riaz.somc/transformers
1
0
Fork 0
mirror of https://github.com/huggingface/transformers.git synced 2025-07-31 02:02:21 +06:00
Code Issues Actions 23 Packages Projects Releases Wiki Activity

Time to Say Goodbye, torch 1.7 and 1.8 (#22291)

Browse Source 
* time to say goodbye, torch 1.7 and 1.8

* clean up torch_int_div

* clean up is_torch_less_than_1_8-9

* update

---------

Co-authored-by: ydshieh <ydshieh@users.noreply.github.com>
This commit is contained in:
Yih-Dar 2023-03-21 19:22:01 +01:00 committed by GitHub
parent 86c7931a70
commit 67c2dbdb54
No known key found for this signature in database
GPG Key ID: 4AEE18F83AFDEB23
42 changed files with 61 additions and 149 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

2
setup.py
View File

@ -171,7 +171,7 @@ _deps = [
"timeout-decorator",
"timm",
"tokenizers>=0.11.1,!=0.11.3,<0.14",
"torch>=1.7,!=1.12.0",
"torch>=1.9,!=1.12.0",
"torchaudio",
"torchvision",
"pyctcdecode>=0.4.0",

2
src/transformers/dependency_versions_table.py
View File

@ -77,7 +77,7 @@ deps = {
"timeout-decorator": "timeout-decorator",
"timm": "timm",
"tokenizers": "tokenizers>=0.11.1,!=0.11.3,<0.14",
"torch": "torch>=1.7,!=1.12.0",
"torch": "torch>=1.9,!=1.12.0",
"torchaudio": "torchaudio",
"torchvision": "torchvision",
"pyctcdecode": "pyctcdecode>=0.4.0",

1
src/transformers/file_utils.py
View File

@ -115,7 +115,6 @@ from .utils import (
is_torch_cuda_available,
is_torch_fx_available,
is_torch_fx_proxy,
is_torch_onnx_dict_inputs_support_available,
is_torch_tf32_available,
is_torch_tpu_available,
is_torchaudio_available,

11
src/transformers/generation/utils.py
View File

@ -32,7 +32,6 @@ from ..models.auto import (
MODEL_FOR_SPEECH_SEQ_2_SEQ_MAPPING,
MODEL_FOR_VISION_2_SEQ_MAPPING,
)
from ..pytorch_utils import torch_int_div
from ..utils import ModelOutput, logging
from .beam_constraints import DisjunctiveConstraint, PhrasalConstraint
from .beam_search import BeamScorer, BeamSearchScorer, ConstrainedBeamSearchScorer
@ -2795,7 +2794,7 @@ class GenerationMixin:
next_token_scores, 2 * num_beams, dim=1, largest=True, sorted=True
)
next_indices = torch_int_div(next_tokens, vocab_size)
next_indices = torch.div(next_tokens, vocab_size, rounding_mode="floor")
next_tokens = next_tokens % vocab_size
# stateless
@ -3129,7 +3128,7 @@ class GenerationMixin:
next_token_scores, _indices = torch.sort(next_token_scores, descending=True, dim=1)
next_tokens = torch.gather(next_tokens, -1, _indices)
next_indices = torch_int_div(next_tokens, vocab_size)
next_indices = torch.div(next_tokens, vocab_size, rounding_mode="floor")
next_tokens = next_tokens % vocab_size
# stateless
@ -3473,7 +3472,7 @@ class GenerationMixin:
next_token_scores, 2 * group_size, dim=1, largest=True, sorted=True
)
next_indices = torch_int_div(next_tokens, vocab_size)
next_indices = torch.div(next_tokens, vocab_size, rounding_mode="floor")
next_tokens = next_tokens % vocab_size
# stateless
@ -3503,7 +3502,9 @@ class GenerationMixin:
# (beam_idx // group_size) -> batch_idx
# (beam_idx % group_size) -> offset of idx inside the group
reordering_indices[batch_group_indices] = (
num_beams * torch_int_div(beam_idx, group_size) + group_start_idx + (beam_idx % group_size)
num_beams * torch.div(beam_idx, group_size, rounding_mode="floor")
+ group_start_idx
+ (beam_idx % group_size)
)
# Store scores, attentions and hidden_states when required

5
src/transformers/modeling_utils.py
View File

@ -539,9 +539,6 @@ def _move_model_to_meta(model, loaded_state_dict_keys, start_prefix):
"""
# meta device was added in pt=1.9
require_version_core("torch>=1.9")
# dematerialize param storage for keys that are going to be replaced by state_dict, by
# putting those on the meta device
for k in loaded_state_dict_keys:
@ -2100,8 +2097,6 @@ 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:
# low_cpu_mem_usage requires PyTorch >= 1.9 to have the meta device.
require_version_core("torch>=1.9")
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")

4
src/transformers/models/big_bird/modeling_big_bird.py
View File

@ -37,7 +37,7 @@ from ...modeling_outputs import (
TokenClassifierOutput,
)
from ...modeling_utils import PreTrainedModel
from ...pytorch_utils import apply_chunking_to_forward, torch_int_div
from ...pytorch_utils import apply_chunking_to_forward
from ...utils import (
ModelOutput,
add_code_sample_docstrings,
@ -972,7 +972,7 @@ class BigBirdBlockSparseAttention(nn.Module):
num_indices_to_pick_from = params.shape[2]
shift = torch.arange(indices.shape[0] * indices.shape[1] * num_indices_to_gather, device=indices.device)
indices_shift = torch_int_div(shift, num_indices_to_gather) * num_indices_to_pick_from
indices_shift = torch.div(shift, num_indices_to_gather, rounding_mode="floor") * num_indices_to_pick_from
flattened_indices = indices.view(-1) + indices_shift
flattened_params = params.reshape(-1, params.shape[-2], params.shape[-1])

