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

Making TF GPT2 compliant with XLA and AMP (#10230)

Browse Source 
* Fix XLA and AMP

* Fix AMP and XLA

* Apply style

* Apply Patrick's comment
This commit is contained in:
Julien Plu
2021-02-18 09:36:01 +01:00
committed by GitHub
parent 5da7c78ed8
commit bdf1669e3f
3 changed files with 25 additions and 138 deletions
Download Patch File Download Diff File Expand all files Collapse all files

8
tests/test_modeling_tf_gpt2.py
View File

@@ -389,14 +389,6 @@ class TFGPT2ModelTest(TFModelTesterMixin, unittest.TestCase):
config_and_inputs = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_gpt2_for_sequence_classification(*config_and_inputs)
def test_mixed_precision(self):
# TODO JP: Make GPT2 float16 compliant
pass
def test_xla_mode(self):
# TODO JP: Make GPT2 XLA compliant
pass
@slow
def test_model_from_pretrained(self):
for model_name in TF_GPT2_PRETRAINED_MODEL_ARCHIVE_LIST[:1]: