s444415/donut
1
0
Fork 0
Code Issues Pull Requests Projects Releases Wiki Activity

Merge branch 'master' of https://git.wmi.amu.edu.pl/s444415/donut

Browse Source
This commit is contained in:
michal.kozlowski 2023-03-17 14:38:23 +01:00
commit d9cbe1f9a8
1 changed files with 3 additions and 3 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

6
train_stream.py
View File

@ -19,7 +19,7 @@ from torchdata.datapipes.iter import IterableWrapper
import json import json
class TestIterator(IterableWrapper): class CustomWrapperIterator(IterableWrapper):
def __init__(self, iterable, deepcopy=True, total_len=None): def __init__(self, iterable, deepcopy=True, total_len=None):
super().__init__(iterable, deepcopy) super().__init__(iterable, deepcopy)
self.total_len = total_len self.total_len = total_len
@ -181,8 +181,8 @@ def main(config, hug_token):
# train_dataset = train_dataset.with_format('torch') # train_dataset = train_dataset.with_format('torch')
# val_dataset = val_dataset.with_format('torch') # val_dataset = val_dataset.with_format('torch')
train_dataset = TestIterator(train_dataset, total_len=train_length) train_dataset = CustomWrapperIterator(train_dataset, total_len=train_length)
val_dataset = TestIterator(val_dataset, total_len=val_length) val_dataset = CustomWrapperIterator(val_dataset, total_len=val_length)
model.config.pad_token_id = processor.tokenizer.pad_token_id model.config.pad_token_id = processor.tokenizer.pad_token_id
model.config.decoder_start_token_id = processor.tokenizer.convert_tokens_to_ids(['<s_cord-v2>'])[0] model.config.decoder_start_token_id = processor.tokenizer.convert_tokens_to_ids(['<s_cord-v2>'])[0]