Is there a more detailed analysis of seed tasks?

[canghongjian]

Why set the number of seed tasks to 175? How did the number of seed tasks affect the final results including the quality of generated instructions and the performance of instructions-tuned model?
I have considered generating more domain-specified instructions recently. The number of seed tasks should be smaller and the content (or format) should be more in line. I wonder is there any I should notice if I craft the seed tasks set myself, for example the number and the content. And do you think models tuned by the domian-specified instructions will do better in the specified domain?
Thanks a lot, wish you a good day :).

[ddzipp]

hey!I have met the same problem with you. Have you ever solved this and generated more specific domain data ?