SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2305.18500 · NeurIPS · 2023

VAST: A Vision-Audio-Subtitle-Text Omni-Modality Foundation Model and Dataset

Jing Liu, Sihan Chen, Handong Li, Qunbo Wang, Zijia Zhao, Mingzhen Sun, Xinxin Zhu

arXiv · PDF · Open in the Atlas

Code that ran

We lifted 42 functions out of this paper's own repositories and ran 15 of them in a sandbox. "Ran" means the function executed on a synthesized input and returned a value. It is not a reproduction of the paper's results.

RepositoryRoleRan
TXH-mercury/VALOR canonical 9 of 23
TXH-mercury/VAST canonical 1 of 11
txh-mercury/vast — 5 of 8
FunctionStatusWhere it lives
swish Ran TXH-mercury/VALOR/model/bert.py
code served (permissive licence) · get_code("0f786c407fb1ee4c")
Contra_head Ran txh-mercury/vast/model/vast.py
code served (permissive licence) · get_code("d0dc6ea5061f7dc5")
GELU Ran txh-mercury/vast/model/vast.py
code served (permissive licence) · get_code("48d525c2e5cbc84a")
GatherLayer Ran txh-mercury/vast/model/vast.py
code served (permissive licence) · get_code("a839c0c942f1b694")
Match_head Ran txh-mercury/vast/model/vast.py
code served (permissive licence) · get_code("2e0587cc12b477e2")
TokenMasker Ran txh-mercury/vast/model/vast.py
code served (permissive licence) · get_code("3e58c21b8cf01cc4")
basic_clean Ran TXH-mercury/VALOR/model/clip_tokenizer.py
code served (permissive licence) · get_code("98f385d847636a3e")
disabled_train Ran TXH-mercury/VAST/model/general_module.py
code served (permissive licence) · get_code("4cb732f513d69dfd")
drop_path Ran TXH-mercury/VALOR/model/videoswin.py
code served (permissive licence) · get_code("55120f2026b56aa2")
gelu Ran TXH-mercury/VALOR/model/transformer.py
code served (permissive licence) · get_code("fdc64f4c72036ae4")
gelu Ran TXH-mercury/VALOR/model/bert.py
code served (permissive licence) · get_code("40e9fee2e0b7e278")
get_pairs Ran TXH-mercury/VALOR/model/clip_tokenizer.py
code served (permissive licence) · get_code("d919ae32e5e4e616")
trunc_normal_ Ran TXH-mercury/VALOR/model/videoswin.py
code served (permissive licence) · get_code("7810fed39bf5c241")
url_to_filename Ran TXH-mercury/VALOR/model/file_utils.py
code served (permissive licence) · get_code("af64ec220e8bcdbc")
whitespace_clean Ran TXH-mercury/VALOR/model/clip_tokenizer.py
code served (permissive licence) · get_code("9542161e9640b858")
MMGeneralModule Not yet run txh-mercury/vast/model/vast.py
code served (permissive licence) · get_code("b6fa06654838e4da")
VAST Not yet run txh-mercury/vast/model/vast.py
code served (permissive licence) · get_code("c22a88dfed0bd02e")
all_gather_list Not yet run TXH-mercury/VAST/utils/distributed.py
code served (permissive licence) · get_code("eee800bfe0a47a9f")
all_gather_with_grad Not yet run txh-mercury/vast/model/vast.py
code served (permissive licence) · get_code("0ec9fc2025c16f65")
build_optimizer Not yet run TXH-mercury/VAST/utils/build_optimizer.py
code served (permissive licence) · get_code("cc4ffe92873b4ee6")
cached_path Not yet run TXH-mercury/VALOR/model/file_utils.py
code served (permissive licence) · get_code("349d780dd4a89a37")
clean Not yet run TXH-mercury/VALOR/inference.py
code served (permissive licence) · get_code("67dae5beb14eebd7")
clones Not yet run TXH-mercury/VALOR/model/transformer.py
code served (permissive licence) · get_code("98110ed59394ae02")
compute_max_audio_sample_num_for_position_embeddings Not yet run TXH-mercury/VAST/utils/args.py
code served (permissive licence) · get_code("0122d74913461cf9")
compute_max_vision_sample_num_for_position_embeddings Not yet run TXH-mercury/VAST/utils/args.py
code served (permissive licence) · get_code("30fd4a1a3eedb19c")
concat_all_gather Not yet run TXH-mercury/VAST/utils/distributed.py
code served (permissive licence) · get_code("73cecca9f3575f09")
cook_refs Not yet run TXH-mercury/VALOR/scorer/bleu_scorer.py
code served (permissive licence) · get_code("6c465efd98518ed2")
cook_test Not yet run TXH-mercury/VALOR/scorer/bleu_scorer.py
code served (permissive licence) · get_code("fb961e80ca3db085")
execCmd Not yet run TXH-mercury/VAST/utils/offline_process_data.py
code served (permissive licence) · get_code("77a45c1e515dbe2e")
filename_to_url Not yet run TXH-mercury/VALOR/model/file_utils.py
code served (permissive licence) · get_code("db0ac56aaf6e35e6")
get_best_name Not yet run TXH-mercury/VAST/utils/pipeline.py
code served (permissive licence) · get_code("36075dd106d2bbbe")
get_padded_tokens Not yet run TXH-mercury/VALOR/inference.py
code served (permissive licence) · get_code("b65e5286fa81fca8")
load_from_pretrained_dir Not yet run TXH-mercury/VAST/utils/build_model.py
code served (permissive licence) · get_code("7d297bdebd282cba")
load_from_resume Not yet run TXH-mercury/VAST/utils/build_model.py
code served (permissive licence) · get_code("647f671c01bde501")
load_vocab Not yet run TXH-mercury/VALOR/model/bert_tokenizer.py
code served (permissive licence) · get_code("fb95c4b13cdf89a2")
parse_with_config Not yet run TXH-mercury/VAST/utils/args.py
code served (permissive licence) · get_code("15b65a6a6e8fd954")
precook Not yet run TXH-mercury/VALOR/scorer/bleu_scorer.py
code served (permissive licence) · get_code("aad08f61905b40a5")
split Not yet run TXH-mercury/VALOR/inference.py
code served (permissive licence) · get_code("d92ebd9f3d882489")
trans Not yet run TXH-mercury/VALOR/model/modeling.py
code served (permissive licence) · get_code("8b5247b663a8f1b5")
valid Not yet run TXH-mercury/VALOR/model/modeling.py
code served (permissive licence) · get_code("3df7530a6ef7c69d")
whitespace_tokenize Not yet run TXH-mercury/VALOR/model/bert_tokenizer.py
code served (permissive licence) · get_code("da7295883cd7da14")
window_partition Not yet run TXH-mercury/VALOR/model/videoswin.py
code served (permissive licence) · get_code("ed4b0e420c5e9bde")

Repositories linked to this paper

Some links come from the archived Papers with Code dataset (CC BY-SA 4.0): attribution and licence.

Abstract

Vision and text have been fully explored in contemporary video-text foundational models, while other modalities such as audio and subtitles in videos have not received sufficient attention. In this paper, we resort to establish connections between multi-modality video tracks, including Vision, Audio, and Subtitle, and Text by exploring an automatically generated large-scale omni-modality video caption dataset called VAST-27M. Specifically, we first collect 27 million opendomain video clips and separately train a vision and an audio captioner to generate vision and audio captions. Then, we employ an off-the-shelf Large Language Model (LLM) to integrate the generated captions, together with subtitles and instructional prompts into omni-modality captions. Based on the proposed VAST-27M dataset, we train an omni-modality video-text foundational model named VAST, which can perceive and process vision, audio, and subtitle modalities from video, and better support various tasks including vision-text, audio-text, and multi-modal video-text tasks (retrieval, captioning and QA). Extensive experiments have been conducted to demonstrate the effectiveness of our proposed VAST-27M corpus and VAST foundation model. VAST achieves 22 new state-of-the-art results on various cross-modality benchmarks. Code, model and dataset will be released at https://github.com/TXH-mercury/VAST.

For agents

The same record, over MCP at https://syntology.ai/mcp:

get_harvested_code_for_paper("2305.18500")
get_code_for_paper("2305.18500")
have("2305.18500")

Connect an agent — have() is free.