Add TVLT (#20725)

* Update image_processing_tvlt.py

* Update modeling_tvlt.py

* Update

* Update modeling_tvlt.py

* Create tvlt.mdx

* Update configuration_tvlt.py

* Update modeling_tvlt.py

* Update test_modeling_tvlt.py

* Update modeling_tvlt.py

* Update modeling_tvlt.py

* Update image_processing_tvlt.py

* Update feature_extraction_tvlt.py

* Update tvlt models

* Update tests

* Update

* Update

* Update tests

* Update README_ko.md

* Update README_ja.md

* Update README_ko.md

* Update README_zh-hans.md

* Update docs/source/en/model_doc/tvlt.mdx
Co-authored-by: NielsRogge <48327001+NielsRogge@users.noreply.github.com>

* Update docs/source/en/model_doc/tvlt.mdx
Co-authored-by: NielsRogge <48327001+NielsRogge@users.noreply.github.com>

* Update src/transformers/models/tvlt/configuration_tvlt.py
Co-authored-by: NielsRogge <48327001+NielsRogge@users.noreply.github.com>

* Update src/transformers/models/tvlt/configuration_tvlt.py
Co-authored-by: NielsRogge <48327001+NielsRogge@users.noreply.github.com>

* Update src/transformers/models/tvlt/configuration_tvlt.py
Co-authored-by: NielsRogge <48327001+NielsRogge@users.noreply.github.com>

* Update src/transformers/models/tvlt/configuration_tvlt.py
Co-authored-by: NielsRogge <48327001+NielsRogge@users.noreply.github.com>

* Update tvlt.mdx

* Update modeling_tvlt.py

* Update configuration_tvlt.py

* Update modeling_tvlt.py

* Update modeling_tvlt.py

* Update modeling_tvlt.py

* Update modeling_tvlt.py

* Add files via upload

* Update model

* Update modeling_tvlt.py

* Update tvlt models

* Update src/transformers/models/tvlt/__init__.py
Co-authored-by: amyeroberts <22614925+amyeroberts@users.noreply.github.com>

* Update src/transformers/models/tvlt/__init__.py
Co-authored-by: amyeroberts <22614925+amyeroberts@users.noreply.github.com>

* Update src/transformers/models/tvlt/image_processing_tvlt.py
Co-authored-by: amyeroberts <22614925+amyeroberts@users.noreply.github.com>

* Update src/transformers/models/tvlt/image_processing_tvlt.py
Co-authored-by: amyeroberts <22614925+amyeroberts@users.noreply.github.com>

* Add files via upload

* Add files via upload

* Delete modeling_tvlt.py

* Delete feature_extraction_tvlt.py

* Delete configuration_tvlt.py

* Delete image_processing_tvlt.py

* Delete processing_tvlt.py

* Update tvlt

* Update src/transformers/models/tvlt/configuration_tvlt.py
Co-authored-by: Sylvain Gugger <35901082+sgugger@users.noreply.github.com>

* Update src/transformers/models/tvlt/image_processing_tvlt.py
Co-authored-by: Arthur <48595927+ArthurZucker@users.noreply.github.com>

* Update src/transformers/models/tvlt/configuration_tvlt.py
Co-authored-by: Arthur <48595927+ArthurZucker@users.noreply.github.com>

* Update tests/models/tvlt/test_modeling_tvlt.py
Co-authored-by: Sylvain Gugger <35901082+sgugger@users.noreply.github.com>

* Update tests/models/tvlt/test_modeling_tvlt.py
Co-authored-by: Sylvain Gugger <35901082+sgugger@users.noreply.github.com>

* Update src/transformers/models/tvlt/modeling_tvlt.py
Co-authored-by: Sylvain Gugger <35901082+sgugger@users.noreply.github.com>

* Update src/transformers/models/tvlt/modeling_tvlt.py
Co-authored-by: Sylvain Gugger <35901082+sgugger@users.noreply.github.com>

* Update README.md
Co-authored-by: Sylvain Gugger <35901082+sgugger@users.noreply.github.com>

* Update README_es.md

* Update README_hd.md

* Update README_ja.md

* Update README_ko.md

* Update README_zh-hans.md

* Update README_zh-hant.md

* Update index.mdx

* Update tvlt.mdx

* Update tvlt.mdx

* Update configuration_tvlt.py

* Update src/transformers/models/tvlt/image_processing_tvlt.py
Co-authored-by: Sylvain Gugger <35901082+sgugger@users.noreply.github.com>

* Update src/transformers/models/tvlt/image_processing_tvlt.py
Co-authored-by: Sylvain Gugger <35901082+sgugger@users.noreply.github.com>

* Update src/transformers/models/tvlt/image_processing_tvlt.py
Co-authored-by: Sylvain Gugger <35901082+sgugger@users.noreply.github.com>

* Update src/transformers/models/tvlt/image_processing_tvlt.py
Co-authored-by: Sylvain Gugger <35901082+sgugger@users.noreply.github.com>

* Update src/transformers/models/tvlt/modeling_tvlt.py
Co-authored-by: Sylvain Gugger <35901082+sgugger@users.noreply.github.com>