3
src/transformers/models/bigbird_pegasus/modeling_bigbird_pegasus.py
View File

@ -36,7 +36,6 @@ from ...modeling_outputs import (
Seq2SeqSequenceClassifierOutput,
)
from ...modeling_utils import PreTrainedModel
from ...pytorch_utils import torch_int_div
from ...utils import (
add_code_sample_docstrings,
add_end_docstrings,
@ -791,7 +790,7 @@ class BigBirdPegasusBlockSparseAttention(nn.Module):
num_indices_to_pick_from = params.shape[2]
shift = torch.arange(indices.shape[0] * indices.shape[1] * num_indices_to_gather, device=indices.device)
indices_shift = torch_int_div(shift, num_indices_to_gather) * num_indices_to_pick_from
indices_shift = torch.div(shift, num_indices_to_gather, rounding_mode="floor") * num_indices_to_pick_from
flattened_indices = indices.view(-1) + indices_shift
flattened_params = params.reshape(-1, params.shape[-2], params.shape[-1])

6
src/transformers/models/conditional_detr/image_processing_conditional_detr.py
View File

@ -68,8 +68,6 @@ if is_torch_available():
import torch
from torch import nn
from transformers.pytorch_utils import torch_int_div
if is_vision_available():
import PIL
@ -1314,7 +1312,7 @@ class ConditionalDetrImageProcessor(BaseImageProcessor):
prob = out_logits.sigmoid()
topk_values, topk_indexes = torch.topk(prob.view(out_logits.shape[0], -1), 300, dim=1)
scores = topk_values
topk_boxes = torch_int_div(topk_indexes, out_logits.shape[2])
topk_boxes = torch.div(topk_indexes, out_logits.shape[2], rounding_mode="floor")
labels = topk_indexes % out_logits.shape[2]
boxes = center_to_corners_format(out_bbox)
boxes = torch.gather(boxes, 1, topk_boxes.unsqueeze(-1).repeat(1, 1, 4))
@ -1360,7 +1358,7 @@ class ConditionalDetrImageProcessor(BaseImageProcessor):
prob = out_logits.sigmoid()
topk_values, topk_indexes = torch.topk(prob.view(out_logits.shape[0], -1), 100, dim=1)
scores = topk_values
topk_boxes = torch_int_div(topk_indexes, out_logits.shape[2])
topk_boxes = torch.div(topk_indexes, out_logits.shape[2], rounding_mode="floor")
labels = topk_indexes % out_logits.shape[2]
boxes = center_to_corners_format(out_bbox)
boxes = torch.gather(boxes, 1, topk_boxes.unsqueeze(-1).repeat(1, 1, 4))

5
src/transformers/models/conditional_detr/modeling_conditional_detr.py
View File

@ -26,7 +26,6 @@ from torch import Tensor, nn
from ...activations import ACT2FN
from ...modeling_outputs import BaseModelOutput, BaseModelOutputWithCrossAttentions, Seq2SeqModelOutput
from ...modeling_utils import PreTrainedModel
from ...pytorch_utils import torch_int_div
from ...utils import (
ModelOutput,
add_start_docstrings,
@ -452,7 +451,7 @@ class ConditionalDetrSinePositionEmbedding(nn.Module):
x_embed = x_embed / (x_embed[:, :, -1:] + 1e-6) * self.scale
dim_t = torch.arange(self.embedding_dim, dtype=torch.float32, device=pixel_values.device)
dim_t = self.temperature ** (2 * torch_int_div(dim_t, 2) / self.embedding_dim)
dim_t = self.temperature ** (2 * torch.div(dim_t, 2, rounding_mode="floor") / self.embedding_dim)
pos_x = x_embed[:, :, :, None] / dim_t
pos_y = y_embed[:, :, :, None] / dim_t
@ -504,7 +503,7 @@ def build_position_encoding(config):
def gen_sine_position_embeddings(pos_tensor):
scale = 2 * math.pi
dim_t = torch.arange(128, dtype=torch.float32, device=pos_tensor.device)
dim_t = 10000 ** (2 * torch_int_div(dim_t, 2) / 128)
dim_t = 10000 ** (2 * torch.div(dim_t, 2, rounding_mode="floor") / 128)
x_embed = pos_tensor[:, :, 0] * scale
y_embed = pos_tensor[:, :, 1] * scale
pos_x = x_embed[:, :, None] / dim_t

3
src/transformers/models/data2vec/modeling_data2vec_audio.py
View File

@ -35,7 +35,6 @@ from ...modeling_outputs import (
XVectorOutput,
)
from ...modeling_utils import PreTrainedModel
from ...pytorch_utils import torch_int_div
from ...utils import add_code_sample_docstrings, add_start_docstrings, add_start_docstrings_to_model_forward, logging
from .configuration_data2vec_audio import Data2VecAudioConfig
@ -731,7 +730,7 @@ class Data2VecAudioPreTrainedModel(PreTrainedModel):
def _conv_out_length(input_length, kernel_size, stride):
# 1D convolutional layer output length formula taken
# from https://pytorch.org/docs/stable/generated/torch.nn.Conv1d.html
return torch_int_div(input_length - kernel_size, stride) + 1
return torch.div(input_length - kernel_size, stride, rounding_mode="floor") + 1
for kernel_size, stride in zip(self.config.conv_kernel, self.config.conv_stride):
input_lengths = _conv_out_length(input_lengths, kernel_size, stride)

6
src/transformers/models/deformable_detr/image_processing_deformable_detr.py
View File

