About the usage of this dataset in SFT training
#1
by
ChengeHarrison
- opened
Is the dataset inclued in FreedomIntelligence/phoenix-sft-data-v1? Can we train a SFT model using these two datasets?
Another question, is the original data(before translation) of the dataset included in FreedomIntelligence/phoenix-sft-data-v1?
This is the Chinese translation of ShareGPT, and it is not used in the Phoenix-sft-data. I think you can use both of them, although it may partially overlap with the Chinese samples of Phoenix-sft-data due to the sampling strategy in Phoenix-sft-data.
For the second question, the original data is Share-GPT, which is included in the Phoenix-sft-data.
Thank you!
liuxuan320
changed discussion status to
closed