* Update modeling_tvlt.py

* Add files via upload

* Update tvlt.mdx

* Update modeling_auto.py

* Add files via upload

* Add files via upload

* Update dummy_pt_objects.py

* Update __init__.py

* Update feature_extraction_tvlt.py

* Update feature_extraction_tvlt.py

* Update image_processing_tvlt.py

* Update modeling_auto.py

* Update test_feature_extraction_tvlt.py

* Update test_processor_tvlt.py

* Update test_feature_extraction_tvlt.py

* Add files via upload

* Update test_image_processor_tvlt.py

* Update tests/models/tvlt/test_processor_tvlt.py
Co-authored-by: amyeroberts <22614925+amyeroberts@users.noreply.github.com>

* Update src/transformers/models/tvlt/modeling_tvlt.py
Co-authored-by: NielsRogge <48327001+NielsRogge@users.noreply.github.com>

* Update src/transformers/models/tvlt/processing_tvlt.py
Co-authored-by: amyeroberts <22614925+amyeroberts@users.noreply.github.com>

* Update src/transformers/models/tvlt/modeling_tvlt.py
Co-authored-by: amyeroberts <22614925+amyeroberts@users.noreply.github.com>

* Update src/transformers/models/tvlt/modeling_tvlt.py
Co-authored-by: amyeroberts <22614925+amyeroberts@users.noreply.github.com>

* Update tests/models/tvlt/test_image_processor_tvlt.py
Co-authored-by: amyeroberts <22614925+amyeroberts@users.noreply.github.com>

* Update src/transformers/models/tvlt/modeling_tvlt.py
Co-authored-by: NielsRogge <48327001+NielsRogge@users.noreply.github.com>

* Update tests/models/tvlt/test_image_processor_tvlt.py
Co-authored-by: amyeroberts <22614925+amyeroberts@users.noreply.github.com>

* Update src/transformers/models/tvlt/modeling_tvlt.py
Co-authored-by: amyeroberts <22614925+amyeroberts@users.noreply.github.com>

* Update tests/models/tvlt/test_image_processor_tvlt.py
Co-authored-by: amyeroberts <22614925+amyeroberts@users.noreply.github.com>

* Update tests/models/tvlt/test_image_processor_tvlt.py
Co-authored-by: amyeroberts <22614925+amyeroberts@users.noreply.github.com>

* Update tests/models/tvlt/test_modeling_tvlt.py
Co-authored-by: amyeroberts <22614925+amyeroberts@users.noreply.github.com>

* Update src/transformers/models/tvlt/modeling_tvlt.py
Co-authored-by: NielsRogge <48327001+NielsRogge@users.noreply.github.com>

* Update src/transformers/models/tvlt/modeling_tvlt.py
Co-authored-by: NielsRogge <48327001+NielsRogge@users.noreply.github.com>

* Update src/transformers/models/tvlt/configuration_tvlt.py
Co-authored-by: amyeroberts <22614925+amyeroberts@users.noreply.github.com>

* Update tests/models/tvlt/test_feature_extraction_tvlt.py
Co-authored-by: amyeroberts <22614925+amyeroberts@users.noreply.github.com>

* Update src/transformers/models/tvlt/processing_tvlt.py
Co-authored-by: amyeroberts <22614925+amyeroberts@users.noreply.github.com>

* Update docs/source/en/model_doc/tvlt.mdx
Co-authored-by: amyeroberts <22614925+amyeroberts@users.noreply.github.com>

* Update src/transformers/models/tvlt/configuration_tvlt.py
Co-authored-by: amyeroberts <22614925+amyeroberts@users.noreply.github.com>

* Update src/transformers/models/tvlt/configuration_tvlt.py
Co-authored-by: amyeroberts <22614925+amyeroberts@users.noreply.github.com>

* Update src/transformers/models/tvlt/configuration_tvlt.py
Co-authored-by: amyeroberts <22614925+amyeroberts@users.noreply.github.com>

* Update src/transformers/models/tvlt/configuration_tvlt.py
Co-authored-by: amyeroberts <22614925+amyeroberts@users.noreply.github.com>

* Update src/transformers/models/tvlt/configuration_tvlt.py
Co-authored-by: amyeroberts <22614925+amyeroberts@users.noreply.github.com>

* Update src/transformers/models/tvlt/configuration_tvlt.py
Co-authored-by: amyeroberts <22614925+amyeroberts@users.noreply.github.com>

* Update src/transformers/models/tvlt/configuration_tvlt.py
Co-authored-by: amyeroberts <22614925+amyeroberts@users.noreply.github.com>

* Update src/transformers/models/tvlt/configuration_tvlt.py
Co-authored-by: amyeroberts <22614925+amyeroberts@users.noreply.github.com>

* Update src/transformers/models/tvlt/configuration_tvlt.py
Co-authored-by: amyeroberts <22614925+amyeroberts@users.noreply.github.com>

* Update src/transformers/models/tvlt/configuration_tvlt.py
Co-authored-by: amyeroberts <22614925+amyeroberts@users.noreply.github.com>

* Update src/transformers/models/tvlt/feature_extraction_tvlt.py
Co-authored-by: amyeroberts <22614925+amyeroberts@users.noreply.github.com>

* Update src/transformers/models/tvlt/configuration_tvlt.py
Co-authored-by: amyeroberts <22614925+amyeroberts@users.noreply.github.com>

