TVP model (#25856)

* tvp model for video grounding

add tokenizer auto

fix param in TVPProcessor

add docs

clear comments and enable different torch dtype

add image processor test and model test and fix code style

* fix conflict

* fix model doc

* fix image processing tests

* fix tvp tests

* remove torch in processor

* fix grammar error

* add more details on tvp.md

* fix model arch for loss, grammar, and processor

* add docstring and do not regard TvpTransformer, TvpVisionModel as individual model

* use pad_image

* update copyright

* control first downsample stride

* reduce first only works for ResNetBottleNeckLayer

* fix param name

* fix style

* add testing

* fix style

* rm init_weight

* fix style

* add post init

* fix comments

* do not test TvpTransformer

* fix warning

* fix style

* fix example

* fix config map

* add link in config

* fix comments

* fix style

* rm useless param

* change attention

* change test

* add notes

* fix comments

* fix tvp

* import checkpointing

* fix gradient checkpointing

* Use a more accurate example in readme

* update

* fix copy

* fix style

* update readme

* delete print

* remove tvp test_forward_signature

* remove TvpTransformer

* fix test init model

* merge main and make style

* fix tests and others

* fix image processor

* fix style and model_input_names

* fix tests

README.md

@@ -489,6 +489,7 @@ Current number of checkpoints: ![](https://img.shields.io/endpoint?url=https://h
 1. **[Transformer-XL](https://huggingface.co/docs/transformers/model_doc/transfo-xl)** (from Google/CMU) released with the paper [Transformer-XL: Attentive Language Models Beyond a Fixed-Length Context](https://arxiv.org/abs/1901.02860) by Zihang Dai*, Zhilin Yang*, Yiming Yang, Jaime Carbonell, Quoc V. Le, Ruslan Salakhutdinov.
 1. **[TrOCR](https://huggingface.co/docs/transformers/model_doc/trocr)** (from Microsoft), released together with the paper [TrOCR: Transformer-based Optical Character Recognition with Pre-trained Models](https://arxiv.org/abs/2109.10282) by Minghao Li, Tengchao Lv, Lei Cui, Yijuan Lu, Dinei Florencio, Cha Zhang, Zhoujun Li, Furu Wei.
 1. **[TVLT](https://huggingface.co/docs/transformers/model_doc/tvlt)** (from UNC Chapel Hill) released with the paper [TVLT: Textless Vision-Language Transformer](https://arxiv.org/abs/2209.14156) by Zineng Tang, Jaemin Cho, Yixin Nie, Mohit Bansal.
+1. **[TVP](https://huggingface.co/docs/transformers/model_doc/tvp)** (from Intel) released with the paper [Text-Visual Prompting for Efficient 2D Temporal Video Grounding](https://arxiv.org/abs/2303.04995) by Yimeng Zhang, Xin Chen, Jinghan Jia, Sijia Liu, Ke Ding.
 1. **[UL2](https://huggingface.co/docs/transformers/model_doc/ul2)** (from Google Research) released with the paper [Unifying Language Learning Paradigms](https://arxiv.org/abs/2205.05131v1) by Yi Tay, Mostafa Dehghani, Vinh Q. Tran, Xavier Garcia, Dara Bahri, Tal Schuster, Huaixiu Steven Zheng, Neil Houlsby, Donald Metzler
 1. **[UMT5](https://huggingface.co/docs/transformers/model_doc/umt5)** (from Google Research) released with the paper [UniMax: Fairer and More Effective Language Sampling for Large-Scale Multilingual Pretraining](https://openreview.net/forum?id=kXwdL1cWOAi) by Hyung Won Chung, Xavier Garcia, Adam Roberts, Yi Tay, Orhan Firat, Sharan Narang, Noah Constant.
 1. **[UniSpeech](https://huggingface.co/docs/transformers/model_doc/unispeech)** (from Microsoft Research) released with the paper [UniSpeech: Unified Speech Representation Learning with Labeled and Unlabeled Data](https://arxiv.org/abs/2101.07597) by Chengyi Wang, Yu Wu, Yao Qian, Kenichi Kumatani, Shujie Liu, Furu Wei, Michael Zeng, Xuedong Huang.

README_es.md

@@ -464,6 +464,7 @@ Número actual de puntos de control: ![](https://img.shields.io/endpoint?url=htt
 1. **[Transformer-XL](https://huggingface.co/docs/transformers/model_doc/transfo-xl)** (from Google/CMU) released with the paper [Transformer-XL: Attentive Language Models Beyond a Fixed-Length Context](https://arxiv.org/abs/1901.02860) by Zihang Dai*, Zhilin Yang*, Yiming Yang, Jaime Carbonell, Quoc V. Le, Ruslan Salakhutdinov.
 1. **[TrOCR](https://huggingface.co/docs/transformers/model_doc/trocr)** (from Microsoft), released together with the paper [TrOCR: Transformer-based Optical Character Recognition with Pre-trained Models](https://arxiv.org/abs/2109.10282) by Minghao Li, Tengchao Lv, Lei Cui, Yijuan Lu, Dinei Florencio, Cha Zhang, Zhoujun Li, Furu Wei.
 1. **[TVLT](https://huggingface.co/docs/transformers/model_doc/tvlt)** (from UNC Chapel Hill) released with the paper [TVLT: Textless Vision-Language Transformer](https://arxiv.org/abs/2209.14156) by Zineng Tang, Jaemin Cho, Yixin Nie, Mohit Bansal.
+1. **[TVP](https://huggingface.co/docs/transformers/model_doc/tvp)** (from Intel) released with the paper [Text-Visual Prompting for Efficient 2D Temporal Video Grounding](https://arxiv.org/abs/2303.04995) by Yimeng Zhang, Xin Chen, Jinghan Jia, Sijia Liu, Ke Ding.
 1. **[UL2](https://huggingface.co/docs/transformers/model_doc/ul2)** (from Google Research) released with the paper [Unifying Language Learning Paradigms](https://arxiv.org/abs/2205.05131v1) by Yi Tay, Mostafa Dehghani, Vinh Q. Tran, Xavier Garcia, Dara Bahri, Tal Schuster, Huaixiu Steven Zheng, Neil Houlsby, Donald Metzler
 1. **[UMT5](https://huggingface.co/docs/transformers/model_doc/umt5)** (from Google Research) released with the paper [UniMax: Fairer and More Effective Language Sampling for Large-Scale Multilingual Pretraining](https://openreview.net/forum?id=kXwdL1cWOAi) by Hyung Won Chung, Xavier Garcia, Adam Roberts, Yi Tay, Orhan Firat, Sharan Narang, Noah Constant.
 1. **[UniSpeech](https://huggingface.co/docs/transformers/model_doc/unispeech)** (from Microsoft Research) released with the paper [UniSpeech: Unified Speech Representation Learning with Labeled and Unlabeled Data](https://arxiv.org/abs/2101.07597) by Chengyi Wang, Yu Wu, Yao Qian, Kenichi Kumatani, Shujie Liu, Furu Wei, Michael Zeng, Xuedong Huang.

README_hd.md

@@ -438,6 +438,7 @@ conda install -c huggingface transformers
 1. **[Transformer-XL](https://huggingface.co/docs/transformers/model_doc/transfo-xl)** (Google/CMU की ओर से) कागज के साथ [संस्करण-एक्स: एक ब्लॉग मॉडल चौकस चौक मॉडल मॉडल] (https://arxivorg/abs/1901.02860) क्वोकोक वी. ले, रुस्लैन सलाखुतदी
 1. **[TrOCR](https://huggingface.co/docs/transformers/model_doc/trocr)** (from Microsoft) released with the paper [TrOCR: Transformer-based Optical Character Recognition with Pre-trained Models](https://arxiv.org/abs/2109.10282) by Minghao Li, Tengchao Lv, Lei Cui, Yijuan Lu, Dinei Florencio, Cha Zhang, Zhoujun Li, Furu Wei.
 1. **[TVLT](https://huggingface.co/docs/transformers/model_doc/tvlt)** (from UNC Chapel Hill) released with the paper [TVLT: Textless Vision-Language Transformer](https://arxiv.org/abs/2209.14156) by Zineng Tang, Jaemin Cho, Yixin Nie, Mohit Bansal.
+1. **[TVP](https://huggingface.co/docs/transformers/model_doc/tvp)** (from Intel) released with the paper [Text-Visual Prompting for Efficient 2D Temporal Video Grounding](https://arxiv.org/abs/2303.04995) by Yimeng Zhang, Xin Chen, Jinghan Jia, Sijia Liu, Ke Ding.
 1. **[UL2](https://huggingface.co/docs/transformers/model_doc/ul2)** (from Google Research) released with the paper [Unifying Language Learning Paradigms](https://arxiv.org/abs/2205.05131v1) by Yi Tay, Mostafa Dehghani, Vinh Q. Tran, Xavier Garcia, Dara Bahri, Tal Schuster, Huaixiu Steven Zheng, Neil Houlsby, Donald Metzler
 1. **[UMT5](https://huggingface.co/docs/transformers/model_doc/umt5)** (Google Research से) Hyung Won Chung, Xavier Garcia, Adam Roberts, Yi Tay, Orhan Firat, Sharan Narang, Noah Constant. द्वाराअनुसंधान पत्र [UniMax: Fairer and More Effective Language Sampling for Large-Scale Multilingual Pretraining](https://openreview.net/forum?id=kXwdL1cWOAi) के साथ जारी किया गया
 1. **[UniSpeech](https://huggingface.co/docs/transformers/model_doc/unispeech)** (माइक्रोसॉफ्ट रिसर्च से) साथ में दिया गया पेपर [UniSpeech: यूनिफाइड स्पीच रिप्रेजेंटेशन लर्निंग विद लेबलेड एंड अनलेबल्ड डेटा](https:/ /arxiv.org/abs/2101.07597) चेंगई वांग, यू वू, याओ कियान, केनिची कुमातानी, शुजी लियू, फुरु वेई, माइकल ज़ेंग, ज़ुएदोंग हुआंग द्वारा।

README_ja.md

@@ -498,6 +498,7 @@ Flax、PyTorch、TensorFlowをcondaでインストールする方法は、それ
 1. **[Transformer-XL](https://huggingface.co/docs/transformers/model_doc/transfo-xl)** (Google/CMU から) Zihang Dai*, Zhilin Yang*, Yiming Yang, Jaime Carbonell, Quoc V. Le, Ruslan Salakhutdinov から公開された研究論文: [Transformer-XL: Attentive Language Models Beyond a Fixed-Length Context](https://arxiv.org/abs/1901.02860)
 1. **[TrOCR](https://huggingface.co/docs/transformers/model_doc/trocr)** (Microsoft から), Minghao Li, Tengchao Lv, Lei Cui, Yijuan Lu, Dinei Florencio, Cha Zhang, Zhoujun Li, Furu Wei から公開された研究論文: [TrOCR: Transformer-based Optical Character Recognition with Pre-trained Models](https://arxiv.org/abs/2109.10282)
 1. **[TVLT](https://huggingface.co/docs/transformers/model_doc/tvlt)** (from UNC Chapel Hill から), Zineng Tang, Jaemin Cho, Yixin Nie, Mohit Bansal から公開された研究論文: [TVLT: Textless Vision-Language Transformer](https://arxiv.org/abs/2209.14156)
+1. **[TVP](https://huggingface.co/docs/transformers/model_doc/tvp)** (Intel から), Yimeng Zhang, Xin Chen, Jinghan Jia, Sijia Liu, Ke Ding から公開された研究論文: [Text-Visual Prompting for Efficient 2D Temporal Video Grounding](https://arxiv.org/abs/2303.04995)
 1. **[UL2](https://huggingface.co/docs/transformers/model_doc/ul2)** (Google Research から) Yi Tay, Mostafa Dehghani, Vinh Q から公開された研究論文: [Unifying Language Learning Paradigms](https://arxiv.org/abs/2205.05131v1) Tran, Xavier Garcia, Dara Bahri, Tal Schuster, Huaixiu Steven Zheng, Neil Houlsby, Donald Metzler
 1. **[UMT5](https://huggingface.co/docs/transformers/model_doc/umt5)** (Google Research から) Hyung Won Chung, Xavier Garcia, Adam Roberts, Yi Tay, Orhan Firat, Sharan Narang, Noah Constant. から公開された研究論文 [UniMax: Fairer and More Effective Language Sampling for Large-Scale Multilingual Pretraining](https://openreview.net/forum?id=kXwdL1cWOAi)
 1. **[UniSpeech](https://huggingface.co/docs/transformers/model_doc/unispeech)** (Microsoft Research から) Chengyi Wang, Yu Wu, Yao Qian, Kenichi Kumatani, Shujie Liu, Furu Wei, Michael Zeng, Xuedong Huang から公開された研究論文: [UniSpeech: Unified Speech Representation Learning with Labeled and Unlabeled Data](https://arxiv.org/abs/2101.07597)

README_ko.md

@@ -413,6 +413,7 @@ Flax, PyTorch, TensorFlow 설치 페이지에서 이들을 conda로 설치하는
 1. **[Transformer-XL](https://huggingface.co/docs/transformers/model_doc/transfo-xl)** (Google/CMU 에서) Zihang Dai*, Zhilin Yang*, Yiming Yang, Jaime Carbonell, Quoc V. Le, Ruslan Salakhutdinov 의 [Transformer-XL: Attentive Language Models Beyond a Fixed-Length Context](https://arxiv.org/abs/1901.02860) 논문과 함께 발표했습니다.
 1. **[TrOCR](https://huggingface.co/docs/transformers/model_doc/trocr)** (Microsoft 에서) Minghao Li, Tengchao Lv, Lei Cui, Yijuan Lu, Dinei Florencio, Cha Zhang, Zhoujun Li, Furu Wei 의 [TrOCR: Transformer-based Optical Character Recognition with Pre-trained Models](https://arxiv.org/abs/2109.10282) 논문과 함께 발표했습니다.
 1. **[TVLT](https://huggingface.co/docs/transformers/model_doc/tvlt)** (from UNC Chapel Hill 에서) Zineng Tang, Jaemin Cho, Yixin Nie, Mohit Bansal 의 [TVLT: Textless Vision-Language Transformer](https://arxiv.org/abs/2209.14156) 논문과 함께 발표했습니다.
+1. **[TVP](https://huggingface.co/docs/transformers/model_doc/tvp)** (Intel 에서) Yimeng Zhang, Xin Chen, Jinghan Jia, Sijia Liu, Ke Ding 의 [Text-Visual Prompting for Efficient 2D Temporal Video Grounding](https://arxiv.org/abs/2303.04995) 논문과 함께 발표했습니다.
 1. **[UL2](https://huggingface.co/docs/transformers/model_doc/ul2)** (Google Research 에서) Yi Tay, Mostafa Dehghani, Vinh Q. Tran, Xavier Garcia, Dara Bahri, Tal Schuster, Huaixiu Steven Zheng, Neil Houlsby, Donald Metzle 의 [Unifying Language Learning Paradigms](https://arxiv.org/abs/2205.05131v1) 논문과 함께 발표했습니다.
 1. **[UMT5](https://huggingface.co/docs/transformers/model_doc/umt5)** (Google Research 에서 제공)은 Hyung Won Chung, Xavier Garcia, Adam Roberts, Yi Tay, Orhan Firat, Sharan Narang, Noah Constant.의 [UniMax: Fairer and More Effective Language Sampling for Large-Scale Multilingual Pretraining](https://openreview.net/forum?id=kXwdL1cWOAi)논문과 함께 발표했습니다.
 1. **[UniSpeech](https://huggingface.co/docs/transformers/model_doc/unispeech)** (Microsoft Research 에서) Chengyi Wang, Yu Wu, Yao Qian, Kenichi Kumatani, Shujie Liu, Furu Wei, Michael Zeng, Xuedong Huang 의 [UniSpeech: Unified Speech Representation Learning with Labeled and Unlabeled Data](https://arxiv.org/abs/2101.07597) 논문과 함께 발표했습니다.

README_zh-hans.md

@@ -437,6 +437,7 @@ conda install -c huggingface transformers
 1. **[Transformer-XL](https://huggingface.co/docs/transformers/model_doc/transfo-xl)** (来自 Google/CMU) 伴随论文 [Transformer-XL: Attentive Language Models Beyond a Fixed-Length Context](https://arxiv.org/abs/1901.02860) 由 Zihang Dai*, Zhilin Yang*, Yiming Yang, Jaime Carbonell, Quoc V. Le, Ruslan Salakhutdinov 发布。
 1. **[TrOCR](https://huggingface.co/docs/transformers/model_doc/trocr)** (来自 Microsoft) 伴随论文 [TrOCR: Transformer-based Optical Character Recognition with Pre-trained Models](https://arxiv.org/abs/2109.10282) 由 Minghao Li, Tengchao Lv, Lei Cui, Yijuan Lu, Dinei Florencio, Cha Zhang, Zhoujun Li, Furu Wei 发布。
 1. **[TVLT](https://huggingface.co/docs/transformers/model_doc/tvlt)** (来自 UNC Chapel Hill) 伴随论文 [TVLT: Textless Vision-Language Transformer](https://arxiv.org/abs/2209.14156) 由 Zineng Tang, Jaemin Cho, Yixin Nie, Mohit Bansal 发布。
+1. **[TVP](https://huggingface.co/docs/transformers/model_doc/tvp)** (来自 Intel) 伴随论文 [Text-Visual Prompting for Efficient 2D Temporal Video Grounding](https://arxiv.org/abs/2303.04995) 由 Yimeng Zhang, Xin Chen, Jinghan Jia, Sijia Liu, Ke Ding 发布.
 1. **[UL2](https://huggingface.co/docs/transformers/model_doc/ul2)** (from Google Research) released with the paper [Unifying Language Learning Paradigms](https://arxiv.org/abs/2205.05131v1) by Yi Tay, Mostafa Dehghani, Vinh Q. Tran, Xavier Garcia, Dara Bahri, Tal Schuster, Huaixiu Steven Zheng, Neil Houlsby, Donald Metzler
 1. **[UMT5](https://huggingface.co/docs/transformers/model_doc/umt5)** (来自 Google Research) 伴随论文 [UniMax: Fairer and More Effective Language Sampling for Large-Scale Multilingual Pretraining](https://openreview.net/forum?id=kXwdL1cWOAi) 由 Hyung Won Chung, Xavier Garcia, Adam Roberts, Yi Tay, Orhan Firat, Sharan Narang, Noah Constant 发布。
 1. **[UniSpeech](https://huggingface.co/docs/transformers/model_doc/unispeech)** (来自 Microsoft Research) 伴随论文 [UniSpeech: Unified Speech Representation Learning with Labeled and Unlabeled Data](https://arxiv.org/abs/2101.07597) 由 Chengyi Wang, Yu Wu, Yao Qian, Kenichi Kumatani, Shujie Liu, Furu Wei, Michael Zeng, Xuedong Huang 发布。

README_zh-hant.md

@@ -449,6 +449,7 @@ conda install -c huggingface transformers
 1. **[Transformer-XL](https://huggingface.co/docs/transformers/model_doc/transfo-xl)** (from Google/CMU) released with the paper [Transformer-XL: Attentive Language Models Beyond a Fixed-Length Context](https://arxiv.org/abs/1901.02860) by Zihang Dai*, Zhilin Yang*, Yiming Yang, Jaime Carbonell, Quoc V. Le, Ruslan Salakhutdinov.
 1. **[TrOCR](https://huggingface.co/docs/transformers/model_doc/trocr)** (from Microsoft) released with the paper [TrOCR: Transformer-based Optical Character Recognition with Pre-trained Models](https://arxiv.org/abs/2109.10282) by Minghao Li, Tengchao Lv, Lei Cui, Yijuan Lu, Dinei Florencio, Cha Zhang, Zhoujun Li, Furu Wei.
 1. **[TVLT](https://huggingface.co/docs/transformers/model_doc/tvlt)** (from UNC Chapel Hill) released with the paper [TVLT: Textless Vision-Language Transformer](https://arxiv.org/abs/2209.14156) by Zineng Tang, Jaemin Cho, Yixin Nie, Mohit Bansal.
+1. **[TVP](https://huggingface.co/docs/transformers/model_doc/tvp)** (from Intel) released with the paper [Text-Visual Prompting for Efficient 2D Temporal Video Grounding](https://arxiv.org/abs/2303.04995) by Yimeng Zhang, Xin Chen, Jinghan Jia, Sijia Liu, Ke Ding.
 1. **[UL2](https://huggingface.co/docs/transformers/model_doc/ul2)** (from Google Research) released with the paper [Unifying Language Learning Paradigms](https://arxiv.org/abs/2205.05131v1) by Yi Tay, Mostafa Dehghani, Vinh Q. Tran, Xavier Garcia, Dara Bahri, Tal Schuster, Huaixiu Steven Zheng, Neil Houlsby, Donald Metzler
 1. **[UMT5](https://huggingface.co/docs/transformers/model_doc/umt5)** (from Google Research) released with the paper [UniMax: Fairer and More Effective Language Sampling for Large-Scale Multilingual Pretraining](https://openreview.net/forum?id=kXwdL1cWOAi) by Hyung Won Chung, Xavier Garcia, Adam Roberts, Yi Tay, Orhan Firat, Sharan Narang, Noah Constant.
 1. **[UniSpeech](https://huggingface.co/docs/transformers/model_doc/unispeech)** (from Microsoft Research) released with the paper [UniSpeech: Unified Speech Representation Learning with Labeled and Unlabeled Data](https://arxiv.org/abs/2101.07597) by Chengyi Wang, Yu Wu, Yao Qian, Kenichi Kumatani, Shujie Liu, Furu Wei, Michael Zeng, Xuedong Huang.

docs/source/en/_toctree.yml

@@ -723,6 +723,8 @@
         title: TrOCR
       - local: model_doc/tvlt
         title: TVLT
+      - local: model_doc/tvp
+        title: TVP
       - local: model_doc/vilt
         title: ViLT
       - local: model_doc/vision-encoder-decoder

docs/source/en/index.md

@@ -264,6 +264,7 @@ Flax), PyTorch, and/or TensorFlow.
 |                  [Transformer-XL](model_doc/transfo-xl)                  |       ✅        |         ✅         |      ❌      |
 |                         [TrOCR](model_doc/trocr)                         |       ✅        |         ❌         |      ❌      |
 |                          [TVLT](model_doc/tvlt)                          |       ✅        |         ❌         |      ❌      |
+|                           [TVP](model_doc/tvp)                           |       ✅        |         ❌         |      ❌      |
 |                           [UL2](model_doc/ul2)                           |       ✅        |         ✅         |      ✅      |
 |                          [UMT5](model_doc/umt5)                          |       ✅        |         ❌         |      ❌      |
 |                     [UniSpeech](model_doc/unispeech)                     |       ✅        |         ❌         |      ❌      |

docs/source/en/model_doc/tvp.md

+<!--Copyright 2023 The Intel Team Authors and HuggingFace Inc. 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.
+-->
+
+# TVP
+
+## Overview
+
+The text-visual prompting (TVP) framework was proposed in the paper [Text-Visual Prompting for Efficient 2D Temporal Video Grounding](https://arxiv.org/abs/2303.04995) by Yimeng Zhang, Xin Chen, Jinghan Jia, Sijia Liu, Ke Ding.
+
+The abstract from the paper is the following:
+
+*In this paper, we study the problem of temporal video grounding (TVG), which aims to predict the starting/ending time points of moments described by a text sentence within a long untrimmed video. Benefiting from fine-grained 3D visual features, the TVG techniques have achieved remarkable progress in recent years. However, the high complexity of 3D convolutional neural networks (CNNs) makes extracting dense 3D visual features time-consuming, which calls for intensive memory and computing resources. Towards efficient TVG, we propose a novel text-visual prompting (TVP) framework, which incorporates optimized perturbation patterns (that we call ‘prompts’) into both visual inputs and textual features of a TVG model. In sharp contrast to 3D CNNs, we show that TVP allows us to effectively co-train vision encoder and language encoder in a 2D TVG model and improves the performance of cross-modal feature fusion using only low-complexity sparse 2D visual features. Further, we propose a Temporal-Distance IoU (TDIoU) loss for efficient learning of TVG. Experiments on two benchmark datasets, Charades-STA and ActivityNet Captions datasets, empirically show that the proposed TVP significantly boosts the performance of 2D TVG (e.g., 9.79% improvement on Charades-STA and 30.77% improvement on ActivityNet Captions) and achieves 5× inference acceleration over TVG using 3D visual features.*
+
+This research addresses temporal video grounding (TVG), which is the process of pinpointing the start and end times of specific events in a long video, as described by a text sentence. Text-visual prompting (TVP), is proposed to enhance TVG. TVP involves integrating specially designed patterns, known as 'prompts', into both the visual (image-based) and textual (word-based) input components of a TVG model. These prompts provide additional spatial-temporal context, improving the model's ability to accurately determine event timings in the video. The approach employs 2D visual inputs in place of 3D ones. Although 3D inputs offer more spatial-temporal detail, they are also more time-consuming to process. The use of 2D inputs with the prompting method aims to provide similar levels of context and accuracy more efficiently.
+
+<img src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/transformers/model_doc/tvp_architecture.png"
+alt="drawing" width="600"/> 
+
+<small> TVP architecture. Taken from the <a href="https://arxiv.org/abs/2209.14156">original paper.</a> </small>
+
+This model was contributed by [Jiqing Feng](https://huggingface.co/Jiqing). The original code can be found [here](https://github.com/intel/TVP).
+
+## Usage tips and examples
+
+Prompts are optimized perturbation patterns, which would be added to input video frames or text features. Universal set refers to using the same exact set of prompts for any input, this means that these prompts are added consistently to all video frames and text features, regardless of the input's content.
+
+TVP consists of a visual encoder and cross-modal encoder. A universal set of visual prompts and text prompts to be integrated into sampled video frames and textual features, respectively. Specially, a set of different visual prompts are applied to uniformly-sampled frames of one untrimmed video in order.
+
+The goal of this model is to incorporate trainable prompts into both visual inputs and textual features to temporal video grounding(TVG) problems.
+In principle, one can apply any visual, cross-modal encoder in the proposed architecture.
+
+The [`TvpProcessor`] wraps [`BertTokenizer`] and [`TvpImageProcessor`] into a single instance to both
+encode the text and prepare the images respectively.
+
+The following example shows how to run temporal video grounding using [`TvpProcessor`] and [`TvpForVideoGrounding`].
+```python
+import av
+import cv2
+import numpy as np
+import torch
+from huggingface_hub import hf_hub_download
+from transformers import AutoProcessor, TvpForVideoGrounding
+
+
+def pyav_decode(container, sampling_rate, num_frames, clip_idx, num_clips, target_fps):
+    '''
+    Convert the video from its original fps to the target_fps and decode the video with PyAV decoder.
+    Args:
+        container (container): pyav container.
+        sampling_rate (int): frame sampling rate (interval between two sampled frames).
+        num_frames (int): number of frames to sample.
+        clip_idx (int): if clip_idx is -1, perform random temporal sampling.
+            If clip_idx is larger than -1, uniformly split the video to num_clips
+            clips, and select the clip_idx-th video clip.
+        num_clips (int): overall number of clips to uniformly sample from the given video.
+        target_fps (int): the input video may have different fps, convert it to
+            the target video fps before frame sampling.
+    Returns:
+        frames (tensor): decoded frames from the video. Return None if the no
+            video stream was found.
+        fps (float): the number of frames per second of the video.
+    '''
+    video = container.streams.video[0]
+    fps = float(video.average_rate)
+    clip_size = sampling_rate * num_frames / target_fps * fps
+    delta = max(num_frames - clip_size, 0)
+    start_idx = delta * clip_idx / num_clips
+    end_idx = start_idx + clip_size - 1
+    timebase = video.duration / num_frames
+    video_start_pts = int(start_idx * timebase)
+    video_end_pts = int(end_idx * timebase)
+    seek_offset = max(video_start_pts - 1024, 0)
+    container.seek(seek_offset, any_frame=False, backward=True, stream=video)
+    frames = {}
+    for frame in container.decode(video=0):
+        if frame.pts < video_start_pts:
+            continue
+        frames[frame.pts] = frame
+        if frame.pts > video_end_pts:
+            break
+    frames = [frames[pts] for pts in sorted(frames)]
+    return frames, fps
+
+
+def decode(container, sampling_rate, num_frames, clip_idx, num_clips, target_fps):
+    '''
+    Decode the video and perform temporal sampling.
+    Args:
+        container (container): pyav container.
+        sampling_rate (int): frame sampling rate (interval between two sampled frames).
+        num_frames (int): number of frames to sample.
+        clip_idx (int): if clip_idx is -1, perform random temporal sampling.
+            If clip_idx is larger than -1, uniformly split the video to num_clips
+            clips, and select the clip_idx-th video clip.
+        num_clips (int): overall number of clips to uniformly sample from the given video.
+        target_fps (int): the input video may have different fps, convert it to
+            the target video fps before frame sampling.
+    Returns:
+        frames (tensor): decoded frames from the video.
+    '''
+    assert clip_idx >= -2, "Not a valied clip_idx {}".format(clip_idx)
+    frames, fps = pyav_decode(container, sampling_rate, num_frames, clip_idx, num_clips, target_fps)
+    clip_size = sampling_rate * num_frames / target_fps * fps
+    index = np.linspace(0, clip_size - 1, num_frames)
+    index = np.clip(index, 0, len(frames) - 1).astype(np.int64)
+    frames = np.array([frames[idx].to_rgb().to_ndarray() for idx in index])
+    frames = frames.transpose(0, 3, 1, 2)
+    return frames
+
+
+file = hf_hub_download(repo_id="Intel/tvp_demo", filename="AK2KG.mp4", repo_type="dataset")
+model = TvpForVideoGrounding.from_pretrained("Intel/tvp-base")
+
+decoder_kwargs = dict(
+    container=av.open(file, metadata_errors="ignore"),
+    sampling_rate=1,
+    num_frames=model.config.num_frames,
+    clip_idx=0,
+    num_clips=1,
+    target_fps=3,
+)
+raw_sampled_frms = decode(**decoder_kwargs)
+
+text = "a person is sitting on a bed."
+processor = AutoProcessor.from_pretrained("Intel/tvp-base")
+model_inputs = processor(
+    text=[text], videos=list(raw_sampled_frms), return_tensors="pt", max_text_length=100#, size=size
+)
+
+model_inputs["pixel_values"] = model_inputs["pixel_values"].to(model.dtype)
+output = model(**model_inputs)
+
+def get_video_duration(filename):
+    cap = cv2.VideoCapture(filename)
+    if cap.isOpened():
+        rate = cap.get(5)
+        frame_num = cap.get(7)
+        duration = frame_num/rate
+        return duration
+    return -1
+
+duration = get_video_duration(file)
+start, end = processor.post_process_video_grounding(output.logits, duration)
+
+print(f"The time slot of the video corresponding to the text \"{text}\" is from {start}s to {end}s")
+```
+
+Tips:
+
+- This implementation of TVP uses [`BertTokenizer`] to generate text embeddings and Resnet-50 model to compute visual embeddings.
+- Checkpoints for pre-trained [tvp-base](https://huggingface.co/Intel/tvp-base) is released.
+- Please refer to [Table 2](https://arxiv.org/pdf/2303.04995.pdf) for TVP's performance on Temporal Video Grounding task.
+
+
+## TvpConfig
+
+[[autodoc]] TvpConfig
+
+## TvpImageProcessor
+
+[[autodoc]] TvpImageProcessor
+    - preprocess
+
+## TvpProcessor
+
+[[autodoc]] TvpProcessor
+    - __call__
+
+## TvpModel
+
+[[autodoc]] TvpModel
+    - forward
+
+## TvpForVideoGrounding
+
+[[autodoc]] TvpForVideoGrounding
+    - forward
\ No newline at end of file

docs/source/ms/index.md

@@ -228,6 +228,7 @@ Dokumentasi disusun kepada lima bahagian:
 1. **[Transformer-XL](model_doc/transfo-xl)** (from Google/CMU) released with the paper [Transformer-XL: Attentive Language Models Beyond a Fixed-Length Context](https://arxiv.org/abs/1901.02860) by Zihang Dai*, Zhilin Yang*, Yiming Yang, Jaime Carbonell, Quoc V. Le, Ruslan Salakhutdinov.
 1. **[TrOCR](model_doc/trocr)** (from Microsoft), released together with the paper [TrOCR: Transformer-based Optical Character Recognition with Pre-trained Models](https://arxiv.org/abs/2109.10282) by Minghao Li, Tengchao Lv, Lei Cui, Yijuan Lu, Dinei Florencio, Cha Zhang, Zhoujun Li, Furu Wei.
 1. **[TVLT](model_doc/tvlt)** (from UNC Chapel Hill) released with the paper [TVLT: Textless Vision-Language Transformer](https://arxiv.org/abs/2209.14156) by Zineng Tang, Jaemin Cho, Yixin Nie, Mohit Bansal.
+1. **[TVP](model_doc/tvp)** (from Intel) released with the paper [Text-Visual Prompting for Efficient 2D Temporal Video Grounding](https://arxiv.org/abs/2303.04995) by Yimeng Zhang, Xin Chen, Jinghan Jia, Sijia Liu, Ke Ding.
 1. **[UL2](model_doc/ul2)** (from Google Research) released with the paper [Unifying Language Learning Paradigms](https://arxiv.org/abs/2205.05131v1) by Yi Tay, Mostafa Dehghani, Vinh Q. Tran, Xavier Garcia, Dara Bahri, Tal Schuster, Huaixiu Steven Zheng, Neil Houlsby, Donald Metzler
 1. **[UniSpeech](model_doc/unispeech)** (from Microsoft Research) released with the paper [UniSpeech: Unified Speech Representation Learning with Labeled and Unlabeled Data](https://arxiv.org/abs/2101.07597) by Chengyi Wang, Yu Wu, Yao Qian, Kenichi Kumatani, Shujie Liu, Furu Wei, Michael Zeng, Xuedong Huang.
 1. **[UniSpeechSat](model_doc/unispeech-sat)** (from Microsoft Research) released with the paper [UNISPEECH-SAT: UNIVERSAL SPEECH REPRESENTATION LEARNING WITH SPEAKER AWARE PRE-TRAINING](https://arxiv.org/abs/2110.05752) by Sanyuan Chen, Yu Wu, Chengyi Wang, Zhengyang Chen, Zhuo Chen, Shujie Liu, Jian Wu, Yao Qian, Furu Wei, Jinyu Li, Xiangzhan Yu.
@@ -431,6 +432,7 @@ Flax), PyTorch, dan/atau TensorFlow.
 |        Transformer-XL         |       ✅       |       ❌       |       ✅        |         ✅         |      ❌      |
 |             TrOCR             |       ❌       |       ❌       |       ✅        |         ❌         |      ❌      |
 |             TVLT              |       ❌       |       ❌       |       ✅        |         ❌         |      ❌      |
+|              TVP              |       ❌       |       ❌       |       ✅        |         ❌         |      ❌      |
 |           UniSpeech           |       ❌       |       ❌       |       ✅        |         ❌         |      ❌      |
 |         UniSpeechSat          |       ❌       |       ❌       |       ✅        |         ❌         |      ❌      |
 |            UPerNet            |       ❌       |       ❌       |       ✅        |         ❌         |      ❌      |

src/transformers/__init__.py

@@ -597,6 +597,11 @@ _import_structure = {
         "TvltFeatureExtractor",
         "TvltProcessor",
     ],
+    "models.tvp": [
+        "TVP_PRETRAINED_CONFIG_ARCHIVE_MAP",
+        "TvpConfig",
+        "TvpProcessor",
+    ],
     "models.umt5": ["UMT5Config"],
     "models.unispeech": [
         "UNISPEECH_PRETRAINED_CONFIG_ARCHIVE_MAP",
@@ -1010,6 +1015,7 @@ else:
     _import_structure["models.segformer"].extend(["SegformerFeatureExtractor", "SegformerImageProcessor"])
     _import_structure["models.swin2sr"].append("Swin2SRImageProcessor")
     _import_structure["models.tvlt"].append("TvltImageProcessor")
+    _import_structure["models.tvp"].append("TvpImageProcessor")
     _import_structure["models.videomae"].extend(["VideoMAEFeatureExtractor", "VideoMAEImageProcessor"])
     _import_structure["models.vilt"].extend(["ViltFeatureExtractor", "ViltImageProcessor", "ViltProcessor"])
     _import_structure["models.vit"].extend(["ViTFeatureExtractor", "ViTImageProcessor"])
@@ -2931,6 +2937,14 @@ else:
             "TvltPreTrainedModel",
         ]
     )
+    _import_structure["models.tvp"].extend(
+        [
+            "TVP_PRETRAINED_MODEL_ARCHIVE_LIST",
+            "TvpForVideoGrounding",
+            "TvpModel",
+            "TvpPreTrainedModel",
+        ]
+    )
     _import_structure["models.umt5"].extend(
         [
             "UMT5EncoderModel",
@@ -4795,6 +4809,11 @@ if TYPE_CHECKING:
     )
     from .models.trocr import TROCR_PRETRAINED_CONFIG_ARCHIVE_MAP, TrOCRConfig, TrOCRProcessor
     from .models.tvlt import TVLT_PRETRAINED_CONFIG_ARCHIVE_MAP, TvltConfig, TvltFeatureExtractor, TvltProcessor
+    from .models.tvp import (
+        TVP_PRETRAINED_CONFIG_ARCHIVE_MAP,
+        TvpConfig,
+        TvpProcessor,
+    )
     from .models.umt5 import UMT5Config
     from .models.unispeech import UNISPEECH_PRETRAINED_CONFIG_ARCHIVE_MAP, UniSpeechConfig
     from .models.unispeech_sat import UNISPEECH_SAT_PRETRAINED_CONFIG_ARCHIVE_MAP, UniSpeechSatConfig
@@ -5165,6 +5184,7 @@ if TYPE_CHECKING:
         from .models.segformer import SegformerFeatureExtractor, SegformerImageProcessor
         from .models.swin2sr import Swin2SRImageProcessor
         from .models.tvlt import TvltImageProcessor
+        from .models.tvp import TvpImageProcessor
         from .models.videomae import VideoMAEFeatureExtractor, VideoMAEImageProcessor
         from .models.vilt import ViltFeatureExtractor, ViltImageProcessor, ViltProcessor
         from .models.vit import ViTFeatureExtractor, ViTImageProcessor
@@ -6755,6 +6775,12 @@ if TYPE_CHECKING:
             TvltModel,
             TvltPreTrainedModel,
         )
+        from .models.tvp import (
+            TVP_PRETRAINED_MODEL_ARCHIVE_LIST,
+            TvpForVideoGrounding,
+            TvpModel,
+            TvpPreTrainedModel,
+        )
         from .models.umt5 import (
             UMT5EncoderModel,
             UMT5ForConditionalGeneration,

src/transformers/models/__init__.py

@@ -207,6 +207,7 @@ from . import (
     transfo_xl,
     trocr,
     tvlt,
+    tvp,
     umt5,
     unispeech,
     unispeech_sat,

src/transformers/models/auto/configuration_auto.py

@@ -214,6 +214,7 @@ CONFIG_MAPPING_NAMES = OrderedDict(
         ("transfo-xl", "TransfoXLConfig"),
         ("trocr", "TrOCRConfig"),
         ("tvlt", "TvltConfig"),
+        ("tvp", "TvpConfig"),
         ("umt5", "UMT5Config"),
         ("unispeech", "UniSpeechConfig"),
         ("unispeech-sat", "UniSpeechSatConfig"),
@@ -420,6 +421,7 @@ CONFIG_ARCHIVE_MAP_MAPPING_NAMES = OrderedDict(
         ("timesformer", "TIMESFORMER_PRETRAINED_CONFIG_ARCHIVE_MAP"),
         ("transfo-xl", "TRANSFO_XL_PRETRAINED_CONFIG_ARCHIVE_MAP"),
         ("tvlt", "TVLT_PRETRAINED_CONFIG_ARCHIVE_MAP"),
+        ("tvp", "TVP_PRETRAINED_CONFIG_ARCHIVE_MAP"),
         ("unispeech", "UNISPEECH_PRETRAINED_CONFIG_ARCHIVE_MAP"),
         ("unispeech-sat", "UNISPEECH_SAT_PRETRAINED_CONFIG_ARCHIVE_MAP"),
         ("van", "VAN_PRETRAINED_CONFIG_ARCHIVE_MAP"),
@@ -660,6 +662,7 @@ MODEL_NAMES_MAPPING = OrderedDict(
         ("transfo-xl", "Transformer-XL"),
         ("trocr", "TrOCR"),
         ("tvlt", "TVLT"),
+        ("tvp", "TVP"),
         ("ul2", "UL2"),
         ("umt5", "UMT5"),
         ("unispeech", "UniSpeech"),

src/transformers/models/auto/image_processing_auto.py

@@ -102,6 +102,7 @@ IMAGE_PROCESSOR_MAPPING_NAMES = OrderedDict(
         ("table-transformer", "DetrImageProcessor"),
         ("timesformer", "VideoMAEImageProcessor"),
         ("tvlt", "TvltImageProcessor"),
+        ("tvp", "TvpImageProcessor"),
         ("upernet", "SegformerImageProcessor"),
         ("van", "ConvNextImageProcessor"),
         ("videomae", "VideoMAEImageProcessor"),

src/transformers/models/auto/modeling_auto.py

@@ -200,6 +200,7 @@ MODEL_MAPPING_NAMES = OrderedDict(
         ("trajectory_transformer", "TrajectoryTransformerModel"),
         ("transfo-xl", "TransfoXLModel"),
         ("tvlt", "TvltModel"),
+        ("tvp", "TvpModel"),
         ("umt5", "UMT5Model"),
         ("unispeech", "UniSpeechModel"),
         ("unispeech-sat", "UniSpeechSatModel"),

src/transformers/models/auto/processing_auto.py

@@ -81,6 +81,7 @@ PROCESSOR_MAPPING_NAMES = OrderedDict(
         ("speecht5", "SpeechT5Processor"),
         ("trocr", "TrOCRProcessor"),
         ("tvlt", "TvltProcessor"),
+        ("tvp", "TvpProcessor"),
         ("unispeech", "Wav2Vec2Processor"),
         ("unispeech-sat", "Wav2Vec2Processor"),
         ("vilt", "ViltProcessor"),

src/transformers/models/auto/tokenization_auto.py

@@ -371,6 +371,7 @@ else:
             ("tapas", ("TapasTokenizer", None)),
             ("tapex", ("TapexTokenizer", None)),
             ("transfo-xl", ("TransfoXLTokenizer", None)),
+            ("tvp", ("BertTokenizer", "BertTokenizerFast" if is_tokenizers_available() else None)),
             (
                 "umt5",
                 (

src/transformers/models/tvp/__init__.py

+# coding=utf-8
+# Copyright 2023 The Intel AIA Team Authors, and HuggingFace Inc. 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.
+from typing import TYPE_CHECKING
+
+from ...utils import OptionalDependencyNotAvailable, _LazyModule, is_torch_available, is_vision_available
+
+
+_import_structure = {
+    "configuration_tvp": [
+        "TVP_PRETRAINED_CONFIG_ARCHIVE_MAP",
+        "TvpConfig",
+    ],
+    "processing_tvp": ["TvpProcessor"],
+}
+
+try:
+    if not is_vision_available():
+        raise OptionalDependencyNotAvailable()
+except OptionalDependencyNotAvailable:
+    pass
+else:
+    _import_structure["image_processing_tvp"] = ["TvpImageProcessor"]
+
+try:
+    if not is_torch_available():
+        raise OptionalDependencyNotAvailable()
+except OptionalDependencyNotAvailable:
+    pass
+else:
+    _import_structure["modeling_tvp"] = [
+        "TVP_PRETRAINED_MODEL_ARCHIVE_LIST",
+        "TvpModel",
+        "TvpPreTrainedModel",
+        "TvpForVideoGrounding",
+    ]
+
+if TYPE_CHECKING:
+    from .configuration_tvp import (
+        TVP_PRETRAINED_CONFIG_ARCHIVE_MAP,
+        TvpConfig,
+    )
+    from .processing_tvp import TvpProcessor
+
+    try:
+        if not is_vision_available():
+            raise OptionalDependencyNotAvailable()
+    except OptionalDependencyNotAvailable:
+        pass
+    else:
+        from .image_processing_tvp import TvpImageProcessor
+
+    try:
+        if not is_torch_available():
+            raise OptionalDependencyNotAvailable()
+    except OptionalDependencyNotAvailable:
+        pass
+    else:
+        from .modeling_tvp import (
+            TVP_PRETRAINED_MODEL_ARCHIVE_LIST,
+            TvpForVideoGrounding,
+            TvpModel,
+            TvpPreTrainedModel,
+        )
+
+else:
+    import sys
+
+    sys.modules[__name__] = _LazyModule(__name__, globals()["__file__"], _import_structure, module_spec=__spec__)

