nroggendorff commited on
Commit
dcf2dd4
·
verified ·
1 Parent(s): 13b4d78

Update train.py

Browse files
Files changed (1) hide show
  1. train.py +2 -2
train.py CHANGED
@@ -25,9 +25,9 @@ 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(3e+5)))
29
  instruct = load_dataset(INSTRUCT_DATASET, split="train", streaming=True)
30
- instruct = Dataset.from_generator(lambda: instruct.take(int(5e+5)))
31
  dataset_dict = DatasetDict({
32
  'pretrain': pretrain,
33
  'instruct': instruct
 
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(3e+4)))
29
  instruct = load_dataset(INSTRUCT_DATASET, split="train", streaming=True)
30
+ instruct = Dataset.from_generator(lambda: instruct.take(int(5e+4)))
31
  dataset_dict = DatasetDict({
32
  'pretrain': pretrain,
33
  'instruct': instruct