* Update src/transformers/models/tvlt/feature_extraction_tvlt.py
Co-authored-by: amyeroberts <22614925+amyeroberts@users.noreply.github.com>

* Update src/transformers/models/tvlt/feature_extraction_tvlt.py
Co-authored-by: amyeroberts <22614925+amyeroberts@users.noreply.github.com>

* Update src/transformers/models/tvlt/feature_extraction_tvlt.py
Co-authored-by: amyeroberts <22614925+amyeroberts@users.noreply.github.com>

* Update feature_extraction_tvlt.py

* Update feature_extraction_tvlt.py

* Update src/transformers/models/tvlt/image_processing_tvlt.py
Co-authored-by: amyeroberts <22614925+amyeroberts@users.noreply.github.com>

* Update src/transformers/models/tvlt/image_processing_tvlt.py
Co-authored-by: amyeroberts <22614925+amyeroberts@users.noreply.github.com>

* Update image_processing_tvlt.py

* Update src/transformers/models/tvlt/modeling_tvlt.py
Co-authored-by: amyeroberts <22614925+amyeroberts@users.noreply.github.com>

* Update modeling_tvlt.py

* Update modeling_tvlt.py

* Update modeling_tvlt.py

* Update test_image_processor_tvlt.py

* Update tests/models/tvlt/test_modeling_tvlt.py
Co-authored-by: amyeroberts <22614925+amyeroberts@users.noreply.github.com>

* Update tests/models/tvlt/test_modeling_tvlt.py
Co-authored-by: amyeroberts <22614925+amyeroberts@users.noreply.github.com>

* Update tests/models/tvlt/test_modeling_tvlt.py
Co-authored-by: amyeroberts <22614925+amyeroberts@users.noreply.github.com>

* Update tests/models/tvlt/test_modeling_tvlt.py
Co-authored-by: amyeroberts <22614925+amyeroberts@users.noreply.github.com>

* Update tests/models/tvlt/test_modeling_tvlt.py
Co-authored-by: amyeroberts <22614925+amyeroberts@users.noreply.github.com>

* Update tests/models/tvlt/test_modeling_tvlt.py
Co-authored-by: amyeroberts <22614925+amyeroberts@users.noreply.github.com>

* Update tests/models/tvlt/test_modeling_tvlt.py
Co-authored-by: amyeroberts <22614925+amyeroberts@users.noreply.github.com>

* Update tests/models/tvlt/test_modeling_tvlt.py
Co-authored-by: amyeroberts <22614925+amyeroberts@users.noreply.github.com>

* Update tests/models/tvlt/test_modeling_tvlt.py
Co-authored-by: amyeroberts <22614925+amyeroberts@users.noreply.github.com>

* Add files via upload

* Add files via upload

* Update modeling_tvlt.py

* Update modeling_tvlt.py

* Update modeling_tvlt.py

* Add files via upload

* Update docs/source/en/model_doc/tvlt.mdx
Co-authored-by: amyeroberts <22614925+amyeroberts@users.noreply.github.com>

* Update image_processing_tvlt.py

* Add files via upload

* Add files via upload

* Update tvlt.mdx

* Update docs/source/en/model_doc/tvlt.mdx
Co-authored-by: NielsRogge <48327001+NielsRogge@users.noreply.github.com>

* Update docs/source/en/model_doc/tvlt.mdx
Co-authored-by: NielsRogge <48327001+NielsRogge@users.noreply.github.com>

* Update src/transformers/models/tvlt/image_processing_tvlt.py
Co-authored-by: NielsRogge <48327001+NielsRogge@users.noreply.github.com>

* Update src/transformers/models/tvlt/image_processing_tvlt.py
Co-authored-by: NielsRogge <48327001+NielsRogge@users.noreply.github.com>

* Update docs/source/en/model_doc/tvlt.mdx
Co-authored-by: Sanchit Gandhi <93869735+sanchit-gandhi@users.noreply.github.com>

* Update docs/source/en/model_doc/tvlt.mdx
Co-authored-by: Sanchit Gandhi <93869735+sanchit-gandhi@users.noreply.github.com>

* Update src/transformers/models/tvlt/configuration_tvlt.py
Co-authored-by: Sanchit Gandhi <93869735+sanchit-gandhi@users.noreply.github.com>

* Add files via upload

* Add files via upload

* Add files via upload

* Add files via upload

* Update modeling_auto.py

* Update tvlt.mdx

* Update dummy_pt_objects.py

* Update feature_extraction_tvlt.py

* Update modeling_tvlt.py

* Update test_feature_extraction_tvlt.py

* Update test_image_processor_tvlt.py

* Update test_feature_extraction_tvlt.py

* Update modeling_tvlt.py

* Update dummy_pt_objects.py

* Update dummy_speech_objects.py

* Add files via upload

* Update README_hd.md

* Update modeling_tvlt.py

* Update modeling_tvlt.py

* Update modeling_tvlt.py

* Update modeling_tvlt.py

* Update modeling_tvlt.py

* Update modeling_tvlt.py

* Update test_modeling_tvlt.py

* Update src/transformers/models/tvlt/configuration_tvlt.py
Co-authored-by: amyeroberts <22614925+amyeroberts@users.noreply.github.com>