@ -68,8 +68,6 @@ if is_torch_available():
import torch
from torch import nn
from ...pytorch_utils import torch_int_div
if is_vision_available():
import PIL
@ -1312,7 +1310,7 @@ class DeformableDetrImageProcessor(BaseImageProcessor):
prob = out_logits.sigmoid()
topk_values, topk_indexes = torch.topk(prob.view(out_logits.shape[0], -1), 100, dim=1)
scores = topk_values
topk_boxes = torch_int_div(topk_indexes, out_logits.shape[2])
topk_boxes = torch.div(topk_indexes, out_logits.shape[2], rounding_mode="floor")
labels = topk_indexes % out_logits.shape[2]
boxes = center_to_corners_format(out_bbox)
boxes = torch.gather(boxes, 1, topk_boxes.unsqueeze(-1).repeat(1, 1, 4))
@ -1357,7 +1355,7 @@ class DeformableDetrImageProcessor(BaseImageProcessor):
prob = out_logits.sigmoid()
topk_values, topk_indexes = torch.topk(prob.view(out_logits.shape[0], -1), 100, dim=1)
scores = topk_values
topk_boxes = torch_int_div(topk_indexes, out_logits.shape[2])
topk_boxes = torch.div(topk_indexes, out_logits.shape[2], rounding_mode="floor")
labels = topk_indexes % out_logits.shape[2]
boxes = center_to_corners_format(out_bbox)
boxes = torch.gather(boxes, 1, topk_boxes.unsqueeze(-1).repeat(1, 1, 4))

6
src/transformers/models/deformable_detr/modeling_deformable_detr.py
View File

@ -41,7 +41,7 @@ from ...file_utils import (
)
from ...modeling_outputs import BaseModelOutput
from ...modeling_utils import PreTrainedModel
from ...pytorch_utils import meshgrid, torch_int_div
from ...pytorch_utils import meshgrid
from ...utils import is_ninja_available, logging
from ..auto import AutoBackbone
from .configuration_deformable_detr import DeformableDetrConfig
@ -497,7 +497,7 @@ class DeformableDetrSinePositionEmbedding(nn.Module):
x_embed = (x_embed - 0.5) / (x_embed[:, :, -1:] + eps) * self.scale
dim_t = torch.arange(self.embedding_dim, dtype=torch.float32, device=pixel_values.device)
dim_t = self.temperature ** (2 * torch_int_div(dim_t, 2) / self.embedding_dim)
dim_t = self.temperature ** (2 * torch.div(dim_t, 2, rounding_mode="floor") / self.embedding_dim)
pos_x = x_embed[:, :, :, None] / dim_t
pos_y = y_embed[:, :, :, None] / dim_t
@ -1552,7 +1552,7 @@ class DeformableDetrModel(DeformableDetrPreTrainedModel):
scale = 2 * math.pi
dim_t = torch.arange(num_pos_feats, dtype=torch.float32, device=proposals.device)
dim_t = temperature ** (2 * torch_int_div(dim_t, 2) / num_pos_feats)
dim_t = temperature ** (2 * torch.div(dim_t, 2, rounding_mode="floor") / num_pos_feats)
# batch_size, num_queries, 4
proposals = proposals.sigmoid() * scale
# batch_size, num_queries, 4, 128

3
src/transformers/models/deta/image_processing_deta.py
View File

@ -63,7 +63,6 @@ from ...utils.generic import ExplicitEnum, TensorType
if is_torch_available():
import torch
from ...pytorch_utils import torch_int_div
if is_torchvision_available():
from torchvision.ops.boxes import batched_nms
@ -967,7 +966,7 @@ class DetaImageProcessor(BaseImageProcessor):
all_scores = prob.view(batch_size, num_queries * num_labels).to(out_logits.device)
all_indexes = torch.arange(num_queries * num_labels)[None].repeat(batch_size, 1).to(out_logits.device)
all_boxes = torch_int_div(all_indexes, out_logits.shape[2])
all_boxes = torch.div(all_indexes, out_logits.shape[2], rounding_mode="floor")
all_labels = all_indexes % out_logits.shape[2]
boxes = center_to_corners_format(out_bbox)

6
src/transformers/models/deta/modeling_deta.py
View File

@ -36,7 +36,7 @@ from ...file_utils import (
)
from ...modeling_outputs import BaseModelOutput
from ...modeling_utils import PreTrainedModel
from ...pytorch_utils import meshgrid, torch_int_div
from ...pytorch_utils import meshgrid
from ...utils import is_torchvision_available, logging, requires_backends
from ..auto import AutoBackbone
from .configuration_deta import DetaConfig
@ -399,7 +399,7 @@ class DetaSinePositionEmbedding(nn.Module):
x_embed = (x_embed - 0.5) / (x_embed[:, :, -1:] + eps) * self.scale
dim_t = torch.arange(self.embedding_dim, dtype=torch.float32, device=pixel_values.device)
dim_t = self.temperature ** (2 * torch_int_div(dim_t, 2) / self.embedding_dim)
dim_t = self.temperature ** (2 * torch.div(dim_t, 2, rounding_mode="floor") / self.embedding_dim)
pos_x = x_embed[:, :, :, None] / dim_t
pos_y = y_embed[:, :, :, None] / dim_t
@ -1463,7 +1463,7 @@ class DetaModel(DetaPreTrainedModel):
scale = 2 * math.pi
dim_t = torch.arange(num_pos_feats, dtype=torch.float32, device=proposals.device)
dim_t = temperature ** (2 * torch_int_div(dim_t, 2) / num_pos_feats)
dim_t = temperature ** (2 * torch.div(dim_t, 2, rounding_mode="floor") / num_pos_feats)
# batch_size, num_queries, 4
proposals = proposals.sigmoid() * scale
# batch_size, num_queries, 4, 128

3
src/transformers/models/detr/modeling_detr.py
View File

@ -26,7 +26,6 @@ from torch import Tensor, nn
from ...activations import ACT2FN
from ...modeling_outputs import BaseModelOutput, BaseModelOutputWithCrossAttentions, Seq2SeqModelOutput
from ...modeling_utils import PreTrainedModel
from ...pytorch_utils import torch_int_div
from ...utils import (
ModelOutput,
add_start_docstrings,
@ -442,7 +441,7 @@ class DetrSinePositionEmbedding(nn.Module):
x_embed = x_embed / (x_embed[:, :, -1:] + 1e-6) * self.scale
dim_t = torch.arange(self.embedding_dim, dtype=torch.float32, device=pixel_values.device)
dim_t = self.temperature ** (2 * torch_int_div(dim_t, 2) / self.embedding_dim)
dim_t = self.temperature ** (2 * torch.div(dim_t, 2, rounding_mode="floor") / self.embedding_dim)
pos_x = x_embed[:, :, :, None] / dim_t
pos_y = y_embed[:, :, :, None] / dim_t

3
src/transformers/models/hubert/modeling_hubert.py
View File

@ -27,7 +27,6 @@ from ...activations import ACT2FN
from ...deepspeed import is_deepspeed_zero3_enabled
from ...modeling_outputs import BaseModelOutput, CausalLMOutput, SequenceClassifierOutput
from ...modeling_utils import PreTrainedModel
from ...pytorch_utils import torch_int_div
from ...utils import (
add_code_sample_docstrings,
add_start_docstrings,
@ -871,7 +870,7 @@ class HubertPreTrainedModel(PreTrainedModel):
def _conv_out_length(input_length, kernel_size, stride):
# 1D convolutional layer output length formula taken
# from https://pytorch.org/docs/stable/generated/torch.nn.Conv1d.html
return torch_int_div(input_length - kernel_size, stride) + 1
return torch.div(input_length - kernel_size, stride, rounding_mode="floor") + 1
for kernel_size, stride in zip(self.config.conv_kernel, self.config.conv_stride):
input_lengths = _conv_out_length(input_lengths, kernel_size, stride)

8
src/transformers/models/layoutlmv2/modeling_layoutlmv2.py
View File

@ -31,7 +31,7 @@ from ...modeling_outputs import (
TokenClassifierOutput,
)
from ...modeling_utils import PreTrainedModel
from ...pytorch_utils import apply_chunking_to_forward, torch_int_div
from ...pytorch_utils import apply_chunking_to_forward
from ...utils import (
add_start_docstrings,
add_start_docstrings_to_model_forward,
@ -770,7 +770,7 @@ class LayoutLMv2Model(LayoutLMv2PreTrainedModel):
return embeddings
def _calc_visual_bbox(self, image_feature_pool_shape, bbox, device, final_shape):
visual_bbox_x = torch_int_div(
visual_bbox_x = torch.div(
torch.arange(
0,
1000 * (image_feature_pool_shape[1] + 1),
@ -779,8 +779,9 @@ class LayoutLMv2Model(LayoutLMv2PreTrainedModel):
dtype=bbox.dtype,
),
self.config.image_feature_pool_shape[1],
rounding_mode="floor",
)
visual_bbox_y = torch_int_div(
visual_bbox_y = torch.div(
torch.arange(
0,
1000 * (self.config.image_feature_pool_shape[0] + 1),
@ -789,6 +790,7 @@ class LayoutLMv2Model(LayoutLMv2PreTrainedModel):
dtype=bbox.dtype,
),
self.config.image_feature_pool_shape[0],
rounding_mode="floor",
)
visual_bbox = torch.stack(
[

4
src/transformers/models/mask2former/image_processing_mask2former.py
View File

@ -57,8 +57,6 @@ if is_torch_available():
import torch
from torch import nn
from ...pytorch_utils import torch_int_div
# Copied from transformers.models.detr.image_processing_detr.max_across_indices
def max_across_indices(values: Iterable[Any]) -> List[Any]:
@ -1009,7 +1007,7 @@ class Mask2FormerImageProcessor(BaseImageProcessor):
scores_per_image, topk_indices = scores.flatten(0, 1).topk(num_queries, sorted=False)
labels_per_image = labels[topk_indices]
topk_indices = torch_int_div(topk_indices, num_classes)
topk_indices = torch.div(topk_indices, num_classes, rounding_mode="floor")
mask_pred = mask_pred[topk_indices]
pred_masks = (mask_pred > 0).float()

4
src/transformers/models/maskformer/image_processing_maskformer.py
View File

@ -61,8 +61,6 @@ if is_torch_available():
import torch
from torch import nn
from ...pytorch_utils import torch_int_div
# Copied from transformers.models.detr.image_processing_detr.max_across_indices
def max_across_indices(values: Iterable[Any]) -> List[Any]:
@ -1077,7 +1075,7 @@ class MaskFormerImageProcessor(BaseImageProcessor):
scores_per_image, topk_indices = scores.flatten(0, 1).topk(num_queries, sorted=False)
labels_per_image = labels[topk_indices]
topk_indices = torch_int_div(topk_indices, num_classes)
topk_indices = torch.div(topk_indices, num_classes, rounding_mode="floor")
mask_pred = mask_pred[topk_indices]
pred_masks = (mask_pred > 0).float()

7
src/transformers/models/mctct/modeling_mctct.py
View File

@ -33,19 +33,12 @@ from ...modeling_utils import (
find_pruneable_heads_and_indices,
prune_linear_layer,
)
from ...pytorch_utils import is_torch_less_than_1_9
from ...utils import logging
from .configuration_mctct import MCTCTConfig
logger = logging.get_logger(__name__)
if is_torch_less_than_1_9:
logger.warning(
f"You are using torch=={torch.__version__}, but torch>=1.9.0 is required to use MCTCTModel. Please upgrade"
" torch."
)
_HIDDEN_STATES_START_POSITION = 1
_CONFIG_FOR_DOC = "MCTCTConfig"

