nroggendorff commited on
Commit
e651924
·
verified ·
1 Parent(s): 12f0b8b

Update train.py

Browse files
Files changed (1) hide show
  1. train.py +1 -1
train.py CHANGED
@@ -25,7 +25,7 @@ PUSH_TO_HUB = True
25
 
26
  def load_data():
27
  pretrain = load_dataset(INPUT_DATASET, "cosmopedia-v2", split="train", streaming=True)
28
- pretrain = Dataset.from_generator(lambda: pretrain.take(int(2e+4)))
29
  instruct = load_dataset(INSTRUCT_DATASET, split="train", streaming=True)
30
  instruct = Dataset.from_generator(lambda: instruct.take(int(8e+4)))
31
  dataset_dict = DatasetDict({
 
25
 
26
  def load_data():
27
  pretrain = load_dataset(INPUT_DATASET, "cosmopedia-v2", split="train", streaming=True)
28
+ pretrain = Dataset.from_generator(lambda: pretrain.take(int(2e+5)))
29
  instruct = load_dataset(INSTRUCT_DATASET, split="train", streaming=True)
30
  instruct = Dataset.from_generator(lambda: instruct.take(int(8e+4)))
31
  dataset_dict = DatasetDict({