Byebye torch 1.10 (#28207)
* fix * fix --------- Co-authored-by: ydshieh <ydshieh@users.noreply.github.com>
This commit is contained in:
Yih-Dar
GitHub
parent
e768616afa
commit
59cd9de39d
|
|
@ -15,7 +15,7 @@ jobs:
|
|||
strategy:
|
||||
fail-fast: false
|
||||
matrix:
|
||||
version: ["1.13", "1.12", "1.11", "1.10"]
|
||||
version: ["1.13", "1.12", "1.11"]
|
||||
runs-on: ubuntu-22.04
|
||||
steps:
|
||||
-
|
||||
|
|
|
|||
|
|
@ -56,21 +56,10 @@ jobs:
|
|||
sha: ${{ github.sha }}
|
||||
secrets: inherit
|
||||
|
||||
run_past_ci_pytorch_1-10:
|
||||
name: PyTorch 1.10
|
||||
if: (cancelled() != true) && ((github.event_name == 'schedule') || ((github.event_name == 'push') && startsWith(github.ref_name, 'run_past_ci')))
|
||||
needs: [run_past_ci_pytorch_1-11]
|
||||
uses: ./.github/workflows/self-past.yml
|
||||
with:
|
||||
framework: pytorch
|
||||
version: "1.10"
|
||||
sha: ${{ github.sha }}
|
||||
secrets: inherit
|
||||
|
||||
run_past_ci_tensorflow_2-11:
|
||||
name: TensorFlow 2.11
|
||||
if: (cancelled() != true) && ((github.event_name == 'push') && startsWith(github.ref_name, 'run_past_ci'))
|
||||
needs: [run_past_ci_pytorch_1-10]
|
||||
needs: [run_past_ci_pytorch_1-11]
|
||||
uses: ./.github/workflows/self-past.yml
|
||||
with:
|
||||
framework: tensorflow
|
||||
|
|
|
|||
|
|
@ -250,7 +250,7 @@ The model itself is a regular [Pytorch `nn.Module`](https://pytorch.org/docs/sta
|
|||
|
||||
### With pip
|
||||
|
||||
This repository is tested on Python 3.8+, Flax 0.4.1+, PyTorch 1.10+, and TensorFlow 2.6+.
|
||||
This repository is tested on Python 3.8+, Flax 0.4.1+, PyTorch 1.11+, and TensorFlow 2.6+.
|
||||
|
||||
You should install 🤗 Transformers in a [virtual environment](https://docs.python.org/3/library/venv.html). If you're unfamiliar with Python virtual environments, check out the [user guide](https://packaging.python.org/guides/installing-using-pip-and-virtual-environments/).
|
||||
|
||||
|
|
|
|||
|
|
@ -225,7 +225,7 @@ El modelo en si es un [Pytorch `nn.Module`](https://pytorch.org/docs/stable/nn.h
|
|||
|
||||
### Con pip
|
||||
|
||||
Este repositorio está probado en Python 3.8+, Flax 0.4.1+, PyTorch 1.10+ y TensorFlow 2.6+.
|
||||
Este repositorio está probado en Python 3.8+, Flax 0.4.1+, PyTorch 1.11+ y TensorFlow 2.6+.
|
||||
|
||||
Deberías instalar 🤗 Transformers en un [ambiente virtual](https://docs.python.org/3/library/venv.html). Si no estas familiarizado con los entornos virtuales de Python, consulta la [guía de usuario](https://packaging.python.org/guides/installing-using-pip-and-virtual-environments/).
|
||||
|
||||
|
|
|
|||
|
|
@ -201,7 +201,7 @@ checkpoint: जाँच बिंदु
|
|||
|
||||
### पिप का उपयोग करना
|
||||
|
||||
इस रिपॉजिटरी का परीक्षण Python 3.8+, Flax 0.4.1+, PyTorch 1.10+ और TensorFlow 2.6+ के तहत किया गया है।
|
||||
इस रिपॉजिटरी का परीक्षण Python 3.8+, Flax 0.4.1+, PyTorch 1.11+ और TensorFlow 2.6+ के तहत किया गया है।
|
||||
|
||||
आप [वर्चुअल एनवायरनमेंट](https://docs.python.org/3/library/venv.html) में 🤗 ट्रांसफॉर्मर इंस्टॉल कर सकते हैं। यदि आप अभी तक पायथन के वर्चुअल एनवायरनमेंट से परिचित नहीं हैं, तो कृपया इसे [उपयोगकर्ता निर्देश](https://packaging.python.org/guides/installing-using-pip-and-virtual-environments/) पढ़ें।
|
||||
|
||||
|
|
|
|||
|
|
@ -259,7 +259,7 @@ Hugging Faceチームによって作られた **[トランスフォーマーを
|
|||
|
||||
### pipにて
|
||||
|
||||
このリポジトリは、Python 3.8+, Flax 0.4.1+, PyTorch 1.10+, TensorFlow 2.6+ でテストされています。
|
||||
このリポジトリは、Python 3.8+, Flax 0.4.1+, PyTorch 1.11+, TensorFlow 2.6+ でテストされています。
|
||||
|
||||
🤗Transformersは[仮想環境](https://docs.python.org/3/library/venv.html)にインストールする必要があります。Pythonの仮想環境に慣れていない場合は、[ユーザーガイド](https://packaging.python.org/guides/installing-using-pip-and-virtual-environments/)を確認してください。
|
||||
|
||||
|
|
|
|||
|
|
@ -176,7 +176,7 @@ limitations under the License.