* Update src/transformers/models/tvlt/feature_extraction_tvlt.py
Co-authored-by: amyeroberts <22614925+amyeroberts@users.noreply.github.com>

* Update src/transformers/models/tvlt/image_processing_tvlt.py
Co-authored-by: amyeroberts <22614925+amyeroberts@users.noreply.github.com>

* Update src/transformers/models/tvlt/image_processing_tvlt.py
Co-authored-by: amyeroberts <22614925+amyeroberts@users.noreply.github.com>

* Update src/transformers/models/tvlt/modeling_tvlt.py
Co-authored-by: amyeroberts <22614925+amyeroberts@users.noreply.github.com>

* Update src/transformers/models/tvlt/image_processing_tvlt.py
Co-authored-by: amyeroberts <22614925+amyeroberts@users.noreply.github.com>

* Update src/transformers/models/tvlt/modeling_tvlt.py
Co-authored-by: amyeroberts <22614925+amyeroberts@users.noreply.github.com>

* Update MAE processing

* Update modeling_tvlt.py

* Update modeling_tvlt.py

* Update modeling

* Update style

* Update src/transformers/models/tvlt/modeling_tvlt.py
Co-authored-by: amyeroberts <22614925+amyeroberts@users.noreply.github.com>

* Update src/transformers/models/tvlt/modeling_tvlt.py
Co-authored-by: amyeroberts <22614925+amyeroberts@users.noreply.github.com>

* Update check_repo.py

* Update tvlt.mdx

* Update __init__.py

* Update tests

* Update tvlt models

* Update configuration_tvlt.py

* Update configuration_tvlt.py

* Update image_processing_tvlt.py

* Update dummy_pt_objects.py

* Add files via upload

* Update test_modeling_tvlt.py

* Update test_feature_extraction_tvlt.py

* Update test_feature_extraction_tvlt.py

* Update test_feature_extraction_tvlt.py

* Update test_feature_extraction_tvlt.py

* Update test_feature_extraction_tvlt.py

* Update test_feature_extraction_tvlt.py

---------
Co-authored-by: NielsRogge <48327001+NielsRogge@users.noreply.github.com>
Co-authored-by: amyeroberts <22614925+amyeroberts@users.noreply.github.com>
Co-authored-by: Sylvain Gugger <35901082+sgugger@users.noreply.github.com>
Co-authored-by: Arthur <48595927+ArthurZucker@users.noreply.github.com>
Co-authored-by: Sanchit Gandhi <93869735+sanchit-gandhi@users.noreply.github.com>

src/transformers/utils/dummy_pt_objects.py

@@ -6068,6 +6068,37 @@ class TrOCRPreTrainedModel(metaclass=DummyObject):
         requires_backends(self, ["torch"])
 
 
+TVLT_PRETRAINED_MODEL_ARCHIVE_LIST = None
+
+
+class TvltForAudioVisualClassification(metaclass=DummyObject):
+    _backends = ["torch"]
+
+    def __init__(self, *args, **kwargs):
+        requires_backends(self, ["torch"])
+
+
+class TvltForPreTraining(metaclass=DummyObject):
+    _backends = ["torch"]
+
+    def __init__(self, *args, **kwargs):
+        requires_backends(self, ["torch"])
+
+
+class TvltModel(metaclass=DummyObject):
+    _backends = ["torch"]
+
+    def __init__(self, *args, **kwargs):
+        requires_backends(self, ["torch"])
+
+
+class TvltPreTrainedModel(metaclass=DummyObject):
+    _backends = ["torch"]
+
+    def __init__(self, *args, **kwargs):
+        requires_backends(self, ["torch"])
+
+
 UNISPEECH_PRETRAINED_MODEL_ARCHIVE_LIST = None
 
 

src/transformers/utils/dummy_speech_objects.py

@@ -28,3 +28,10 @@ class SpeechT5FeatureExtractor(metaclass=DummyObject):
 
     def __init__(self, *args, **kwargs):
         requires_backends(self, ["speech"])
+
+
+class TvltFeatureExtractor(metaclass=DummyObject):
+    _backends = ["speech"]
+
+    def __init__(self, *args, **kwargs):
+        requires_backends(self, ["speech"])

src/transformers/utils/dummy_vision_objects.py

@@ -415,6 +415,13 @@ class Swin2SRImageProcessor(metaclass=DummyObject):
         requires_backends(self, ["vision"])
 
 
+class TvltImageProcessor(metaclass=DummyObject):
+    _backends = ["vision"]
+
+    def __init__(self, *args, **kwargs):
+        requires_backends(self, ["vision"])
+
+
 class VideoMAEFeatureExtractor(metaclass=DummyObject):
     _backends = ["vision"]
 

