nroggendorff commited on
Commit
1848b96
·
verified ·
1 Parent(s): b1e38e3

Update train.py

Browse files
Files changed (1) hide show
  1. train.py +2 -2
train.py CHANGED
@@ -24,9 +24,9 @@ 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(6e+4)))
28
  instruct = load_dataset(INSTRUCT_DATASET, split="train", streaming=True)
29
- instruct = Dataset.from_generator(lambda: instruct.take(int(8e+4)))
30
  dataset_dict = DatasetDict({
31
  'pretrain': pretrain,
32
  'instruct': instruct
 
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({
31
  'pretrain': pretrain,
32
  'instruct': instruct