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

device agnostic models testing (#27146)

Browse Source 
* device agnostic models testing

* add decorator `require_torch_fp16`

* make style

* apply review suggestion

* Oops, the fp16 decorator was misused
This commit is contained in:
Hz, Ji
2023-11-01 01:12:14 +08:00
committed by GitHub
parent 77930f8a01
commit 50378cbf6c
51 changed files with 369 additions and 154 deletions
Download Patch File Download Diff File Expand all files Collapse all files

5
tests/models/opt/test_modeling_opt.py
View File

@@ -22,7 +22,7 @@ import unittest
import timeout_decorator # noqa
from transformers import OPTConfig, is_torch_available
from transformers.testing_utils import require_torch, require_torch_fp16, require_torch_gpu, slow, torch_device
from transformers.testing_utils import require_torch, require_torch_accelerator, require_torch_fp16, slow, torch_device
from ...generation.test_utils import GenerationTesterMixin
from ...test_configuration_common import ConfigTester
@@ -514,7 +514,8 @@ class OPTGenerationTest(unittest.TestCase):
self.assertListEqual(predicted_outputs, EXPECTED_OUTPUTS)
@require_torch_gpu
@require_torch_accelerator
@require_torch_fp16
def test_batched_nan_fp16(self):
# a bug manifested starting at models facebook/opt-1.3 and larger when running batched generations,
# therefore not using a tiny model, but the smallest model the problem was seen with which is opt-1.3b.