|
|||
|
||||
### pip로 설치하기
|
||||
|
||||
이 저장소는 Python 3.8+, Flax 0.4.1+, PyTorch 1.10+, TensorFlow 2.6+에서 테스트 되었습니다.
|
||||
이 저장소는 Python 3.8+, Flax 0.4.1+, PyTorch 1.11+, TensorFlow 2.6+에서 테스트 되었습니다.
|
||||
|
||||
[가상 환경](https://docs.python.org/3/library/venv.html)에 🤗 Transformers를 설치하세요. Python 가상 환경에 익숙하지 않다면, [사용자 가이드](https://packaging.python.org/guides/installing-using-pip-and-virtual-environments/)를 확인하세요.
|
||||
|
||||
|
|
|
|||
|
|
@ -258,7 +258,7 @@ O modelo em si é um [Pytorch `nn.Module`](https://pytorch.org/docs/stable/nn.ht
|
|||
|
||||
### Com pip
|
||||
|
||||
Este repositório é testado no Python 3.8+, Flax 0.4.1+, PyTorch 1.10+ e TensorFlow 2.6+.
|
||||
Este repositório é testado no Python 3.8+, Flax 0.4.1+, PyTorch 1.11+ e TensorFlow 2.6+.
|
||||
|
||||
Você deve instalar o 🤗 Transformers em um [ambiente virtual](https://docs.python.org/3/library/venv.html). Se você não está familiarizado com ambientes virtuais em Python, confira o [guia do usuário](https://packaging.python.org/guides/installing-using-pip-and-virtual-environments/).
|
||||
|
||||
|
|
|
|||
|
|
@ -248,7 +248,7 @@ Hugging Face Hub. Мы хотим, чтобы Transformers позволил ра
|
|||
|
||||
### С помощью pip
|
||||
|
||||
Данный репозиторий протестирован на Python 3.8+, Flax 0.4.1+, PyTorch 1.10+ и TensorFlow 2.6+.
|
||||
Данный репозиторий протестирован на Python 3.8+, Flax 0.4.1+, PyTorch 1.11+ и TensorFlow 2.6+.
|
||||
|
||||
Устанавливать 🤗 Transformers следует в [виртуальной среде](https://docs.python.org/3/library/venv.html). Если вы не знакомы с виртуальными средами Python, ознакомьтесь с [руководством пользователя](https://packaging.python.org/guides/installing-using-pip-and-virtual-environments/).
|
||||
|
||||
|
|
|
|||
|
|
@ -251,7 +251,7 @@ limitations under the License.
|
|||
|
||||
### పిప్ తో
|
||||
|
||||
ఈ రిపోజిటరీ పైథాన్ 3.8+, ఫ్లాక్స్ 0.4.1+, PyTorch 1.10+ మరియు TensorFlow 2.6+లో పరీక్షించబడింది.
|
||||
ఈ రిపోజిటరీ పైథాన్ 3.8+, ఫ్లాక్స్ 0.4.1+, PyTorch 1.11+ మరియు TensorFlow 2.6+లో పరీక్షించబడింది.
|
||||
|
||||
మీరు [వర్చువల్ వాతావరణం](https://docs.python.org/3/library/venv.html)లో 🤗 ట్రాన్స్ఫార్మర్లను ఇన్స్టాల్ చేయాలి. మీకు పైథాన్ వర్చువల్ పరిసరాల గురించి తెలియకుంటే, [యూజర్ గైడ్](https://packaging.python.org/guides/installing-using-pip-and-virtual-environments/) చూడండి.