4
src/transformers/models/oneformer/image_processing_oneformer.py
View File

@ -58,8 +58,6 @@ if is_torch_available():
import torch
from torch import nn
from ...pytorch_utils import torch_int_div
# Copied from transformers.models.detr.image_processing_detr.max_across_indices
def max_across_indices(values: Iterable[Any]) -> List[Any]:
@ -1122,7 +1120,7 @@ class OneFormerImageProcessor(BaseImageProcessor):
scores_per_image, topk_indices = scores.flatten(0, 1).topk(num_queries, sorted=False)
labels_per_image = labels[topk_indices]
topk_indices = torch_int_div(topk_indices, num_classes)
topk_indices = torch.div(topk_indices, num_classes, rounding_mode="floor")
# mask_pred = mask_pred.unsqueeze(1).repeat(1, self.sem_seg_head.num_classes, 1).flatten(0, 1)
mask_pred = masks_queries_logits[i][topk_indices]

3
src/transformers/models/sew/modeling_sew.py
View File

@ -28,7 +28,6 @@ from ...activations import ACT2FN
from ...deepspeed import is_deepspeed_zero3_enabled
from ...modeling_outputs import BaseModelOutput, CausalLMOutput, SequenceClassifierOutput
from ...modeling_utils import PreTrainedModel
from ...pytorch_utils import torch_int_div
from ...utils import add_code_sample_docstrings, add_start_docstrings, add_start_docstrings_to_model_forward, logging
from .configuration_sew import SEWConfig
@ -770,7 +769,7 @@ class SEWPreTrainedModel(PreTrainedModel):
def _conv_out_length(input_length, kernel_size, stride):
# 1D convolutional layer output length formula taken
# from https://pytorch.org/docs/stable/generated/torch.nn.Conv1d.html
return torch_int_div(input_length - kernel_size, stride) + 1
return torch.div(input_length - kernel_size, stride, rounding_mode="floor") + 1
for kernel_size, stride in zip(self.config.conv_kernel, self.config.conv_stride):
input_lengths = _conv_out_length(input_lengths, kernel_size, stride)

4
src/transformers/models/sew_d/modeling_sew_d.py
View File

@ -29,7 +29,7 @@ from ...activations import ACT2FN
from ...deepspeed import is_deepspeed_zero3_enabled
from ...modeling_outputs import BaseModelOutput, CausalLMOutput, SequenceClassifierOutput
from ...modeling_utils import PreTrainedModel
from ...pytorch_utils import softmax_backward_data, torch_int_div
from ...pytorch_utils import softmax_backward_data
from ...utils import add_code_sample_docstrings, add_start_docstrings, add_start_docstrings_to_model_forward, logging
from .configuration_sew_d import SEWDConfig
@ -1305,7 +1305,7 @@ class SEWDPreTrainedModel(PreTrainedModel):
def _conv_out_length(input_length, kernel_size, stride):
# 1D convolutional layer output length formula taken
# from https://pytorch.org/docs/stable/generated/torch.nn.Conv1d.html
return torch_int_div(input_length - kernel_size, stride) + 1
return torch.div(input_length - kernel_size, stride, rounding_mode="floor") + 1
for kernel_size, stride in zip(self.config.conv_kernel, self.config.conv_stride):
input_lengths = _conv_out_length(input_lengths, kernel_size, stride)

3
src/transformers/models/speecht5/modeling_speecht5.py
View File

@ -34,7 +34,6 @@ from ...modeling_outputs import (
Seq2SeqSpectrogramOutput,
)
from ...modeling_utils import PreTrainedModel
from ...pytorch_utils import torch_int_div
from ...utils import add_start_docstrings, add_start_docstrings_to_model_forward, logging, replace_return_docstrings
from .configuration_speecht5 import SpeechT5Config, SpeechT5HifiGanConfig
@ -620,7 +619,7 @@ class SpeechT5SpeechEncoderPrenet(nn.Module):
def _conv_out_length(input_length, kernel_size, stride):
# 1D convolutional layer output length formula taken
# from https://pytorch.org/docs/stable/generated/torch.nn.Conv1d.html
return torch_int_div(input_length - kernel_size, stride) + 1
return torch.div(input_length - kernel_size, stride, rounding_mode="floor") + 1
for kernel_size, stride in zip(self.config.conv_kernel, self.config.conv_stride):
input_lengths = _conv_out_length(input_lengths, kernel_size, stride)

3
src/transformers/models/table_transformer/modeling_table_transformer.py
View File

@ -26,7 +26,6 @@ from torch import Tensor, nn
from ...activations import ACT2FN
from ...modeling_outputs import BaseModelOutput, BaseModelOutputWithCrossAttentions, Seq2SeqModelOutput
from ...modeling_utils import PreTrainedModel
from ...pytorch_utils import torch_int_div
from ...utils import (
ModelOutput,
add_start_docstrings,
@ -380,7 +379,7 @@ class TableTransformerSinePositionEmbedding(nn.Module):
x_embed = x_embed / (x_embed[:, :, -1:] + 1e-6) * self.scale
dim_t = torch.arange(self.embedding_dim, dtype=torch.float32, device=pixel_values.device)
dim_t = self.temperature ** (2 * torch_int_div(dim_t, 2) / self.embedding_dim)
dim_t = self.temperature ** (2 * torch.div(dim_t, 2, rounding_mode="floor") / self.embedding_dim)
pos_x = x_embed[:, :, :, None] / dim_t
pos_y = y_embed[:, :, :, None] / dim_t

3
src/transformers/models/tapas/modeling_tapas.py
View File

@ -33,7 +33,6 @@ from ...pytorch_utils import (
apply_chunking_to_forward,
find_pruneable_heads_and_indices,
prune_linear_layer,
torch_int_div,
)
from ...utils import (
ModelOutput,
@ -1637,7 +1636,7 @@ class ProductIndexMap(IndexMap):
def project_outer(self, index):
"""Projects an index with the same index set onto the outer components."""
indices = torch_int_div(index.indices, self.inner_index.num_segments).type(torch.long)
indices = torch.div(index.indices, self.inner_index.num_segments, rounding_mode="floor").type(torch.long)
return IndexMap(indices=indices, num_segments=self.outer_index.num_segments, batch_dims=index.batch_dims)
def project_inner(self, index):

