Spaces:
Runtime error
Runtime error
Update train.py
Browse files
train.py
CHANGED
@@ -25,8 +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(
|
29 |
-
instruct = load_dataset(INSTRUCT_DATASET, split="train")
|
|
|
30 |
dataset_dict = DatasetDict({
|
31 |
'pretrain': pretrain,
|
32 |
'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(50000))
|
29 |
+
instruct = load_dataset(INSTRUCT_DATASET, split="train", streaming=True)
|
30 |
+
instruct = Dataset.from_generator(lambda: instruct.take(20000))
|
31 |
dataset_dict = DatasetDict({
|
32 |
'pretrain': pretrain,
|
33 |
'instruct': instruct
|