|
||||
|
||||
|
|
|
|||
|
|
@ -201,7 +201,7 @@ checkpoint: 检查点
|
|||
|
||||
### 使用 pip
|
||||
|
||||
这个仓库已在 Python 3.8+、Flax 0.4.1+、PyTorch 1.10+ 和 TensorFlow 2.6+ 下经过测试。
|
||||
这个仓库已在 Python 3.8+、Flax 0.4.1+、PyTorch 1.11+ 和 TensorFlow 2.6+ 下经过测试。
|
||||
|
||||
你可以在[虚拟环境](https://docs.python.org/3/library/venv.html)中安装 🤗 Transformers。如果你还不熟悉 Python 的虚拟环境，请阅此[用户说明](https://packaging.python.org/guides/installing-using-pip-and-virtual-environments/)。
|
||||
|
||||
|
|
|
|||
|
|
@ -213,7 +213,7 @@ Tokenizer 為所有的預訓練模型提供了預處理，並可以直接轉換
|
|||
|
||||
### 使用 pip
|
||||
|
||||
這個 Repository 已在 Python 3.8+、Flax 0.4.1+、PyTorch 1.10+ 和 TensorFlow 2.6+ 下經過測試。
|
||||
這個 Repository 已在 Python 3.8+、Flax 0.4.1+、PyTorch 1.11+ 和 TensorFlow 2.6+ 下經過測試。
|
||||
|
||||
你可以在[虛擬環境](https://docs.python.org/3/library/venv.html)中安裝 🤗 Transformers。如果你還不熟悉 Python 的虛擬環境，請閱此[使用者指引](https://packaging.python.org/guides/installing-using-pip-and-virtual-environments/)。
|
||||
|
||||
|
|
|
|||
2
setup.py
2
setup.py
|
|
@ -175,7 +175,7 @@ _deps = [
|
|||
"timeout-decorator",
|
||||
"timm",
|
||||
"tokenizers>=0.14,<0.19",
|
||||
"torch>=1.10,!=1.12.0",
|
||||
"torch>=1.11,!=1.12.0",
|
||||
"torchaudio",
|
||||
"torchvision",
|
||||
"pyctcdecode>=0.4.0",
|
||||
|
|
|
|||
|
|
@ -273,40 +273,22 @@ def convert_pytorch(nlp: Pipeline, opset: int, output: Path, use_external_format
|
|||
import torch
|
||||
from torch.onnx import export
|
||||
|
||||
from transformers.pytorch_utils import is_torch_less_than_1_11
|
||||
|
||||
print(f"Using framework PyTorch: {torch.__version__}")
|
||||
|
||||
with torch.no_grad():
|
||||
input_names, output_names, dynamic_axes, tokens = infer_shapes(nlp, "pt")
|
||||
ordered_input_names, model_args = ensure_valid_input(nlp.model, tokens, input_names)
|
||||
|
||||
# PyTorch deprecated the `enable_onnx_checker` and `use_external_data_format` arguments in v1.11,
|
||||
# so we check the torch version for backwards compatibility
|
||||
if is_torch_less_than_1_11:
|
||||
export(
|
||||
nlp.model,
|
||||
model_args,
|
||||
f=output.as_posix(),
|
||||
input_names=ordered_input_names,
|
||||
output_names=output_names,
|
||||
dynamic_axes=dynamic_axes,
|
||||
do_constant_folding=True,
|
||||
use_external_data_format=use_external_format,
|
||||
enable_onnx_checker=True,
|
||||
opset_version=opset,
|
||||
)
|
||||
else:
|
||||
export(
|
||||
nlp.model,
|
||||
model_args,
|
||||
f=output.as_posix(),
|
||||
input_names=ordered_input_names,
|
||||
output_names=output_names,
|
||||
dynamic_axes=dynamic_axes,
|
||||
do_constant_folding=True,
|
||||
opset_version=opset,
|
||||
)
|
||||
export(
|
||||
nlp.model,
|
||||
model_args,
|
||||
f=output.as_posix(),
|
||||
input_names=ordered_input_names,
|
||||
output_names=output_names,
|
||||
dynamic_axes=dynamic_axes,
|
||||
do_constant_folding=True,
|
||||
opset_version=opset,
|
||||
)
|
||||
|
||||
|
||||
def convert_tensorflow(nlp: Pipeline, opset: int, output: Path):
|
||||
|
|
|
|||
|
|
@ -80,7 +80,7 @@ deps = {
|
|||
"timeout-decorator": "timeout-decorator",
|
||||
"timm": "timm",
|
||||
"tokenizers": "tokenizers>=0.14,<0.19",
|
||||
"torch": "torch>=1.10,!=1.12.0",
|
||||
"torch": "torch>=1.11,!=1.12.0",
|
||||
"torchaudio": "torchaudio",
|
||||
"torchvision": "torchvision",
|
||||
"pyctcdecode": "pyctcdecode>=0.4.0",
|
||||
|
|
|
|||
|
|
@ -97,7 +97,6 @@ from .utils.import_utils import (
|
|||
is_torchdynamo_compiling,
|
||||
)
|
||||
from .utils.quantization_config import AwqConfig, BitsAndBytesConfig, GPTQConfig, QuantizationMethod
|
||||
from .utils.versions import require_version_core
|
||||
|
||||
|
||||
XLA_USE_BF16 = os.environ.get("XLA_USE_BF16", "0").upper()
|
||||
|
|
@ -2898,10 +2897,6 @@ class PreTrainedModel(nn.Module, ModuleUtilsMixin, GenerationMixin, PushToHubMix
|
|||
raise ValueError("Passing along a `device_map` requires `low_cpu_mem_usage=True`")
|
||||
|
||||
if low_cpu_mem_usage:
|
||||
if device_map is not None:
|
||||
# The max memory utils require PyTorch >= 1.10 to have torch.cuda.mem_get_info.
|
||||
require_version_core("torch>=1.10")
|
||||
|
||||
if is_deepspeed_zero3_enabled():
|
||||
raise ValueError(
|
||||
"DeepSpeed Zero-3 is not compatible with `low_cpu_mem_usage=True` or with passing a `device_map`."
|
||||
|
|
|
|||
|
|
@ -43,23 +43,10 @@ if is_vision_available():
|
|||
if is_torch_available():
|
||||
import torch
|
||||
|
||||
from transformers.pytorch_utils import is_torch_greater_or_equal_than_1_11
|
||||
else:
|
||||
is_torch_greater_or_equal_than_1_11 = False
|
||||
|
||||
|
||||
logger = logging.get_logger(__name__)
|
||||
DEFAULT_FONT_PATH = "ybelkada/fonts"
|
||||
|
||||
|
||||
def _check_torch_version():
|
||||
if is_torch_available() and not is_torch_greater_or_equal_than_1_11:
|
||||
raise ImportError(
|
||||
f"You are using torch=={torch.__version__}, but torch>=1.11.0 is required to use "
|
||||
"Pix2StructImageProcessor. Please upgrade torch."
|
||||
)
|
||||
|
||||
|
||||
# adapted from: https://discuss.pytorch.org/t/tf-image-extract-patches-in-pytorch/171409/2
|
||||
def torch_extract_patches(image_tensor, patch_height, patch_width):
|
||||
"""
|
||||
|
|
@ -75,7 +62,6 @@ def torch_extract_patches(image_tensor, patch_height, patch_width):
|
|||
The width of the patches to extract.
|
||||
"""
|
||||
requires_backends(torch_extract_patches, ["torch"])
|
||||
_check_torch_version()
|
||||
|
||||
image_tensor = image_tensor.unsqueeze(0)
|
||||
patches = torch.nn.functional.unfold(image_tensor, (patch_height, patch_width), stride=(patch_height, patch_width))
|
||||
|
|
@ -262,7 +248,6 @@ class Pix2StructImageProcessor(BaseImageProcessor):
|
|||
A sequence of `max_patches` flattened patches.
|
||||
"""
|
||||
requires_backends(self.extract_flattened_patches, "torch")
|
||||
_check_torch_version()
|
||||
|
||||
# convert to torch
|
||||
image = to_channel_dimension_format(image, ChannelDimension.FIRST, input_data_format)
|
||||
|
|
|
|||
|
|
@ -33,7 +33,6 @@ from .config import OnnxConfig
|
|||
|
||||
if is_torch_available():
|
||||
from ..modeling_utils import PreTrainedModel
|
||||
from ..pytorch_utils import is_torch_less_than_1_11
|
||||
|
||||
if is_tf_available():
|
||||
from ..modeling_tf_utils import TFPreTrainedModel
|
||||
|
|
@ -167,49 +166,16 @@ def export_pytorch(
|
|||
|
||||
config.patch_ops()
|
||||
|
||||
# PyTorch deprecated the `enable_onnx_checker` and `use_external_data_format` arguments in v1.11,
|
||||
# so we check the torch version for backwards compatibility
|
||||
if is_torch_less_than_1_11:
|
||||
# export can work with named args but the dict containing named args
|
||||
# has to be the last element of the args tuple.
|
||||
try:
|
||||
onnx_export(
|
||||
model,
|
||||
(model_inputs,),
|
||||
f=output.as_posix(),
|
||||
input_names=list(config.inputs.keys()),
|
||||
output_names=onnx_outputs,
|
||||
dynamic_axes=dict(chain(config.inputs.items(), config.outputs.items())),
|
||||
do_constant_folding=True,
|
||||
use_external_data_format=config.use_external_data_format(model.num_parameters()),
|
||||
enable_onnx_checker=True,
|
||||
opset_version=opset,
|
||||
)
|
||||
except RuntimeError as err:
|
||||
message = str(err)
|
||||
if (
|
||||
message
|
||||
== "Exporting model exceed maximum protobuf size of 2GB. Please call torch.onnx.export without"
|
||||
" setting use_external_data_format parameter."
|
||||
):
|
||||
message = (
|
||||
"Exporting model exceed maximum protobuf size of 2GB. Please call torch.onnx.export"
|
||||
" without setting use_external_data_format parameter or try with torch 1.10+."
|
||||
)
|
||||
raise RuntimeError(message)
|
||||
else:
|
||||
raise err
|
||||
else:
|
||||
onnx_export(
|
||||
model,
|
||||
(model_inputs,),
|
||||
f=output.as_posix(),
|
||||
input_names=list(config.inputs.keys()),
|
||||
output_names=onnx_outputs,
|
||||
dynamic_axes=dict(chain(config.inputs.items(), config.outputs.items())),
|
||||
do_constant_folding=True,
|
||||
opset_version=opset,
|
||||
)
|
||||
onnx_export(
|
||||
model,
|
||||
(model_inputs,),
|
||||
f=output.as_posix(),
|
||||
input_names=list(config.inputs.keys()),
|
||||
output_names=onnx_outputs,
|
||||
dynamic_axes=dict(chain(config.inputs.items(), config.outputs.items())),
|
||||
do_constant_folding=True,
|
||||
opset_version=opset,
|
||||
)
|
||||
|
||||
config.restore_ops()
|
||||
|
||||
|
|
|
|||
|
|
@ -32,9 +32,6 @@ is_torch_greater_or_equal_than_2_1 = parsed_torch_version_base >= version.parse(
|
|||
is_torch_greater_or_equal_than_2_0 = parsed_torch_version_base >= version.parse("2.0")
|
||||
is_torch_greater_or_equal_than_1_13 = parsed_torch_version_base >= version.parse("1.13")
|
||||
is_torch_greater_or_equal_than_1_12 = parsed_torch_version_base >= version.parse("1.12")
|
||||
is_torch_greater_or_equal_than_1_11 = parsed_torch_version_base >= version.parse("1.11")
|
||||
is_torch_less_than_1_11 = parsed_torch_version_base < version.parse("1.11")
|
||||
is_torch_1_8_0 = parsed_torch_version_base == version.parse("1.8.0")
|
||||
|
||||
|
||||
def softmax_backward_data(parent, grad_output, output, dim, self):
|
||||
|
|
@ -45,10 +42,7 @@ def softmax_backward_data(parent, grad_output, output, dim, self):
|
|||
|
||||
from torch import _softmax_backward_data
|
||||
|
||||
if is_torch_less_than_1_11:
|
||||
return _softmax_backward_data(grad_output, output, parent.dim, self)
|
||||
else:
|
||||
return _softmax_backward_data(grad_output, output, parent.dim, self.dtype)
|
||||
return _softmax_backward_data(grad_output, output, parent.dim, self.dtype)
|
||||
|
||||
|
||||
def prune_linear_layer(layer: nn.Linear, index: torch.LongTensor, dim: int = 0) -> nn.Linear:
|
||||
|
|
|
|||
|
|
@ -64,7 +64,7 @@ from .modelcard import TrainingSummary
|
|||
from .modeling_utils import PreTrainedModel, load_sharded_checkpoint, unwrap_model
|
||||
from .models.auto.modeling_auto import MODEL_FOR_CAUSAL_LM_MAPPING_NAMES, MODEL_MAPPING_NAMES
|
||||
from .optimization import Adafactor, get_scheduler
|
||||
from .pytorch_utils import ALL_LAYERNORM_LAYERS, is_torch_less_than_1_11
|
||||
from .pytorch_utils import ALL_LAYERNORM_LAYERS
|
||||
from .tokenization_utils_base import PreTrainedTokenizerBase
|
||||
from .trainer_callback import (
|
||||
CallbackHandler,
|
||||
|
|
@ -1794,7 +1794,7 @@ class Trainer:
|
|||
if version.parse(accelerate_version) > version.parse("0.23.0"):
|
||||
sampler_kinds.append(SeedableRandomSampler)
|
||||
is_random_sampler = isinstance(sampler, tuple(sampler_kinds))
|
||||
if is_torch_less_than_1_11 or not is_random_sampler:
|
||||
if not is_random_sampler:
|
||||
# We just need to begin an iteration to create the randomization of the sampler.
|
||||
for _ in train_dataloader:
|
||||
break
|
||||
|
|
|
|||
|
|
@ -1439,15 +1439,6 @@ class TrainingArguments:
|
|||
raise ValueError(
|
||||
"Your setup doesn't support bf16/gpu. You need torch>=1.10, using Ampere GPU with cuda>=11.0"
|
||||
)
|
||||
elif is_torch_npu_available():
|
||||
# npu
|
||||
from .pytorch_utils import is_torch_greater_or_equal_than_1_11
|
||||
|
||||
if not is_torch_greater_or_equal_than_1_11:
|
||||
raise ValueError(
|
||||
"Your setup doesn't support bf16/npu. You need torch>=1.11, using Ascend NPU with "
|
||||
"`torch_npu` installed"
|
||||
)
|
||||
elif not is_torch_xpu_available():
|
||||
# xpu
|
||||
from .pytorch_utils import is_torch_greater_or_equal_than_1_12
|
||||
|
|
|
|||
|
|
@ -64,6 +64,7 @@ USE_JAX = os.environ.get("USE_FLAX", "AUTO").upper()
|
|||
|
||||
FORCE_TF_AVAILABLE = os.environ.get("FORCE_TF_AVAILABLE", "AUTO").upper()
|
||||
|
||||
# `transformers` requires `torch>=1.11` but this variable is exposed publicly, and we can't simply remove it.
|
||||
# This is the version of torch required to run torch.fx features and torch.onnx with dictionary inputs.
|
||||
TORCH_FX_REQUIRED_VERSION = version.parse("1.10")
|
||||
|
||||
|
|
|
|||
|
|
@ -40,7 +40,6 @@ if is_torch_available():
|
|||
LongT5Model,
|
||||
)
|
||||
from transformers.models.longt5.modeling_longt5 import LONGT5_PRETRAINED_MODEL_ARCHIVE_LIST
|
||||
from transformers.pytorch_utils import is_torch_less_than_1_11
|
||||
|
||||
|
||||
class LongT5ModelTester:
|
||||
|
|
@ -595,10 +594,6 @@ class LongT5ModelTest(ModelTesterMixin, GenerationTesterMixin, PipelineTesterMix
|
|||
model = LongT5Model.from_pretrained(model_name)
|
||||
self.assertIsNotNone(model)
|
||||
|
||||
@unittest.skipIf(
|
||||
not is_torch_available() or is_torch_less_than_1_11,
|
||||
"Test failed with torch < 1.11 with an exception in a C++ file.",
|
||||
)
|
||||
@slow
|
||||
def test_export_to_onnx(self):
|
||||
config_and_inputs = self.model_tester.prepare_config_and_inputs()
|
||||
|
|
|
|||
|
|
@ -28,10 +28,6 @@ from ...test_image_processing_common import ImageProcessingTestMixin, prepare_im
|
|||
if is_torch_available():
|
||||
import torch
|
||||
|
||||
from transformers.pytorch_utils import is_torch_greater_or_equal_than_1_11
|
||||
else:
|
||||
is_torch_greater_or_equal_than_1_11 = False
|
||||
|
||||
if is_vision_available():
|
||||
from PIL import Image
|
||||
|
||||
|
|
@ -85,10 +81,6 @@ class Pix2StructImageProcessingTester(unittest.TestCase):
|
|||
)
|
||||
|
||||
|
||||
@unittest.skipIf(
|
||||
not is_torch_greater_or_equal_than_1_11,
|
||||
reason="`Pix2StructImageProcessor` requires `torch>=1.11.0`.",
|
||||
)
|
||||
@require_torch
|
||||
@require_vision
|
||||
class Pix2StructImageProcessingTest(ImageProcessingTestMixin, unittest.TestCase):
|
||||
|
|
@ -290,10 +282,6 @@ class Pix2StructImageProcessingTest(ImageProcessingTestMixin, unittest.TestCase)
|
|||
)
|
||||
|
||||
|
||||
@unittest.skipIf(
|
||||
not is_torch_greater_or_equal_than_1_11,
|
||||
reason="`Pix2StructImageProcessor` requires `torch>=1.11.0`.",
|
||||
)
|
||||
@require_torch
|
||||
@require_vision
|
||||