tests/models/tvlt/test_feature_extraction_tvlt.py

+# coding=utf-8
+# Copyright 2023 HuggingFace Inc.
+#
+# 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.
+""" Testing suite for the TVLT feature extraction. """
+
+import itertools
+import os
+import random
+import tempfile
+import unittest
+
+import numpy as np
+
+from transformers import is_datasets_available, is_speech_available
+from transformers.testing_utils import check_json_file_has_correct_format, require_torch, require_torchaudio
+from transformers.utils.import_utils import is_torch_available
+
+from ...test_sequence_feature_extraction_common import SequenceFeatureExtractionTestMixin
+
+
+if is_torch_available():
+    import torch
+
+if is_datasets_available():
+    from datasets import load_dataset
+
+if is_speech_available():
+    from transformers import TvltFeatureExtractor
+
+global_rng = random.Random()
+
+
+def floats_list(shape, scale=1.0, rng=None, name=None):
+    """Creates a random float32 tensor"""
+    if rng is None:
+        rng = global_rng
+
+    values = []
+    for batch_idx in range(shape[0]):
+        values.append([])
+        for _ in range(shape[1]):
+            values[-1].append(rng.random() * scale)
+
+    return values
+
+
+class TvltFeatureExtractionTester(unittest.TestCase):
+    def __init__(
+        self,
+        parent,
+        batch_size=7,
+        min_seq_length=400,
+        max_seq_length=2000,
+        spectrogram_length=2048,
+        feature_size=128,
+        num_audio_channels=1,
+        hop_length=512,
+        chunk_length=30,
+        sampling_rate=44100,
+    ):
+        self.parent = parent
+        self.batch_size = batch_size
+        self.min_seq_length = min_seq_length
+        self.max_seq_length = max_seq_length
+        self.seq_length_diff = (self.max_seq_length - self.min_seq_length) // (self.batch_size - 1)
+        self.spectrogram_length = spectrogram_length
+        self.feature_size = feature_size
+        self.num_audio_channels = num_audio_channels
+        self.hop_length = hop_length
+        self.chunk_length = chunk_length
+        self.sampling_rate = sampling_rate
+
+    def prepare_feat_extract_dict(self):
+        return {
+            "spectrogram_length": self.spectrogram_length,
+            "feature_size": self.feature_size,
+            "num_audio_channels": self.num_audio_channels,
+            "hop_length": self.hop_length,
+            "chunk_length": self.chunk_length,
+            "sampling_rate": self.sampling_rate,
+        }
+
+    def prepare_inputs_for_common(self, equal_length=False, numpify=False):
+        def _flatten(list_of_lists):
+            return list(itertools.chain(*list_of_lists))
+
+        if equal_length:
+            speech_inputs = [floats_list((self.max_seq_length, self.feature_size)) for _ in range(self.batch_size)]
+        else:
+            # make sure that inputs increase in size
+            speech_inputs = [
+                floats_list((x, self.feature_size))
+                for x in range(self.min_seq_length, self.max_seq_length, self.seq_length_diff)
+            ]
+        if numpify:
+            speech_inputs = [np.asarray(x) for x in speech_inputs]
+        return speech_inputs
+
+
+@require_torch
+@require_torchaudio
+class TvltFeatureExtractionTest(SequenceFeatureExtractionTestMixin, unittest.TestCase):
+    feature_extraction_class = TvltFeatureExtractor if is_speech_available() else None
+
+    def setUp(self):
+        self.feat_extract_tester = TvltFeatureExtractionTester(self)
+
+    def test_feat_extract_properties(self):
+        feature_extractor = self.feature_extraction_class(**self.feat_extract_dict)
+        self.assertTrue(hasattr(feature_extractor, "spectrogram_length"))
+        self.assertTrue(hasattr(feature_extractor, "feature_size"))
+        self.assertTrue(hasattr(feature_extractor, "num_audio_channels"))
+        self.assertTrue(hasattr(feature_extractor, "hop_length"))
+        self.assertTrue(hasattr(feature_extractor, "chunk_length"))
+        self.assertTrue(hasattr(feature_extractor, "sampling_rate"))
+
+    def test_feat_extract_from_and_save_pretrained(self):
+        feat_extract_first = self.feature_extraction_class(**self.feat_extract_dict)
+
+        with tempfile.TemporaryDirectory() as tmpdirname:
+            saved_file = feat_extract_first.save_pretrained(tmpdirname)[0]
+            check_json_file_has_correct_format(saved_file)
+            feat_extract_second = self.feature_extraction_class.from_pretrained(tmpdirname)
+
+        dict_first = feat_extract_first.to_dict()
+        dict_second = feat_extract_second.to_dict()
+        mel_1 = dict_first.pop("mel_filters")
+        mel_2 = dict_second.pop("mel_filters")
+        self.assertTrue(np.allclose(mel_1, mel_2))
+        self.assertEqual(dict_first, dict_second)
+
+    def test_feat_extract_to_json_file(self):
+        feat_extract_first = self.feature_extraction_class(**self.feat_extract_dict)
+
+        with tempfile.TemporaryDirectory() as tmpdirname:
+            json_file_path = os.path.join(tmpdirname, "feat_extract.json")
+            feat_extract_first.to_json_file(json_file_path)
+            feat_extract_second = self.feature_extraction_class.from_json_file(json_file_path)
+
+        dict_first = feat_extract_first.to_dict()
+        dict_second = feat_extract_second.to_dict()
+        mel_1 = dict_first.pop("mel_filters")
+        mel_2 = dict_second.pop("mel_filters")
+        self.assertTrue(np.allclose(mel_1, mel_2))
+        self.assertEqual(dict_first, dict_second)
+
+    def test_call(self):
+        # Initialize feature_extractor
+        feature_extractor = self.feature_extraction_class(**self.feat_extract_dict)
+
+        # create three inputs of length 800, 1000, and 1200
+        speech_inputs = [floats_list((1, x))[0] for x in range(8000, 14000, 20000)]
+        np_speech_inputs = [np.asarray(speech_input) for speech_input in speech_inputs]
+
+        # Test not batched input
+        encoded_audios = feature_extractor(np_speech_inputs[0], return_tensors="np", sampling_rate=44100).audio_values
+
+        self.assertTrue(encoded_audios.ndim == 4)
+        self.assertTrue(encoded_audios.shape[-1] == feature_extractor.feature_size)
+        self.assertTrue(encoded_audios.shape[-2] <= feature_extractor.spectrogram_length)
+        self.assertTrue(encoded_audios.shape[-3] == feature_extractor.num_channels)
+
+        # Test batched
+        encoded_audios = feature_extractor(np_speech_inputs, return_tensors="np", sampling_rate=44100).audio_values
+
+        self.assertTrue(encoded_audios.ndim == 4)
+        self.assertTrue(encoded_audios.shape[-1] == feature_extractor.feature_size)
+        self.assertTrue(encoded_audios.shape[-2] <= feature_extractor.spectrogram_length)
+        self.assertTrue(encoded_audios.shape[-3] == feature_extractor.num_channels)
+
+        # Test audio masking
+        encoded_audios = feature_extractor(
+            np_speech_inputs, return_tensors="np", sampling_rate=44100, mask_audio=True
+        ).audio_values
+
+        self.assertTrue(encoded_audios.ndim == 4)
+        self.assertTrue(encoded_audios.shape[-1] == feature_extractor.feature_size)
+        self.assertTrue(encoded_audios.shape[-2] <= feature_extractor.spectrogram_length)
+        self.assertTrue(encoded_audios.shape[-3] == feature_extractor.num_channels)
+
+    def _load_datasamples(self, num_samples):
+        ds = load_dataset("hf-internal-testing/librispeech_asr_dummy", "clean", split="validation")
+        # automatic decoding with librispeech
+        speech_samples = ds.sort("id").select(range(num_samples))[:num_samples]["audio"]
+
+        return [x["array"] for x in speech_samples]
+
+    def test_integration(self):
+        input_speech = self._load_datasamples(1)
+        feaure_extractor = TvltFeatureExtractor()
+        audio_values = feaure_extractor(input_speech, return_tensors="pt").audio_values
+
+        self.assertTrue(audio_values.shape, [1, 1, 192, 128])
+
+        expected_slice = torch.tensor([[-0.3032, -0.2708], [-0.4434, -0.4007]])
+        self.assertTrue(torch.allclose(audio_values[0, 0, :2, :2], expected_slice, atol=1e-4))

