nroggendorff commited on
Commit
2d0aeb0
·
verified ·
1 Parent(s): c36a6eb

Update train.py

Browse files
Files changed (1) hide show
  1. train.py +2 -2
train.py CHANGED
@@ -27,8 +27,8 @@ PUSH_TO_HUB = True
27
 
28
  def load_data():
29
  if not INSTRUCT_FINETUNE_BOOL:
30
- dataset = load_dataset(INPUT_DATASET, "cosmopedia-v2", split="train")#, streaming=True)
31
- # dataset = Dataset.from_generator(lambda: dataset.take(int(5e+6)))
32
  # dataset = dataset.shard(num_shards=len(dataset) // SHARD_SIZE, index=INIT)
33
  else:
34
  dataset = load_dataset(INSTRUCT_DATASET, split="train")#, streaming=True)
 
27
 
28
  def load_data():
29
  if not INSTRUCT_FINETUNE_BOOL:
30
+ dataset = load_dataset(INPUT_DATASET, "cosmopedia-v2", split="train", streaming=True)
31
+ dataset = Dataset.from_generator(lambda: dataset.take(int(15e+6)))
32
  # dataset = dataset.shard(num_shards=len(dataset) // SHARD_SIZE, index=INIT)
33
  else:
34
  dataset = load_dataset(INSTRUCT_DATASET, split="train")#, streaming=True)