Hi,
I am manually correcting 10000 records but my session has been closed due to the rebooting the machine. Is there any way to re-open the session exactly where i left (i was at 2478th document using ner.correct)?. If i open the session again, does it shuffle the documents and start again?
Hi! By default, Prodigy won't shuffle your data and the manual recipes like ner.correct will stream in examples in order, so when you restart the server, examples that are already in the dataset will be skipped and you can start annotating again where you left off.
If you're at example 2478, this means the 2478th line in your source data or the 2478th sentence, depending on whether you have sentence segmentation enabled or disabled. So if you know the exact number, you could also use an edited source file that only includes the examples starting with that row. It shouldn't make a big difference, but it means you don't have to re-load the already annotated examples just to skip them.
thanks. sorry its a 2478th json document in the jsonl file. Its for NER tagging