class Pix2StructImageProcessingTestFourChannels(ImageProcessingTestMixin, unittest.TestCase):
|
||||
|
|
|
|||
|
|
@ -49,9 +49,6 @@ if is_torch_available():
|
|||
Pix2StructVisionModel,
|
||||
)
|
||||
from transformers.models.pix2struct.modeling_pix2struct import PIX2STRUCT_PRETRAINED_MODEL_ARCHIVE_LIST
|
||||
from transformers.pytorch_utils import is_torch_greater_or_equal_than_1_11
|
||||
else:
|
||||
is_torch_greater_or_equal_than_1_11 = False
|
||||
|
||||
|
||||
if is_vision_available():
|
||||
|
|
@ -746,10 +743,6 @@ def prepare_img():
|
|||
return im
|
||||
|
||||
|
||||
@unittest.skipIf(
|
||||
not is_torch_greater_or_equal_than_1_11,
|
||||
reason="`Pix2StructImageProcessor` requires `torch>=1.11.0`.",
|
||||
)
|
||||
@require_vision
|
||||
@require_torch
|
||||
@slow
|
||||
|
|
|
|||
|
|
@ -19,14 +19,9 @@ import numpy as np
|
|||
import pytest
|
||||
|
||||
from transformers.testing_utils import require_torch, require_vision
|
||||
from transformers.utils import is_torch_available, is_vision_available
|
||||
from transformers.utils import is_vision_available
|
||||
|
||||
|
||||
if is_torch_available():
|
||||
from transformers.pytorch_utils import is_torch_greater_or_equal_than_1_11
|
||||
else:
|
||||
is_torch_greater_or_equal_than_1_11 = False
|
||||
|
||||
if is_vision_available():
|
||||
from PIL import Image
|
||||
|
||||
|
|
@ -39,10 +34,6 @@ if is_vision_available():
|
|||
)
|
||||
|
||||
|
||||
@unittest.skipIf(
|
||||
not is_torch_greater_or_equal_than_1_11,
|
||||
reason="`Pix2StructImageProcessor` requires `torch>=1.11.0`.",
|
||||
)
|
||||
@require_vision
|
||||
@require_torch
|
||||
class Pix2StructProcessorTest(unittest.TestCase):
|
||||
|
|
|
|||
|
|
@ -45,10 +45,6 @@ if is_torch_available():
|
|||
|
||||
from transformers import Pop2PianoForConditionalGeneration
|
||||
from transformers.models.pop2piano.modeling_pop2piano import POP2PIANO_PRETRAINED_MODEL_ARCHIVE_LIST
|
||||
from transformers.pytorch_utils import is_torch_1_8_0
|
||||
|
||||
else:
|
||||
is_torch_1_8_0 = False
|
||||
|
||||
|
||||
@require_torch
|
||||
|
|
@ -616,10 +612,6 @@ class Pop2PianoModelTest(ModelTesterMixin, GenerationTesterMixin, PipelineTester
|
|||
self.assertIsNotNone(model)
|
||||
|
||||
@require_onnx
|
||||
@unittest.skipIf(
|
||||
is_torch_1_8_0,
|
||||
reason="Test has a segmentation fault on torch 1.8.0",
|
||||
)
|
||||
def test_export_to_onnx(self):
|
||||
config_and_inputs = self.model_tester.prepare_config_and_inputs()
|
||||
model = Pop2PianoForConditionalGeneration(config_and_inputs[0]).to(torch_device)
|
||||
|
|
|
|||
|
|
@ -906,7 +906,6 @@ class UniSpeechSatModelIntegrationTest(unittest.TestCase):
|
|||
)
|
||||
self.assertEqual(labels[0, :, 0].sum(), 270)
|
||||
self.assertEqual(labels[0, :, 1].sum(), 647)
|
||||
# TODO: update the tolerance after the CI moves to torch 1.10
|
||||
self.assertTrue(torch.allclose(outputs.logits[:, :4], expected_logits, atol=1e-2))
|
||||
|
||||
def test_inference_speaker_verification(self):
|
||||
|
|
@ -931,5 +930,4 @@ class UniSpeechSatModelIntegrationTest(unittest.TestCase):
|
|||
# id10002 vs id10004
|
||||
self.assertAlmostEqual(cosine_sim(embeddings[2], embeddings[3]).item(), 0.5616, 3)
|
||||
|
||||
# TODO: update the tolerance after the CI moves to torch 1.10
|
||||