tests/models/tvlt/test_image_processor_tvlt.py

+# coding=utf-8
+# Copyright 2023 HuggingFace Inc.
+#
+# 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.
+""" Testing suite for the TVLT image processor. """
+
+import unittest
+
+import numpy as np
+
+from transformers.testing_utils import require_torch, require_vision
+from transformers.utils import is_torch_available, is_vision_available
+
+from ...test_image_processing_common import ImageProcessingSavingTestMixin
+
+
+if is_torch_available():
+    import torch
+
+if is_vision_available():
+    from PIL import Image
+
+    from transformers import TvltImageProcessor
+
+
+def prepare_video(image_processor_tester, width=10, height=10, numpify=False, torchify=False):
+    """This function prepares a video as a list of PIL images/NumPy arrays/PyTorch tensors."""
+
+    video = []
+    for i in range(image_processor_tester.num_frames):
+        video.append(np.random.randint(255, size=(image_processor_tester.num_channels, width, height), dtype=np.uint8))
+
+    if not numpify and not torchify:
+        # PIL expects the channel dimension as last dimension
+        video = [Image.fromarray(np.moveaxis(frame, 0, -1)) for frame in video]
+
+    if torchify:
+        video = [torch.from_numpy(frame) for frame in video]
+
+    return video
+
+
+def prepare_video_inputs(image_processor_tester, equal_resolution=False, numpify=False, torchify=False):
+    """This function prepares a batch of videos: a list of list of PIL images, or a list of list of numpy arrays if
+    one specifies numpify=True, or a list of list of PyTorch tensors if one specifies torchify=True.
+    One can specify whether the videos are of the same resolution or not.
+    """
+
+    assert not (numpify and torchify), "You cannot specify both numpy and PyTorch tensors at the same time"
+
+    video_inputs = []
+    for i in range(image_processor_tester.batch_size):
+        if equal_resolution:
+            width = height = image_processor_tester.max_resolution
+        else:
+            width, height = np.random.choice(
+                np.arange(image_processor_tester.min_resolution, image_processor_tester.max_resolution), 2
+            )
+            video = prepare_video(
+                image_processor_tester=image_processor_tester,
+                width=width,
+                height=height,
+                numpify=numpify,
+                torchify=torchify,
+            )
+        video_inputs.append(video)
+
+    return video_inputs
+
+
+class TvltImageProcessorTester(unittest.TestCase):
+    def __init__(
+        self,
+        parent,
+        batch_size=7,
+        num_channels=3,
+        num_frames=4,
+        image_size=18,
+        min_resolution=30,
+        max_resolution=400,
+        do_resize=True,
+        size=None,
+        do_normalize=True,
+        image_mean=[0.5, 0.5, 0.5],
+        image_std=[0.5, 0.5, 0.5],
+        do_center_crop=True,
+        crop_size=None,
+    ):
+        size = size if size is not None else {"shortest_edge": 18}
+        crop_size = crop_size if crop_size is not None else {"height": 18, "width": 18}
+
+        self.parent = parent
+        self.batch_size = batch_size
+        self.num_channels = num_channels
+        self.num_frames = num_frames
+        self.image_size = image_size
+        self.min_resolution = min_resolution
+        self.max_resolution = max_resolution
+        self.do_resize = do_resize
+        self.size = size
+        self.do_normalize = do_normalize
+        self.image_mean = image_mean
+        self.image_std = image_std
+        self.do_center_crop = do_center_crop
+        self.crop_size = crop_size
+
+    def prepare_image_processor_dict(self):
+        return {
+            "image_mean": self.image_mean,
+            "image_std": self.image_std,
+            "do_normalize": self.do_normalize,
+            "do_resize": self.do_resize,
+            "size": self.size,
+            "do_center_crop": self.do_center_crop,
+            "crop_size": self.crop_size,
+        }
+
+
+@require_torch
+@require_vision
+class TvltImageProcessorTest(ImageProcessingSavingTestMixin, unittest.TestCase):
+    image_processing_class = TvltImageProcessor if is_vision_available() else None
+
+    def setUp(self):
+        self.image_processor_tester = TvltImageProcessorTester(self)
+
+    @property
+    def image_processor_dict(self):
+        return self.image_processor_tester.prepare_image_processor_dict()
+
+    def test_image_processor_properties(self):
+        image_processor = self.image_processing_class(**self.image_processor_dict)
+        self.assertTrue(hasattr(image_processor, "image_mean"))
