Tasks SotA Datasets Papers Methods Submit About

Papers With Code 2

A community resource for machine learning research: papers, code, benchmarks, and state-of-the-art results.

Explore

Notable Benchmarks All SotA Datasets Papers Methods

Community

Submit Results About

Data sourced from the PWC Archive (CC-BY-SA 4.0). Built by the community, for the community.

Models/PaLI

PaLI

Reported on 36 benchmarks across 3 tasks · 2 papers · 17 SOTA

Note: results are matched by exact model name. Different papers may use the same name for different model variants.

Natural Language Processing30 results

Visual Question Answering (VQA)onTextVQA test-standard
overall· 2022-09-14
73.1
SOTA
PaLI: A Jointly-Scaled Multilingual Language-Image Model arXiv:2209.06794
Visual Question Answering (VQA)onVizWiz 2020 VQA
overall· 2022-09-14
73.3
SOTA
PaLI: A Jointly-Scaled Multilingual Language-Image Model arXiv:2209.06794
Visual Question Answering (VQA)onVQA v2 test-dev
Accuracy· 2022-09-14
84.3
SOTA
PaLI: A Jointly-Scaled Multilingual Language-Image Model arXiv:2209.06794
Image Captioningonnocaps near-domain
B3· 2022-09-14
58.99
SOTA
PaLI: A Jointly-Scaled Multilingual Language-Image Model arXiv:2209.06794
Image Captioningonnocaps near-domain
B4· 2022-09-14
39.98
SOTA
PaLI: A Jointly-Scaled Multilingual Language-Image Model arXiv:2209.06794
Image Captioningonnocaps near-domain
METEOR· 2022-09-14
33.47
SOTA
PaLI: A Jointly-Scaled Multilingual Language-Image Model arXiv:2209.06794
Image Captioningonnocaps near-domain
ROUGE-L· 2022-09-14
63.99
SOTA
PaLI: A Jointly-Scaled Multilingual Language-Image Model arXiv:2209.06794
Image Captioningonnocaps out-of-domain
B4· 2022-09-14
32
SOTA
PaLI: A Jointly-Scaled Multilingual Language-Image Model arXiv:2209.06794
Image Captioningonnocaps out-of-domain
CIDEr· 2022-09-14
126.67
SOTA
PaLI: A Jointly-Scaled Multilingual Language-Image Model arXiv:2209.06794
Image Captioningonnocaps out-of-domain
METEOR· 2022-09-14
30.99
SOTA
PaLI: A Jointly-Scaled Multilingual Language-Image Model arXiv:2209.06794
Image Captioningonnocaps out-of-domain
ROUGE-L· 2022-09-14
61.35
SOTA
PaLI: A Jointly-Scaled Multilingual Language-Image Model arXiv:2209.06794
Image Captioningonnocaps in-domain
CIDEr· 2022-09-14
149.1
SOTA
PaLI: A Jointly-Scaled Multilingual Language-Image Model arXiv:2209.06794
Image Captioningonnocaps in-domain
METEOR· 2022-09-14
34.22
SOTA
PaLI: A Jointly-Scaled Multilingual Language-Image Model arXiv:2209.06794
Image Captioningonnocaps in-domain
ROUGE-L· 2022-09-14
64.39
SOTA
PaLI: A Jointly-Scaled Multilingual Language-Image Model arXiv:2209.06794
Visual Question Answering (VQA)onInfoSeek
Accuracy· 2023-02-23
19.7
best: 30.65 (RA-VQAv2 w/ PreFLMR)
Can Pre-trained Vision and Language Models Answer Visual Information-Seeking Questions?arXiv:2302.11713
Image Captioningonnocaps near-domain
B1· 2022-09-14
88.57
best: 88.9 (GIT2, Single Model)
PaLI: A Jointly-Scaled Multilingual Language-Image Model arXiv:2209.06794
Image Captioningonnocaps near-domain
B2· 2022-09-14
75.56
best: 75.86 (GIT2, Single Model)
PaLI: A Jointly-Scaled Multilingual Language-Image Model arXiv:2209.06794
Image Captioningonnocaps near-domain
CIDEr· 2022-09-14
124.35
best: 125.51 (GIT2, Single Model)
PaLI: A Jointly-Scaled Multilingual Language-Image Model arXiv:2209.06794
Image Captioningonnocaps near-domain
SPICE· 2022-09-14
15.75
best: 16.11 (GIT2, Single Model)
PaLI: A Jointly-Scaled Multilingual Language-Image Model arXiv:2209.06794
Image Captioningonnocaps near-domain
SPICE· 2022-09-14
15.75
best: 16.11 (GIT2, Single Model)
PaLI: A Jointly-Scaled Multilingual Language-Image Model arXiv:2209.06794
Image Captioningonnocaps out-of-domain
B1· 2022-09-14
86.28
PaLI: A Jointly-Scaled Multilingual Language-Image Model arXiv:2209.06794
Image Captioningonnocaps out-of-domain
B2· 2022-09-14
71.19
best: 71.28 (GIT, Single Model)
PaLI: A Jointly-Scaled Multilingual Language-Image Model arXiv:2209.06794
Image Captioningonnocaps out-of-domain
B3· 2022-09-14
52.63
best: 52.66 (GIT, Single Model)
PaLI: A Jointly-Scaled Multilingual Language-Image Model arXiv:2209.06794
Image Captioningonnocaps out-of-domain
SPICE· 2022-09-14
15.49
best: 15.7 (GIT, Single Model)
PaLI: A Jointly-Scaled Multilingual Language-Image Model arXiv:2209.06794
Image Captioningonnocaps in-domain
B1· 2022-09-14
88.02
best: 88.86 (GIT2, Single Model)
PaLI: A Jointly-Scaled Multilingual Language-Image Model arXiv:2209.06794
Image Captioningonnocaps in-domain
B2· 2022-09-14
75.21
best: 76.1 (GIT, Single Model)
PaLI: A Jointly-Scaled Multilingual Language-Image Model arXiv:2209.06794
Image Captioningonnocaps in-domain
B3· 2022-09-14
59.38
best: 60.53 (GIT, Single Model)
PaLI: A Jointly-Scaled Multilingual Language-Image Model arXiv:2209.06794
Image Captioningonnocaps in-domain
B4· 2022-09-14
41.16
best: 41.65 (GIT, Single Model)
PaLI: A Jointly-Scaled Multilingual Language-Image Model arXiv:2209.06794
Image Captioningonnocaps in-domain
CIDEr· 2022-09-14
121.09
best: 149.1
PaLI: A Jointly-Scaled Multilingual Language-Image Model arXiv:2209.06794
Image Captioningonnocaps in-domain
SPICE· 2022-09-14
15.69
best: 16.36 (GIT2, Single Model)
PaLI: A Jointly-Scaled Multilingual Language-Image Model arXiv:2209.06794

Computer Vision8 results

Zero-Shot Transfer Image ClassificationonObjectNet
Top 5 Accuracy· 2022-09-14
58.35
SOTA
PaLI: A Jointly-Scaled Multilingual Language-Image Model arXiv:2209.06794
Zero-Shot Transfer Image ClassificationonImageNet-S
Accuracy (Private)· 2022-09-14
63.83
SOTA
PaLI: A Jointly-Scaled Multilingual Language-Image Model arXiv:2209.06794
Zero-Shot Transfer Image ClassificationonImageNet-S
Top 5 Accuracy· 2022-09-14
79.3
SOTA
PaLI: A Jointly-Scaled Multilingual Language-Image Model arXiv:2209.06794
Zero-Shot Transfer Image ClassificationonImageNet V2
Accuracy (Private)· 2022-09-14
64.46
best: 81.2 (BASIC (Lion))
PaLI: A Jointly-Scaled Multilingual Language-Image Model arXiv:2209.06794
Zero-Shot Transfer Image ClassificationonImageNet-A
Accuracy (Private)· 2022-09-14
44.7
best: 90.2 (CoCa)
PaLI: A Jointly-Scaled Multilingual Language-Image Model arXiv:2209.06794
Zero-Shot Transfer Image ClassificationonImageNet
Accuracy (Private)· uses extra data· 2022-09-14
72.11
best: 88.5 (M2-Encoder)
PaLI: A Jointly-Scaled Multilingual Language-Image Model arXiv:2209.06794
Zero-Shot Transfer Image ClassificationonImageNet-R
Accuracy· 2022-09-14
81.97
best: 96.8 (BASIC (Lion))
PaLI: A Jointly-Scaled Multilingual Language-Image Model arXiv:2209.06794
Zero-Shot Transfer Image ClassificationonObjectNet
Accuracy (Private)· 2022-09-14
42.62
best: 87.6 (LiT-22B)
PaLI: A Jointly-Scaled Multilingual Language-Image Model arXiv:2209.06794