nroggendorff commited on
Commit
6c32e51
·
verified ·
1 Parent(s): 6ced1a5

Update train.py

Browse files
Files changed (1) hide show
  1. train.py +1 -1
train.py CHANGED
@@ -24,7 +24,7 @@ PUSH_TO_HUB = True
24
 
25
  def load_data():
26
  pretrain = load_dataset(INPUT_DATASET, "cosmopedia-v2", split="train", streaming=True)
27
- pretrain = Dataset.from_generator(lambda: pretrain.take(int(3e+5)))
28
  instruct = load_dataset(INSTRUCT_DATASET, split="train", streaming=True)
29
  instruct = Dataset.from_generator(lambda: instruct.take(int(5e+5)))
30
  dataset_dict = DatasetDict({
 
24
 
25
  def load_data():
26
  pretrain = load_dataset(INPUT_DATASET, "cosmopedia-v2", split="train", streaming=True)
27
+ pretrain = Dataset.from_generator(lambda: pretrain.take(int(3e+6)))
28
  instruct = load_dataset(INSTRUCT_DATASET, split="train", streaming=True)
29
  instruct = Dataset.from_generator(lambda: instruct.take(int(5e+5)))
30
  dataset_dict = DatasetDict({