feat(transformers) Pipeline class-classification; Model ViT (#1178)

* init pipelines `image-classification`, `image-segmentation`

* add vit for img_cls pipeline testing

* debug, passed vit generate

* passed pipeline; add ut

* debug; vit passed ut

* fix bug for img classification

* add copyright

* update after merging master

* delete jit decorator

Author: chenyingshu

mindone/transformers/__init__.py

@@ -493,6 +493,7 @@
     UMT5Model,
     UMT5PreTrainedModel,
 )
+from .models.vit import ViTForImageClassification, ViTForMaskedImageModeling, ViTModel, ViTPreTrainedModel
 from .models.vits import VitsModel, VitsPreTrainedModel
 from .models.wav2vec2 import (
     Wav2Vec2FeatureExtractor,

mindone/transformers/models/__init__.py

@@ -84,6 +84,7 @@
     switch_transformers,
     t5,
     umt5,
+    vit,
     vits,
     wav2vec2,
     xlm_roberta,

mindone/transformers/models/auto/configuration_auto.py

@@ -107,6 +107,7 @@
         ("speecht5", "SpeechT5Config"),
         ("t5", "T5Config"),
         ("umt5", "UMT5Config"),
+        ("vit", "ViTConfig"),
         ("wav2vec2", "Wav2Vec2Config"),
         ("mvp", "MvpConfig"),
         ("whisper", "WhisperConfig"),
@@ -200,6 +201,7 @@
         ("t5", "T5"),
         ("t5v1.1", "T5v1.1"),
         ("umt5", "UMT5"),
+        ("vit", "ViT"),
         ("wav2vec2", "Wav2Vec2"),
         ("whisper", "Whisper"),
         ("convbert", "ConvBERT"),

mindone/transformers/models/auto/modeling_auto.py

@@ -103,6 +103,7 @@
         ("speecht5", "SpeechT5Model"),
         ("t5", "T5Model"),
         ("umt5", "UMT5Model"),
+        ("vit", "ViTModel"),
         ("wav2vec2", "Wav2Vec2Model"),
         ("whisper", "WhisperModel"),
         ("xlm-roberta", "XLMRobertaModel"),
@@ -223,10 +224,15 @@
         ("imagegpt", "ImageGPTModel"),
         ("levit", "LevitModel"),
         ("siglip_vision_model", "SiglipVisionModel"),
+        ("vit", "ViTModel"),
     ]
 )
 
-MODEL_FOR_MASKED_IMAGE_MODELING_MAPPING_NAMES = OrderedDict()
+MODEL_FOR_MASKED_IMAGE_MODELING_MAPPING_NAMES = OrderedDict(
+    [
+        ("vit", "ViTForMaskedImageModeling"),
+    ]
+)
 
 
 MODEL_FOR_CAUSAL_IMAGE_MODELING_MAPPING_NAMES = OrderedDict(
@@ -248,6 +254,7 @@
             ("LevitForImageClassification", "LevitForImageClassificationWithTeacher"),
         ),
         ("siglip", "SiglipForImageClassification"),
+        ("vit", "ViTForImageClassification"),
     ]
 )
 

mindone/transformers/models/vit/__init__.py

@@ -0,0 +1,17 @@
+# Copyright 2024 The HuggingFace Team. All rights reserved.
+#
+# This code is adapted from https://github.com/huggingface/transformers
+# with modifications to run transformers on mindspore.
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+from .modeling_vit import *