Vocabulary-free Image Classification paper page: https://
huggingface.co/papers/2306.00
917
…
demo: https://
altndrr-vic.hf.space Recent advances in large vision-language models have revolutionized the image classification paradigm. Despite showing impressive zero-shot capabilities, a pre-defined set of
MACHINE LEARNING
-
Vocabulary-free Image Classification Using Vision-Language Models
By
–
-
VideoComposer: Controllable Video Synthesis with Temporal Conditions
By
–
VideoComposer: Compositional Video Synthesis
— AK (@_akhaliq) 5 juin 2023
with Motion Controllability
project page: https://t.co/n5DgXaCRgH
presents VideoCompoer that allows users to flexibly compose a video with textual conditions, spatial conditions, and more importantly temporal conditions.… pic.twitter.com/IEDi9wHo5pComposer: Compositional Synthesis
with Motion Controllability project page: https://
videocomposer.github.io presents VideoCompoer that allows users to flexibly compose a video with textual conditions, spatial conditions, and more importantly temporal conditions. -
StableRep: Synthetic Images Improve Visual Representation Learning
By
–
StableRep: Synthetic Images from Text-to-Image Models Make Strong Visual Representation Learners Another paper demonstrating the power of training ML models using *only* synthetically generated content leading to good improvements in learning efficiency! https://
arxiv.org/abs/2306.00984 -
Computer Vision Evolution: From Legacy to AI-Driven Production Solutions
By
–
Know about the journey of computer vision in production from vintage techniques to modern AI-driven solutions! Learn about its evolution, breakthroughs, and future possibilities. Read more: https://
bit.ly/3MTqBMM @_DigitalIndia @GoI_MeitY @NeGD_GoI @nasscom -
AAAI Bridge Program on Continual Causality Proceedings Published
By
–
Volume 208 Proceedings of the AAAI Bridge Program on Continual Causality https://
proceedings.mlr.press/v208/ Is now available on PMLR. -
Paraconsistent Logic Advances AI Reasoning Capabilities
By
–
Advancing AI's reasoning capabilities with paraconsistent logic! The innovative approach overcomes logical inconsistencies, enabling AI systems to make informed decisions even in complex and uncertain scenarios. Read more: https://
bit.ly/43osNTx @jeevprabnivash -
DaTaSeg: Universal Multi-Dataset Multi-Task Segmentation Model
By
–
DaTaSeg: Taming a Universal Multi-Dataset Multi-Task Segmentation Model paper page: https://
huggingface.co/papers/2306.01
736
… Observing the close relationship among panoptic, semantic and instance segmentation tasks, we propose to train a universal multi-dataset multi-task segmentation model: -
HQ-SAM: High Quality Segment Anything Model
By
–
Segment Anything in High Quality
— AK (@_akhaliq) 5 juin 2023
paper page: https://t.co/IG4IAN4a4V
propose HQ-SAM, equipping SAM with the ability to accurately segment any object, while maintaining SAM's original promptable design, efficiency, and zero-shot generalizability. Our careful design reuses and… pic.twitter.com/6xG0FES7hvSegment Anything in High Quality paper page: https://
huggingface.co/papers/2306.01
567
… propose HQ-SAM, equipping SAM with the ability to accurately segment any object, while maintaining SAM's original promptable design, efficiency, and zero-shot generalizability. Our careful design reuses and -
Fine-Grained Human Feedback Improves Language Model Reward Training
By
–
Fine-Grained Human Feedback Gives Better Rewards for Language Model Training paper page: https://
huggingface.co/papers/2306.01
693
… use fine-grained human feedback (e.g., which sentence is false, which sub-sentence is irrelevant) as an explicit training signal. We introduce Fine-Grained RLHF, a -
Large Language Models for Private Synthetic Text Generation
By
–
Harnessing large-language models to generate private synthetic text paper page: https://
huggingface.co/papers/2306.01
684
… Differentially private (DP) training methods like DP-SGD can protect sensitive training data by ensuring that ML models will not reveal private information. An alternative