src/transformers/models/tvp/configuration_tvp.py

+# coding=utf-8
+# Copyright 2023 The Intel AIA Team Authors, and HuggingFace Inc. 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.
+""" TVP model configuration"""
+
+import copy
+
+from ...configuration_utils import PretrainedConfig
+from ...utils import logging
+from ..auto import CONFIG_MAPPING
+
+
+logger = logging.get_logger(__name__)
+
+
+TVP_PRETRAINED_CONFIG_ARCHIVE_MAP = {
+    "Intel/tvp-base": "https://huggingface.co/Intel/tvp-base/resolve/main/config.json",
+}
+
+
+class TvpConfig(PretrainedConfig):
+    r"""
+    This is the configuration class to store the configuration of a [`TvpModel`]. It is used to instantiate an Tvp
+    model according to the specified arguments, defining the model architecture. Instantiating a configuration with the
+    defaults will yield a similar configuration to that of the Tvp
+    [Intel/tvp-base](https://huggingface.co/Intel/tvp-base) architecture.
+
+    Configuration objects inherit from [`PretrainedConfig`] and can be used to control the model outputs. Read the
+    documentation from [`PretrainedConfig`] for more information.
+
+
+    Args:
+        backbone_config (`PretrainedConfig` or `dict`, *optional*):
+            The configuration of the backbone model.
+        distance_loss_weight (`float`, *optional*, defaults to 1.0):
+            The weight of distance loss.
+        duration_loss_weight (`float`, *optional*, defaults to 0.1):
+            The weight of duration loss.
+        visual_prompter_type (`str`, *optional*, defaults to `"framepad"`):
+            Visual prompt type. The type of padding. Framepad means padding on each frame. Should be one of "framepad"
+            or "framedownpad"
+        visual_prompter_apply (`str`, *optional*, defaults to `"replace"`):
+            The way of applying visual prompt. Replace means use the value of prompt to change the original value in
+            visual inputs. Should be one of "replace", or "add", or "remove".
+        visual_prompt_size (`int`, *optional*, defaults to 96):
+            The size of visual prompt.
+        max_img_size (`int`, *optional*, defaults to 448):
+            The maximum size of frame.
+        num_frames (`int`, *optional*, defaults to 48):
+            The number of frames extracted from a video.
+        vocab_size (`int`, *optional*, defaults to 30522):
+            Vocabulary size of the Tvp text model. Defines the number of different tokens that can be represented by
+            the `inputs_ids` passed when calling [`TvpModel`].
+        hidden_size (`int`, *optional*, defaults to 768):
+            Dimensionality of the encoder layers.
+        intermediate_size (`int`, *optional*, defaults to 3072):
+            Dimensionality of the "intermediate" (i.e., feed-forward) layer in the Transformer encoder.
+        num_hidden_layers (`int`, *optional*, defaults to 12):
+            Number of hidden layers in the Transformer encoder.
+        num_attention_heads (`int`, *optional*, defaults to 12):
+            Number of attention heads for each attention layer in the Transformer encoder.
+        max_position_embeddings (`int`, *optional*, defaults to 512):
+            The maximum sequence length that this model might ever be used with. Typically set this to something large
+            just in case (e.g., 512 or 1024 or 2048).
+        max_grid_col_position_embeddings (`int`, *optional*, defaults to 100):
+            The largest number of horizontal patches from a video frame.
+        max_grid_row_position_embeddings (`int`, *optional*, defaults to 100):
+            The largest number of vertical patches from a video frame.
+        hidden_dropout_prob (`float`, *optional*, defaults to 0.1):
+            The dropout probability of hidden layers.
+        hidden_act (`str` or `function`, *optional*, defaults to `"gelu"`):
+            The non-linear activation function (function or string) in the encoder and pooler. If string, `"gelu"`,
+            `"relu"`, `"selu"` and `"gelu_new"` ``"quick_gelu"` are supported.
+        layer_norm_eps (`float`, *optional*, defaults to 1e-12):
+            The epsilon used by the layer normalization layers.
+        initializer_range (`float`, *optional*, defaults to 0.02):
+            The standard deviation of the truncated_normal_initializer for initializing all weight matrices.
+        attention_probs_dropout_prob (`float`, *optional*, defaults to 0.1):
+            The dropout probability of attention layers.
+    """
+
+    model_type = "tvp"
+
+    def __init__(
+        self,
+        backbone_config=None,
+        distance_loss_weight=1.0,
+        duration_loss_weight=0.1,
+        visual_prompter_type="framepad",
+        visual_prompter_apply="replace",
+        visual_prompt_size=96,
+        max_img_size=448,
+        num_frames=48,
+        vocab_size=30522,
+        hidden_size=768,
+        intermediate_size=3072,
+        num_hidden_layers=12,
+        num_attention_heads=12,
+        max_position_embeddings=512,
+        max_grid_col_position_embeddings=100,
+        max_grid_row_position_embeddings=100,
+        hidden_dropout_prob=0.1,
+        hidden_act="gelu",
+        layer_norm_eps=1e-12,
+        initializer_range=0.02,
+        attention_probs_dropout_prob=0.1,
+        **kwargs,
+    ):
+        super().__init__(**kwargs)
+
+        if backbone_config is None:
+            logger.info("`backbone_config` is `None`. Initializing the config with the default `ResNet` backbone.")
+            backbone_config = CONFIG_MAPPING["resnet"](out_features=["stage4"])
+        elif isinstance(backbone_config, dict):
+            backbone_model_type = backbone_config.get("model_type")
+            config_class = CONFIG_MAPPING[backbone_model_type]
+            backbone_config = config_class.from_dict(backbone_config)
+
+        self.backbone_config = backbone_config
+        self.distance_loss_weight = distance_loss_weight
+        self.duration_loss_weight = duration_loss_weight
+        self.visual_prompter_type = visual_prompter_type
+        self.visual_prompter_apply = visual_prompter_apply
+        self.visual_prompt_size = visual_prompt_size
+        self.max_img_size = max_img_size
+        self.num_frames = num_frames
+        self.vocab_size = vocab_size
+        self.hidden_size = hidden_size
+        self.intermediate_size = intermediate_size
+        self.num_hidden_layers = num_hidden_layers
+        self.num_attention_heads = num_attention_heads
+        self.max_position_embeddings = max_position_embeddings
+        self.max_grid_col_position_embeddings = max_grid_col_position_embeddings
+        self.max_grid_row_position_embeddings = max_grid_row_position_embeddings
+        self.layer_norm_eps = layer_norm_eps
+        self.hidden_dropout_prob = hidden_dropout_prob
+        self.hidden_act = hidden_act
+        self.initializer_range = initializer_range
+        self.attention_probs_dropout_prob = attention_probs_dropout_prob
+
+    @classmethod
+    def from_backbone_config(cls, backbone_config: PretrainedConfig, **kwargs):
+        """Instantiate a [`TvpConfig`] (or a derived class) from a pre-trained backbone model configuration.
+
+        Args:
+            backbone_config ([`PretrainedConfig`]):
+                The backbone configuration.
+        Returns:
+            [`TvpConfig`]: An instance of a configuration object
+        """
+        return cls(backbone_config=backbone_config, **kwargs)
+
+    def to_dict(self):
+        """
+        Serializes this instance to a Python dictionary. Override the default [`~PretrainedConfig.to_dict`].
+
+        Returns:
+            `Dict[str, any]`: Dictionary of all the attributes that make up this configuration instance,
+        """
+        output = copy.deepcopy(self.__dict__)
+        if output["backbone_config"] is not None:
+            output["backbone_config"] = self.backbone_config.to_dict()
+        output["model_type"] = self.__class__.model_type
+        return output