clearing that up.

Update: Just released v1.11.3, which should fix the underlying issue!