3
src/transformers/models/unispeech/modeling_unispeech.py
View File

@ -29,7 +29,6 @@ from ...activations import ACT2FN
from ...deepspeed import is_deepspeed_zero3_enabled
from ...modeling_outputs import BaseModelOutput, CausalLMOutput, SequenceClassifierOutput, Wav2Vec2BaseModelOutput
from ...modeling_utils import PreTrainedModel
from ...pytorch_utils import torch_int_div
from ...utils import (
ModelOutput,
add_code_sample_docstrings,
@ -981,7 +980,7 @@ class UniSpeechPreTrainedModel(PreTrainedModel):
def _conv_out_length(input_length, kernel_size, stride):
# 1D convolutional layer output length formula taken
# from https://pytorch.org/docs/stable/generated/torch.nn.Conv1d.html
return torch_int_div(input_length - kernel_size, stride) + 1
return torch.div(input_length - kernel_size, stride, rounding_mode="floor") + 1
for kernel_size, stride in zip(self.config.conv_kernel, self.config.conv_stride):
input_lengths = _conv_out_length(input_lengths, kernel_size, stride)

3
src/transformers/models/unispeech_sat/modeling_unispeech_sat.py
View File

@ -36,7 +36,6 @@ from ...modeling_outputs import (
XVectorOutput,
)
from ...modeling_utils import PreTrainedModel
from ...pytorch_utils import torch_int_div
from ...utils import (
ModelOutput,
add_code_sample_docstrings,
@ -995,7 +994,7 @@ class UniSpeechSatPreTrainedModel(PreTrainedModel):
def _conv_out_length(input_length, kernel_size, stride):
# 1D convolutional layer output length formula taken
# from https://pytorch.org/docs/stable/generated/torch.nn.Conv1d.html
return torch_int_div(input_length - kernel_size, stride) + 1
return torch.div(input_length - kernel_size, stride, rounding_mode="floor") + 1
for kernel_size, stride in zip(self.config.conv_kernel, self.config.conv_stride):
input_lengths = _conv_out_length(input_lengths, kernel_size, stride)

3
src/transformers/models/wav2vec2/modeling_wav2vec2.py
View File

@ -37,7 +37,6 @@ from ...modeling_outputs import (
XVectorOutput,
)
from ...modeling_utils import PreTrainedModel
from ...pytorch_utils import torch_int_div
from ...utils import (
ModelOutput,
add_code_sample_docstrings,
@ -1098,7 +1097,7 @@ class Wav2Vec2PreTrainedModel(PreTrainedModel):
def _conv_out_length(input_length, kernel_size, stride):
# 1D convolutional layer output length formula taken
# from https://pytorch.org/docs/stable/generated/torch.nn.Conv1d.html
return torch_int_div(input_length - kernel_size, stride) + 1
return torch.div(input_length - kernel_size, stride, rounding_mode="floor") + 1
for kernel_size, stride in zip(self.config.conv_kernel, self.config.conv_stride):
input_lengths = _conv_out_length(input_lengths, kernel_size, stride)

3
src/transformers/models/wav2vec2_conformer/modeling_wav2vec2_conformer.py
View File

@ -35,7 +35,6 @@ from ...modeling_outputs import (
XVectorOutput,
)
from ...modeling_utils import PreTrainedModel
from ...pytorch_utils import torch_int_div
from ...utils import (
ModelOutput,
add_code_sample_docstrings,
@ -1143,7 +1142,7 @@ class Wav2Vec2ConformerPreTrainedModel(PreTrainedModel):
def _conv_out_length(input_length, kernel_size, stride):
# 1D convolutional layer output length formula taken
# from https://pytorch.org/docs/stable/generated/torch.nn.Conv1d.html
return torch_int_div(input_length - kernel_size, stride) + 1
return torch.div(input_length - kernel_size, stride, rounding_mode="floor") + 1
for kernel_size, stride in zip(self.config.conv_kernel, self.config.conv_stride):
input_lengths = _conv_out_length(input_lengths, kernel_size, stride)

3
src/transformers/models/wavlm/modeling_wavlm.py
View File

@ -36,7 +36,6 @@ from ...modeling_outputs import (
XVectorOutput,
)
from ...modeling_utils import PreTrainedModel
from ...pytorch_utils import torch_int_div
from ...utils import add_code_sample_docstrings, add_start_docstrings, add_start_docstrings_to_model_forward, logging
from .configuration_wavlm import WavLMConfig
@ -1019,7 +1018,7 @@ class WavLMPreTrainedModel(PreTrainedModel):
def _conv_out_length(input_length, kernel_size, stride):
# 1D convolutional layer output length formula taken
# from https://pytorch.org/docs/stable/generated/torch.nn.Conv1d.html
return torch_int_div(input_length - kernel_size, stride) + 1
return torch.div(input_length - kernel_size, stride, rounding_mode="floor") + 1
for kernel_size, stride in zip(self.config.conv_kernel, self.config.conv_stride):
input_lengths = _conv_out_length(input_lengths, kernel_size, stride)

4
src/transformers/onnx/convert.py
View File

@ -26,7 +26,6 @@ from ..utils import (
TensorType,
is_tf_available,
is_torch_available,
is_torch_onnx_dict_inputs_support_available,
logging,
)
from .config import OnnxConfig
@ -339,9 +338,6 @@ def export(
if is_torch_available():
from ..utils import torch_version
if not is_torch_onnx_dict_inputs_support_available():
raise AssertionError(f"Unsupported PyTorch version, minimum required is 1.8.0, got: {torch_version}")
if not config.is_torch_support_available:
logger.warning(
f"Unsupported PyTorch version for this model. Minimum required is {config.torch_onnx_minimum_version},"