+        self.assertTrue(hasattr(image_processor, "image_std"))
+        self.assertTrue(hasattr(image_processor, "do_normalize"))
+        self.assertTrue(hasattr(image_processor, "do_resize"))
+        self.assertTrue(hasattr(image_processor, "do_center_crop"))
+        self.assertTrue(hasattr(image_processor, "size"))
+
+    def test_call_pil(self):
+        # Initialize image_processor
+        image_processor = self.image_processing_class(**self.image_processor_dict)
+        # create random PIL videos
+        video_inputs = prepare_video_inputs(self.image_processor_tester, equal_resolution=False)
+        for video in video_inputs:
+            self.assertIsInstance(video, list)
+            self.assertIsInstance(video[0], Image.Image)
+
+        # Test not batched input
+        encoded_videos = image_processor(video_inputs[0], return_tensors="pt").pixel_values
+        self.assertEqual(
+            encoded_videos.shape,
+            (
+                1,
+                self.image_processor_tester.num_frames,
+                self.image_processor_tester.num_channels,
+                self.image_processor_tester.crop_size["height"],
+                self.image_processor_tester.crop_size["width"],
+            ),
+        )
+
+        # Test batched
+        encoded_videos = image_processor(video_inputs, return_tensors="pt").pixel_values
+        self.assertEqual(
+            encoded_videos.shape,
+            (
+                self.image_processor_tester.batch_size,
+                self.image_processor_tester.num_frames,
+                self.image_processor_tester.num_channels,
+                self.image_processor_tester.crop_size["height"],
+                self.image_processor_tester.crop_size["width"],
+            ),
+        )
+
+    def test_call_numpy(self):
+        # Initialize image_processor
+        image_processor = self.image_processing_class(**self.image_processor_dict)
+        # create random numpy tensors
+        video_inputs = prepare_video_inputs(self.image_processor_tester, equal_resolution=False, numpify=True)
+        for video in video_inputs:
+            self.assertIsInstance(video, list)
+            self.assertIsInstance(video[0], np.ndarray)
+
+        # Test not batched input
+        encoded_videos = image_processor(video_inputs[0], return_tensors="pt").pixel_values
+        self.assertEqual(
+            encoded_videos.shape,
+            (
+                1,
+                self.image_processor_tester.num_frames,
+                self.image_processor_tester.num_channels,
+                self.image_processor_tester.crop_size["height"],
+                self.image_processor_tester.crop_size["width"],
+            ),
+        )
+
+        # Test batched
+        encoded_videos = image_processor(video_inputs, return_tensors="pt").pixel_values
+        self.assertEqual(
+            encoded_videos.shape,
+            (
+                self.image_processor_tester.batch_size,
+                self.image_processor_tester.num_frames,
+                self.image_processor_tester.num_channels,
+                self.image_processor_tester.crop_size["height"],
+                self.image_processor_tester.crop_size["width"],
+            ),
+        )
+
+    def test_call_pytorch(self):
+        # Initialize image_processor
+        image_processor = self.image_processing_class(**self.image_processor_dict)
+        # create random PyTorch tensors
+        video_inputs = prepare_video_inputs(self.image_processor_tester, equal_resolution=False, torchify=True)
+        for video in video_inputs:
+            self.assertIsInstance(video, list)
+            self.assertIsInstance(video[0], torch.Tensor)
+
+        # Test not batched input
+        encoded_videos = image_processor(video_inputs[0], return_tensors="pt").pixel_values
+        self.assertEqual(
+            encoded_videos.shape,
+            (
+                1,
+                self.image_processor_tester.num_frames,
+                self.image_processor_tester.num_channels,
+                self.image_processor_tester.crop_size["height"],
+                self.image_processor_tester.crop_size["width"],
+            ),
+        )
+
+        # Test batched
+        encoded_videos = image_processor(video_inputs, return_tensors="pt").pixel_values
+        self.assertEqual(
+            encoded_videos.shape,
+            (
+                self.image_processor_tester.batch_size,
+                self.image_processor_tester.num_frames,
+                self.image_processor_tester.num_channels,
+                self.image_processor_tester.crop_size["height"],
+                self.image_processor_tester.crop_size["width"],
+            ),
+        )

