Artificial Intelligence l l AI Updates
Kanalga Telegramβda oβtish
News about AI & DL & ML!!! Admin: @Gayrat_Tangriberganov
Ko'proq ko'rsatish1 614
Obunachilar
Ma'lumot yo'q24 soatlar
+47 kunlar
+530 kunlar
Postlar arxiv
π GitHub_Link
βοΈ Fine-tuning Florence-2 - Microsoft's Cutting-edge Vision Language Models
#VLMs
Join my channel:
ππππππ
https://t.me/Artificial_Intelligence_Updates
π GitHub_Link
βοΈ Data processing with ML and LLM π₯
#LLMs
Join my channel:
ππππππ
https://t.me/Artificial_Intelligence_Updates
π GitHub_Link
βοΈ VIA: A Spatiotemporal Video Adaptation Framework for Global and Local Video Editing π₯
#VideoEditing
Join my channel:
ππππππ
https://t.me/Artificial_Intelligence_Updates
π GitHub_Link
βοΈ Awesome Evaluation of Visual Generation π₯
Join my channel:
ππππππ
https://t.me/Artificial_Intelligence_Updates
π GitHub_Link
π Blog_Link
NVIDIA introduces DoRA (Weight-Decomposed Low-Rank Adaptation), a high-performing alternative to LoRA for fine-tuning.
Key highlights:
- DoRA consistently outperforms LoRA across various tasks, including commonsense reasoning, multi-turn benchmarks, and image-text understanding.
- It improves both learning capacity and stability without introducing additional inference overhead.
- Excels in LLM, VLM, compressed LLM, and #diffusion model applications.
- It's designed to work seamlessly with existing LoRA implementations
DoRA decomposes #pretrained weights into magnitude and directional components, allowing for more nuanced fine-tuning.
This method closely mimics full fine-tuning behavior while maintaining the parameter efficiency of LoRA.
Hugging Face #developers can easily implement DoRA by setting 'use_dora=True' in their LoraConfig, making it readily accessible to improve LLM.
Join my channel:
ππππππ
https://t.me/Artificial_Intelligence_Updates
π GitHub_Link
βοΈ Odd-One-Out: Anomaly Detection by Comparing with Neighbors β‘οΈ
#AnomalyDetection
Join my channel:
ππππππ
https://t.me/Artificial_Intelligence_Updates
π GitHub_Link
RT-DETR is now supported in Hugging Face Transformers! π
RT-DETR, short for βReal-Time DEtection TRansformerβ, is a computer vision model developed at Peking University and Baidu, Inc. capable of real-time object detection. The authors claim better performance than YOLO models in both speed and accuracy. The model comes with an Apache 2.0 license, meaning people can freely use it for commercial applications. π₯
RT-DETR is a follow-up work of DETR, a model developed by AI at Meta that successfully used Transformers for the first time for object detection. The latter has been in the Transformers library since 2020. After this, lots of improvements have been made to enable faster convergence and inference speed. RT-DETR is an important example of that as it unlocks real-time inference at high accuracy!
Big congrats to Daniel Choi for contributing this model!
Join my channel:
ππππππ
https://t.me/Artificial_Intelligence_Updates
π GitHub_Link
βοΈ Doduo: Dense Visual Correspondence from Unsupervised Semantic-Aware Flow
#UnsupervisedLearning
Join my channel:
ππππππ
https://t.me/Artificial_Intelligence_Updates
π GitHub_Link
βοΈ GiT: the first successful LLM-like general vision model unifies various vision tasks only with a vanilla ViT π₯
#LLMs #ViT
Join my channel:
ππππππ
https://t.me/Artificial_Intelligence_Updates
π GitHub_Link
βοΈ Large-Vocabulary Continuous Sign Language Recognition
Join my channel:
ππππππ
https://t.me/Artificial_Intelligence_Updates
π GitHub_Link
βοΈ ManiWAV: Learning Robot Manipulation from In-the-Wild Audio-Visual Data
#Robotics
Join my channel:
ππππππ
https://t.me/Artificial_Intelligence_Updates
π GitHub_Link
βοΈ OPTIMUS: Imitating Task and Motion Planning with Visuomotor Transformers
#Robotics
Join my channel:
ππππππ
https://t.me/Artificial_Intelligence_Updates
π GitHub_Link
βοΈ UniDexGrasp++: Improving Dexterous Grasping Policy Learning via Geometry-aware Curriculum and Iterative Generalist-Specialist Learning
#Robotics
Join my channel:
ππππππ
https://t.me/Artificial_Intelligence_Updates
π GitHub_Link
βοΈ Plan-Seq-Learn: Language Model Guided RL for Solving Long Horizon Robotics Tasks
#Robotics
Join my channel:
ππππππ
https://t.me/Artificial_Intelligence_Updates
π GitHub_Link
βοΈ Dex Retargeting: Various retargeting optimizers to translate human hand motion to robot hand motion
#MotionRetargeting
Join my channel:
ππππππ
https://t.me/Artificial_Intelligence_Updates
π GitHub_Link
βοΈ Generative Region-Language Pretraining for Open-Ended Object Detection
#ObjectDetection
Join my channel:
ππππππ
https://t.me/Artificial_Intelligence_Updates
π GitHub_Link
βοΈ NeRF On-the-go: Exploiting Uncertainty for Distractor-free NeRFs in the Wild
#NeRF
Join my channel:
ππππππ
https://t.me/Artificial_Intelligence_Updates
π GitHub_Link
βοΈ From a Bird's Eye View to See: Joint Camera and Subject Registration without the Camera Calibration
Join my channel:
ππππππ
https://t.me/Artificial_Intelligence_Updates
π GitHub_Link
βοΈ SeaBird: Segmentation in Bird's View with Dice Loss Improves Monocular 3D Detection of Large Objects
#3D #ObjectDetection
Join my channel:
ππππππ
https://t.me/Artificial_Intelligence_Updates
π GitHub_Link
βοΈ AsyncDepth: Better Monocular 3D Detectors with LiDAR from the Past
#LiDAR #Depth #3D
Join my channel:
ππππππ
https://t.me/Artificial_Intelligence_Updates
Endi mavjud! Telegram Tadqiqoti 2025 β yilning asosiy insaytlari 