12
src/transformers/pytorch_utils.py
View File

@ -27,22 +27,10 @@ logger = logging.get_logger(__name__)
parsed_torch_version_base = version.parse(version.parse(torch.__version__).base_version)
is_torch_less_than_1_8 = parsed_torch_version_base < version.parse("1.8.0")
is_torch_less_than_1_9 = parsed_torch_version_base < version.parse("1.9.0")
is_torch_greater_or_equal_than_1_10 = parsed_torch_version_base >= version.parse("1.10")
is_torch_less_than_1_11 = parsed_torch_version_base < version.parse("1.11")
def torch_int_div(tensor1, tensor2):
"""
A function that performs integer division across different versions of PyTorch.
"""
if is_torch_less_than_1_8:
return tensor1 // tensor2
else:
return torch.div(tensor1, tensor2, rounding_mode="floor")
def softmax_backward_data(parent, grad_output, output, dim, self):
"""
A function that calls the internal `_softmax_backward_data` PyTorch method and that adjusts the arguments according

1
src/transformers/utils/__init__.py
View File

@ -151,7 +151,6 @@ from .import_utils import (
is_torch_fx_available,
is_torch_fx_proxy,
is_torch_neuroncore_available,
is_torch_onnx_dict_inputs_support_available,
is_torch_tensorrt_fx_available,
is_torch_tf32_available,
is_torch_tpu_available,

9
src/transformers/utils/import_utils.py
View File

@ -276,7 +276,6 @@ except importlib_metadata.PackageNotFoundError:
# 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")
TORCH_ONNX_DICT_INPUTS_MINIMUM_VERSION = version.parse("1.8")
def is_kenlm_available():
@ -388,7 +387,7 @@ def is_torch_tf32_available():
torch_version = None
_torch_fx_available = _torch_onnx_dict_inputs_support_available = False
_torch_fx_available = False
if _torch_available:
torch_version = version.parse(importlib_metadata.version("torch"))
_torch_fx_available = (torch_version.major, torch_version.minor) >= (
@ -396,8 +395,6 @@ if _torch_available:
TORCH_FX_REQUIRED_VERSION.minor,
)
_torch_onnx_dict_inputs_support_available = torch_version >= TORCH_ONNX_DICT_INPUTS_MINIMUM_VERSION
def is_torch_fx_available():
return _torch_fx_available
@ -407,10 +404,6 @@ def is_bs4_available():
return importlib.util.find_spec("bs4") is not None
def is_torch_onnx_dict_inputs_support_available():
return _torch_onnx_dict_inputs_support_available
def is_tf_available():
return _tf_available

9
tests/generation/test_beam_search.py
View File

@ -32,7 +32,6 @@ if is_torch_available():
DisjunctiveConstraint,
PhrasalConstraint,
)
from transformers.pytorch_utils import torch_int_div
class BeamSearchTester:
@ -161,7 +160,9 @@ class BeamSearchTester:
expected_output_scores = cut_expected_tensor(next_scores)
# add num_beams * batch_idx
offset = torch_int_div(torch.arange(self.num_beams * self.batch_size, device=torch_device), self.num_beams)
offset = torch.div(
torch.arange(self.num_beams * self.batch_size, device=torch_device), self.num_beams, rounding_mode="floor"
)
expected_output_indices = cut_expected_tensor(next_indices) + offset * self.num_beams
self.parent.assertListEqual(expected_output_tokens.tolist(), output_tokens.tolist())
@ -398,7 +399,9 @@ class ConstrainedBeamSearchTester:
expected_output_scores = cut_expected_tensor(next_scores)
# add num_beams * batch_idx
offset = torch_int_div(torch.arange(self.num_beams * self.batch_size, device=torch_device), self.num_beams)
offset = torch.div(
torch.arange(self.num_beams * self.batch_size, device=torch_device), self.num_beams, rounding_mode="floor"
)
expected_output_indices = cut_expected_tensor(next_indices) + offset * self.num_beams
self.parent.assertListEqual(expected_output_tokens.tolist(), output_tokens.tolist())

6
tests/models/bloom/test_modeling_bloom.py
View File

@ -38,10 +38,9 @@ if is_torch_available():
BloomModel,
BloomTokenizerFast,
)
from transformers.pytorch_utils import is_torch_greater_or_equal_than_1_10, is_torch_less_than_1_9
from transformers.pytorch_utils import is_torch_greater_or_equal_than_1_10
else:
is_torch_greater_or_equal_than_1_10 = False
is_torch_less_than_1_9 = True
@require_torch
@ -751,9 +750,6 @@ class BloomEmbeddingTest(unittest.TestCase):
self.assertAlmostEqual(EMBEDDINGS_DS_AFTER_LN[key][idx], output_dict_norm[key][idx], places=1)
@require_torch
@unittest.skipIf(
is_torch_less_than_1_9, reason="Test failed with torch < 1.9 (`min_cuda` not implemented for `BFloat16`)"
)
def test_hidden_states_transformers(self):
cuda_available = torch.cuda.is_available()
model = BloomModel.from_pretrained(self.path_bigscience_model, use_cache=False, torch_dtype="auto").to(

4
tests/models/mctct/test_modeling_mctct.py
View File

@ -32,9 +32,6 @@ if is_torch_available():
import torch
from transformers import MCTCTForCTC, MCTCTModel, MCTCTProcessor
from transformers.pytorch_utils import is_torch_less_than_1_9
else:
is_torch_less_than_1_9 = True
class MCTCTModelTester:
@ -265,7 +262,6 @@ class MCTCTModelTester:
@require_torch
@unittest.skipIf(is_torch_less_than_1_9, "MCTCT is only available in torch v1.9+")
class MCTCTModelTest(ModelTesterMixin, PipelineTesterMixin, unittest.TestCase):
all_model_classes = (MCTCTForCTC, MCTCTModel) if is_torch_available() else ()
pipeline_model_mapping = (

