nroggendorff commited on
Commit
4bfbe72
·
verified ·
1 Parent(s): bb15100

Update train.py

Browse files
Files changed (1) hide show
  1. train.py +1 -1
train.py CHANGED
@@ -26,7 +26,7 @@ PUSH_TO_HUB = True
26
  def load_data():
27
  if not INSTRUCT_FINETUNE_BOOL:
28
  dataset = load_dataset(INPUT_DATASET, "cosmopedia-v2", split="train", streaming=True)
29
- dataset = Dataset.from_generator(lambda: dataset.take(int(3e+4)))
30
  else:
31
  dataset = load_dataset(INSTRUCT_DATASET, split="train", streaming=True)
32
  dataset = Dataset.from_generator(lambda: dataset.take(int(5e+4)))
 
26
  def load_data():
27
  if not INSTRUCT_FINETUNE_BOOL:
28
  dataset = load_dataset(INPUT_DATASET, "cosmopedia-v2", split="train", streaming=True)
29
+ dataset = Dataset.from_generator(lambda: dataset.take(int(6e+4)))
30
  else:
31
  dataset = load_dataset(INSTRUCT_DATASET, split="train", streaming=True)
32
  dataset = Dataset.from_generator(lambda: dataset.take(int(5e+4)))