self.assertAlmostEqual(outputs.loss.item(), 18.5925, 2)
|
||||
|
|
|
|||
|
|
@ -1928,7 +1928,6 @@ class Wav2Vec2ModelIntegrationTest(unittest.TestCase):
|
|||
)
|
||||
self.assertEqual(labels[0, :, 0].sum(), 555)
|
||||
self.assertEqual(labels[0, :, 1].sum(), 299)
|
||||
# TODO: update the tolerance after the CI moves to torch 1.10
|
||||
self.assertTrue(torch.allclose(outputs.logits[:, :4], expected_logits, atol=1e-2))
|
||||
|
||||
def test_inference_speaker_verification(self):
|
||||
|
|
@ -1953,7 +1952,6 @@ class Wav2Vec2ModelIntegrationTest(unittest.TestCase):
|
|||
# id10002 vs id10004
|
||||
self.assertAlmostEqual(cosine_sim(embeddings[2], embeddings[3]).numpy(), 0.7594, 3)
|
||||
|
||||
# TODO: update the tolerance after the CI moves to torch 1.10
|
||||
self.assertAlmostEqual(outputs.loss.item(), 17.7963, 2)
|
||||
|
||||
@require_torchaudio
|
||||
|
|
|
|||
|
|
@ -515,7 +515,6 @@ class WavLMModelIntegrationTest(unittest.TestCase):
|
|||
EXPECTED_HIDDEN_STATES_SLICE = torch.tensor(
|
||||
[[[0.0577, 0.1161], [0.0579, 0.1165]], [[0.0199, 0.1237], [0.0059, 0.0605]]]
|
||||
)
|
||||
# TODO: update the tolerance after the CI moves to torch 1.10
|
||||
self.assertTrue(torch.allclose(hidden_states_slice, EXPECTED_HIDDEN_STATES_SLICE, atol=5e-2))
|
||||
|
||||
def test_inference_large(self):
|
||||
|
|
@ -567,7 +566,6 @@ class WavLMModelIntegrationTest(unittest.TestCase):
|
|||
)
|
||||
self.assertEqual(labels[0, :, 0].sum(), 258)
|
||||
self.assertEqual(labels[0, :, 1].sum(), 647)
|
||||
# TODO: update the tolerance after the CI moves to torch 1.10
|
||||
self.assertTrue(torch.allclose(outputs.logits[:, :4], expected_logits, atol=1e-2))
|
||||
|
||||
def test_inference_speaker_verification(self):
|
||||
|
|
@ -592,5 +590,4 @@ class WavLMModelIntegrationTest(unittest.TestCase):
|
|||
# id10002 vs id10004
|
||||
self.assertAlmostEqual(cosine_sim(embeddings[2], embeddings[3]).item(), 0.4780, 3)
|
||||
|
||||
# TODO: update the tolerance after the CI moves to torch 1.10
|
||||
self.assertAlmostEqual(outputs.loss.item(), 18.4154, 2)
|
||||
|
|
|
|||
|
|
@ -20,7 +20,6 @@ from transformers import MODEL_FOR_VISION_2_SEQ_MAPPING, TF_MODEL_FOR_VISION_2_S
|
|||
from transformers.pipelines import pipeline
|
||||
from transformers.testing_utils import (
|
||||
is_pipeline_test,
|
||||
is_torch_available,
|
||||
require_tf,
|
||||
require_torch,
|
||||
require_vision,
|
||||
|
|
@ -30,12 +29,6 @@ from transformers.testing_utils import (
|
|||
from .test_pipelines_common import ANY
|
||||
|
||||
|
||||
if is_torch_available():
|
||||
from transformers.pytorch_utils import is_torch_greater_or_equal_than_1_11
|
||||
else:
|
||||
is_torch_greater_or_equal_than_1_11 = False
|
||||
|
||||
|
||||
if is_vision_available():
|
||||
from PIL import Image
|
||||
else:
|
||||
|
|
@ -217,9 +210,6 @@ class ImageToTextPipelineTests(unittest.TestCase):
|
|||
with self.assertRaises(ValueError):
|
||||
outputs = pipe([image, image], prompt=[prompt, prompt])
|
||||
|
||||
@unittest.skipIf(
|
||||
not is_torch_greater_or_equal_than_1_11, reason="`Pix2StructImageProcessor` requires `torch>=1.11.0`."
|
||||
)
|
||||
@slow
|
||||
@require_torch
|
||||
def test_conditional_generation_pt_pix2struct(self):
|
||||
|
|
|
|||
Loading…
Reference in New Issue