SUMIN/HuggingFace_transformer
1
0
Fork 0
Code Issues Pull Requests Actions 7 Packages Projects Releases Wiki Activity

fix Trainer.train(resume_from_checkpoint=False) is causing an exception (#12981)

Browse Source 
* fix #12970

* Update tests/test_trainer.py

Co-authored-by: Sylvain Gugger <35901082+sgugger@users.noreply.github.com>

* Update tests/test_trainer.py

Co-authored-by: Sylvain Gugger <35901082+sgugger@users.noreply.github.com>

* Update tests/test_trainer.py

Co-authored-by: Sylvain Gugger <35901082+sgugger@users.noreply.github.com>

* remove unnecessary issue link

* fix test formatting

Co-authored-by: Sylvain Gugger <35901082+sgugger@users.noreply.github.com>
This commit is contained in:
Philip May
2021-08-03 10:10:33 +02:00
committed by GitHub
parent 790f1c9545
commit b7439675b8
2 changed files with 15 additions and 0 deletions
Download Patch File Download Diff File Expand all files Collapse all files

14
tests/test_trainer.py
View File

@@ -827,6 +827,20 @@ class TrainerIntegrationTest(TestCasePlus, TrainerIntegrationCommon):
self.assertAlmostEqual(a, a1, delta=1e-8)
self.assertAlmostEqual(b, b1, delta=1e-8)
# regression for this issue: https://github.com/huggingface/transformers/issues/12970
def test_training_with_resume_from_checkpoint_flase(self):
train_dataset = RegressionDataset(length=128)
eval_dataset = RegressionDataset()
config = RegressionModelConfig(a=0, b=2)
model = RegressionRandomPreTrainedModel(config)
tmp_dir = self.get_auto_remove_tmp_dir()
args = RegressionTrainingArguments(tmp_dir, save_steps=5, learning_rate=0.1)
trainer = Trainer(model, args, train_dataset=train_dataset, eval_dataset=eval_dataset)
trainer.train(resume_from_checkpoint=False)
@require_torch_up_to_2_gpus
def test_resume_training_with_gradient_accumulation(self):
# This test will fail for more than 2 GPUs since the batch size will get bigger and with the number of