Update synthetic.py

This commit is contained in:
Daniel Han 2025-04-30 08:49:04 -07:00
commit a7e6d72c60

View file

@ -224,8 +224,8 @@ class SyntheticDataKit:
):
assert(hasattr(self, "model_name"))
assert(hasattr(self, "max_seq_length"))
assert(hasattr(max_generation_tokens < self.max_seq_length))
assert(max_generation_tokens < self.max_seq_length)
locations = "pdf,html,youtube,docx,ppt,txt,output,generated,cleaned,final"
locations = locations.split(",")
for path in locations: