Prefix conditioning is a novel method for pre-training visual language models using both classification and caption datasets to provide their complementary benefits. Learn how it uses prefix tokens to disentangle dataset biases from visual concepts ↓
@googleai
-
HALP: Learned Cache Eviction for YouTube CDN Efficiency
By
–
Learn how HALP, a scalable cache eviction framework based on learned rewards that uses preference learning with automated feedback, improves infrastructure efficiency and user video playback latency for YouTube’s content delivery network → https://t.co/RXsDlmo1Ru pic.twitter.com/V5qKhhIv9S
— Google AI (@GoogleAI) 23 juin 2023Learn how HALP, a scalable cache eviction framework based on learned rewards that uses preference learning with automated feedback, improves infrastructure efficiency and user video playback latency for YouTube’s content delivery network → https://
goo.gle/4476JwC -
SoundStorm Synthesizes High-Quality Natural Dialogues
By
–
Also, check out this example of how SoundStorm can synthesize high-quality, natural dialogues: https://t.co/vYX5GZ2fJE pic.twitter.com/AkWSLFTSu7
— Google AI (@GoogleAI) 23 juin 2023Also, check out this example of how SoundStorm can synthesize high-quality, natural dialogues:
-
SoundStorm: Parallel Decoding for Efficient Audio Generation
By
–
Many generative audio models rely on auto-regressive decoding, which produces tokens one by one and can be slow. Read all about SoundStorm, a new method tailored for audio tokens that uses parallel decoding for efficient and high-quality audio generation ↓
-
Google Imagen Editor: Text-Guided Image Editing Tool
By
–
Stop by the #CVPR2023 Google booth at 12:30pm today to speak with @wangsu_googleai about Imagen Editor, a novel tool for text-guided image editing, and to learn about EditBench, a benchmark for quality assessment of image-text alignment.
-
Google Demonstrates On-Device Large Diffusion Models at CVPR
By
–
Curious how we manage to execute large diffusion models (LDM) on-device? Drop by the Google booth at #CVPR2023 today at 10am to see our LDM in action and learn how it was made possible!
-
Ming-Hsuan Yang Wins Longuet-Higgins Prize for Object Tracking
By
–
Congratulations to Ming-Hsuan Yang and co-authors of the paper "Online Object Tracking: A Benchmark" for winning the 2023 Longuet-Higgins Prize at #CVPR2023! This prize recognizes a notable CVPR paper published 10 years ago that has withstood the test of time.
-
40 AI Research Proposals Selected for Funding Globally
By
–
Looking back, in 2022 we received 200+ applications from over 100 universities globally, and as we reopen applications for a new cycle, we are excited to share that 40 proposals were selected for funding. Please help us congratulate our 2022 recipients ↓
-
Award for Inclusion Research Program Opens Globally for Computing
By
–
Applications are open globally for the Award for Inclusion Research program! This supports academic research in computing & technology that addresses the needs of historically marginalized groups for positive social impact. Learn more & apply by July 13 ↓
-
Google Reveals REVEAL: Retrieval-Augmented Visual-Language Model
By
–
At #CVPR2023? Drop by the Google booth at 12:30pm today to hear @alirezafathi talk about REVEAL (https://t.co/MTP69suQ6d), an end-to-end retrieval-augmented visual-language model that learns to use multi-source, multi-modal data to answer knowledge-intensive queries. pic.twitter.com/XqunzICYya
— Google AI (@GoogleAI) 21 juin 2023At #CVPR2023? Drop by the Google booth at 12:30pm today to hear @alirezafathi talk about REVEAL (
http://
goo.gle/3qcZwwc), an end-to-end retrieval-augmented visual-language model that learns to use multi-source, multi-modal data to answer knowledge-intensive queries.