SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2107.06383 · 2021

How Much Can CLIP Benefit Vision-and-Language Tasks?

arXiv · PDF · Open in the Atlas

Code that ran

We lifted 10 functions out of this paper's own repositories and ran 7 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
clip-vil/CLIP-ViL canonical 0 of 1
jianjieluo/openai-clip-feature pwc_unofficial 7 of 9
FunctionStatusWhere it lives
basic_clean Ran jianjieluo/openai-clip-feature/clip/simple_tokenizer.py
pointer only (licence: MIT) · get_code("98f385d847636a3e")
custom_path2id Ran jianjieluo/openai-clip-feature/clip_visual_feats.py
pointer only (licence: MIT) · get_code("012bdd69e64bf2d1")
get_pairs Ran jianjieluo/openai-clip-feature/clip/simple_tokenizer.py
pointer only (licence: MIT) · get_code("d919ae32e5e4e616")
load_json Ran jianjieluo/openai-clip-feature/basic_utils.py
pointer only (licence: MIT) · get_code("2d946250dd2f5a4f")
load_jsonl Ran jianjieluo/openai-clip-feature/basic_utils.py
pointer only (licence: MIT) · get_code("f1d1cccccf038785")
load_pickle Ran jianjieluo/openai-clip-feature/basic_utils.py
pointer only (licence: MIT) · get_code("570ad34bd1af44a8")
whitespace_clean Ran jianjieluo/openai-clip-feature/clip/simple_tokenizer.py
pointer only (licence: MIT) · get_code("9542161e9640b858")
build_model Not yet run clip-vil/CLIP-ViL/CLIP-ViL-Direct/vqa/clip/model.py
code served (permissive licence) · get_code("0a59bad0d461993c")
build_model Not yet run jianjieluo/openai-clip-feature/clip/model.py
pointer only (licence: MIT) · get_code("25a074ccf6e3f98a")
load Not yet run jianjieluo/openai-clip-feature/clip/clip.py
pointer only (licence: MIT) · get_code("2da7ec0975be872a")

Repositories linked to this paper

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

Abstract

Most existing Vision-and-Language (V&L) models rely on pre-trained visual encoders, using a relatively small set of manually-annotated data (as compared to web-crawled data), to perceive the visual world. However, it has been observed that large-scale pretraining usually can result in better generalization performance, e.g., CLIP (Contrastive Language-Image Pre-training), trained on a massive amount of image-caption pairs, has shown a strong zero-shot capability on various vision tasks. To further study the advantage brought by CLIP, we propose to use CLIP as the visual encoder in various V&L models in two typical scenarios: 1) plugging CLIP into task-specific fine-tuning; 2) combining CLIP with V&L pre-training and transferring to downstream tasks. We show that CLIP significantly outperforms widely-used visual encoders trained with in-domain annotated data, such as BottomUp-TopDown. We achieve competitive or better results on diverse V&L tasks, while establishing new state-of-the-art results on Visual Question Answering, Visual Entailment, and V&L Navigation tasks. We release our code at https://github.com/clip-vil/CLIP-ViL.

For agents

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

get_harvested_code_for_paper("2107.06383")
get_code_for_paper("2107.06383")
have("2107.06383")

Connect an agent — have() is free.