tests/models/tvlt/test_processor_tvlt.py

+# Copyright 2023 The HuggingFace Team. All rights reserved.
+#
+# 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.
+
+import shutil
+import tempfile
+import unittest
+
+import numpy as np
+import pytest
+
+from transformers import is_speech_available, is_vision_available
+from transformers.testing_utils import require_torch
+
+
+if is_vision_available():
+    from transformers import TvltImageProcessor
+
+if is_speech_available():
+    from transformers import TvltFeatureExtractor
+
+from transformers import TvltProcessor
+
+
+@require_torch
+class TvltProcessorTest(unittest.TestCase):
+    def setUp(self):
+        self.checkpoint = "ZinengTang/tvlt-base"
+        self.tmpdirname = tempfile.mkdtemp()
+
+    def get_image_processor(self, **kwargs):
+        return TvltImageProcessor.from_pretrained(self.checkpoint, **kwargs)
+
+    def get_feature_extractor(self, **kwargs):
+        return TvltFeatureExtractor.from_pretrained(self.checkpoint, **kwargs)
+
+    def tearDown(self):
+        shutil.rmtree(self.tmpdirname)
+
+    def test_save_load_pretrained_default(self):
+        image_processor = self.get_image_processor()
+        feature_extractor = self.get_feature_extractor()
+
+        processor = TvltProcessor(image_processor=image_processor, feature_extractor=feature_extractor)
+        processor.save_pretrained(self.tmpdirname)
+        processor = TvltProcessor.from_pretrained(self.tmpdirname)
+
+        self.assertIsInstance(processor.feature_extractor, TvltFeatureExtractor)
+        self.assertIsInstance(processor.image_processor, TvltImageProcessor)
+
+    def test_feature_extractor(self):
+        image_processor = self.get_image_processor()
+        feature_extractor = self.get_feature_extractor()
+
+        processor = TvltProcessor(image_processor=image_processor, feature_extractor=feature_extractor)
+
+        audio = np.ones([12000])
+
+        audio_dict = feature_extractor(audio, return_tensors="np")
+        input_processor = processor(audio=audio, return_tensors="np")
+
+        for key in audio_dict.keys():
+            self.assertAlmostEqual(audio_dict[key].sum(), input_processor[key].sum(), delta=1e-2)
+
+    def test_image_processor(self):
+        image_processor = self.get_image_processor()
+        feature_extractor = self.get_feature_extractor()
+
+        processor = TvltProcessor(image_processor=image_processor, feature_extractor=feature_extractor)
+
+        images = np.ones([3, 224, 224])
+
+        image_dict = image_processor(images, return_tensors="np")
+        input_processor = processor(images=images, return_tensors="np")
+
+        for key in image_dict.keys():
+            self.assertAlmostEqual(image_dict[key].sum(), input_processor[key].sum(), delta=1e-2)
+
+    def test_processor(self):
+        image_processor = self.get_image_processor()
+        feature_extractor = self.get_feature_extractor()
+
+        processor = TvltProcessor(image_processor=image_processor, feature_extractor=feature_extractor)
+
+        audio = np.ones([12000])
+        images = np.ones([3, 224, 224])
+
+        inputs = processor(audio=audio, images=images)
+
+        self.assertListEqual(list(inputs.keys()), ["audio_values", "audio_mask", "pixel_values", "pixel_mask"])
+
+        # test if it raises when no input is passed
+        with pytest.raises(ValueError):
+            processor()
+
+    def test_model_input_names(self):
+        image_processor = self.get_image_processor()
+        feature_extractor = self.get_feature_extractor()
+
+        processor = TvltProcessor(image_processor=image_processor, feature_extractor=feature_extractor)
+
+        self.assertListEqual(
+            processor.model_input_names,
+            image_processor.model_input_names + feature_extractor.model_input_names,
+            msg="`processor` and `image_processor`+`feature_extractor` model input names do not match",
+        )

utils/check_repo.py

@@ -287,6 +287,7 @@ IGNORE_NON_AUTO_CONFIGURED = PRIVATE_MODELS.copy() + [
     "AltCLIPTextModel",
     "AltCLIPVisionModel",
     "AltRobertaModel",
+    "TvltForAudioVisualClassification",
     "SpeechT5ForSpeechToSpeech",
     "SpeechT5ForTextToSpeech",
     "SpeechT5HifiGan",