5
tests/models/swin/test_modeling_swin.py
View File

@ -33,9 +33,7 @@ if is_torch_available():
from transformers import SwinBackbone, SwinForImageClassification, SwinForMaskedImageModeling, SwinModel
from transformers.models.swin.modeling_swin import SWIN_PRETRAINED_MODEL_ARCHIVE_LIST
from transformers.pytorch_utils import is_torch_less_than_1_9
else:
is_torch_less_than_1_9 = True
if is_vision_available():
from PIL import Image
@ -266,7 +264,6 @@ class SwinModelTest(ModelTesterMixin, PipelineTesterMixin, unittest.TestCase):
def test_multi_gpu_data_parallel_forward(self):
pass
@unittest.skipIf(is_torch_less_than_1_9, reason="This test fails for SwinModel when torch < 1.9")
def test_training_gradient_checkpointing(self):
super().test_training_gradient_checkpointing()

4
tests/models/trajectory_transformer/test_modeling_trajectory_transformer.py
View File

@ -36,9 +36,6 @@ if is_torch_available():
from transformers.models.trajectory_transformer.modeling_trajectory_transformer import (
TRAJECTORY_TRANSFORMER_PRETRAINED_MODEL_ARCHIVE_LIST,
)
from transformers.pytorch_utils import is_torch_less_than_1_9
else:
is_torch_less_than_1_9 = True
class TrajectoryTransformerModelTester:
@ -199,7 +196,6 @@ class TrajectoryTransformerModelTest(ModelTesterMixin, GenerationTesterMixin, Pi
).loss
loss.backward()
@unittest.skipIf(is_torch_less_than_1_9, reason="This test fails for TrajectoryTransformerModel when torch < 1.9")
def test_training_gradient_checkpointing(self):
if not self.model_tester.is_training:
return

19
tests/models/wav2vec2/test_modeling_wav2vec2.py
View File

@ -71,9 +71,6 @@ if is_torch_available():
_compute_mask_indices,
_sample_negative_indices,
)
from transformers.pytorch_utils import is_torch_less_than_1_9, torch_int_div
else:
is_torch_less_than_1_9 = True
if is_torchaudio_available():
@ -1217,7 +1214,9 @@ class Wav2Vec2UtilsTest(unittest.TestCase):
sequence_length = 10
hidden_size = 4
num_negatives = 3
sequence = torch_int_div(torch.arange(sequence_length * hidden_size, device=torch_device), hidden_size)
sequence = torch.div(
torch.arange(sequence_length * hidden_size, device=torch_device), hidden_size, rounding_mode="floor"
)
features = sequence.view(sequence_length, hidden_size) # each value in vector consits of same value
features = features[None, :].expand(batch_size, sequence_length, hidden_size).contiguous()
@ -1245,7 +1244,9 @@ class Wav2Vec2UtilsTest(unittest.TestCase):
mask = torch.ones((batch_size, sequence_length), dtype=torch.long, device=torch_device)
mask[-1, sequence_length // 2 :] = 0
sequence = torch_int_div(torch.arange(sequence_length * hidden_size, device=torch_device), hidden_size)
sequence = torch.div(
torch.arange(sequence_length * hidden_size, device=torch_device), hidden_size, rounding_mode="floor"
)
features = sequence.view(sequence_length, hidden_size) # each value in vector consits of same value
features = features[None, :].expand(batch_size, sequence_length, hidden_size).contiguous()
@ -1651,10 +1652,6 @@ class Wav2Vec2ModelIntegrationTest(unittest.TestCase):
@require_pyctcdecode
@require_torchaudio
@unittest.skipIf(
is_torch_less_than_1_9,
reason="`torchaudio.functional.resample` needs torchaudio >= 0.9 which requires torch >= 0.9",
)
def test_wav2vec2_with_lm(self):
ds = load_dataset("common_voice", "es", split="test", streaming=True)
sample = next(iter(ds))
@ -1679,10 +1676,6 @@ class Wav2Vec2ModelIntegrationTest(unittest.TestCase):
@require_pyctcdecode
@require_torchaudio
@unittest.skipIf(
is_torch_less_than_1_9,
reason="`torchaudio.functional.resample` needs torchaudio >= 0.9 which requires torch >= 0.9",
)
def test_wav2vec2_with_lm_pool(self):
ds = load_dataset("common_voice", "es", split="test", streaming=True)
sample = next(iter(ds))

10
tests/onnx/test_onnx_v2.py
View File

@ -14,7 +14,6 @@ from transformers.onnx import (
OnnxConfig,
OnnxConfigWithPast,
ParameterFormat,
export,
validate_model_outputs,
)
from transformers.onnx.utils import (
@ -40,15 +39,6 @@ class OnnxUtilsTestCaseV2(TestCase):
Cover all the utilities involved to export ONNX models
"""
@require_torch
@patch("transformers.onnx.convert.is_torch_onnx_dict_inputs_support_available", return_value=False)
def test_ensure_pytorch_version_ge_1_8_0(self, mock_is_torch_onnx_dict_inputs_support_available):
"""
Ensure we raise an Exception if the pytorch version is unsupported (< 1.8.0)
"""
self.assertRaises(AssertionError, export, None, None, None, None, None)
mock_is_torch_onnx_dict_inputs_support_available.assert_called()
def test_compute_effective_axis_dimension(self):
"""
When exporting ONNX model with dynamic axis (batch or sequence) we set batch_size and/or sequence_length = -1.