Hacker News for AI research PixWizard: Versatile Image-to-Image Visual Assistant with Open-Language Instructions Hot right now
GENERATIVE AI
-
Sapiens: Meta’s New Human-Centric Vision Model Family
By
–
Sapiens, a powerful new model family from Meta Reality Labs, has been introduced for human-centric vision tasks such as 2D pose estimation, body-part segmentation, depth estimation, and surface normal prediction.
— Satya Mallick (@LearnOpenCV) 24 septembre 2024
Human Vision Models like Sapiens are game changers in Metaverse,… pic.twitter.com/4TuOUpgkLISapiens, a powerful new model family from Meta Reality Labs, has been introduced for human-centric vision tasks such as 2D pose estimation, body-part segmentation, depth estimation, and surface normal prediction. Human Vision Models like Sapiens are game changers in Metaverse,
-
Meta Enables Influencers to Create AI Versions of Themselves
By
–
Meta Will Enable Influencers To Create AI Versions of Themselves
#AI #AIio #BigData #ML #NLU #Futureofwork @briansolis @Xbond49 @VitalikButerin @DataScienceCtrl @Benioff @johnnosta @IrmaRaste @Tiffani_Bova http://
ow.ly/4et230sGwQn -
Qwen2.5-34B Block Replication vs Layer Replication Comparison
By
–
I agree. I'm mostly curious about the difference between replicating blocks vs. individual layers with Qwen2.5-34B-Instruct.
-
ByteDance Releases PixelDance and Seaweed Video Generation Models
By
–
ByteDance, the powerhouse behind TikTok, just dropped TWO bomb video gen models – PixelDance & Seaweed! We took 'em for a spin and yes, they're as epic as China's Sora! Get ready to be blown away! 🌟📷 #PixelDance #Seaweed #TechTrend #RunwayGen3 #SoraAI #TikTok pic.twitter.com/iIvITZwO3X
— 机器之心 JIQIZHIXIN (@jiqizhixin) 24 septembre 2024ByteDance, the powerhouse behind TikTok, just dropped TWO bomb video gen models – PixelDance & Seaweed! We took 'em for a spin and yes, they're as epic as China's Sora! Get ready to be blown away! #PixelDance #Seaweed #TechTrend #RunwayGen3 #SoraAI #TikTok
-
Echo Merge: Optimizing Layer Replication in AI Models
By
–
For the echo merge, the idea is that self-merges might be slightly "insane" because replicating blocks of several layers disrupts the information flow. By only replicating individual layers in a distributed way, I'm trying to minimize the input/output difference they introduce.
-
Echo Merge Experiment Tests Self-Merge Information Flow
By
–
I don't expect it to have superior performance in general, but in some narrow domains. The echo merge is an experiment to check if the "insanity" of these self-merges comes from a broken information flow.
-
Top AI Stories: Superintelligence, Autonomous Vehicles, Video Generation
By
–
Top stories in AI today: -Superintelligence may be here sooner than expected
-Nvidia and Alibaba to improve autonomous cars
-AI video-to-video generation with Runway
-IBM, NASA team up on new AI climate model
-6 new AI tools & 4 new AI jobs Read more: http://
therundown.ai/p/superintelli
gence-in-a-few-thousand-days-sam-altman
… -
BigQwen2.5-125B: Self-Merged Large Language Model
By
–
BigQwen2.5-125B I made a few self-merges with the new Qwen2.5 models: – The 125B model is based on Qwen2.5-72B-Instruct and follows Meta-Llama-3-120B-Instruct's (
https://
huggingface.co/mlabonne/Meta-
Llama-3-120B-Instruct
…) recipe, where blocks of 10 layers are repeated 6 times. – The 47B model is based on -
RDUs: Beyond GPUs for Generative and Agentic AI
By
–
Thank you to our partner @AITECHio for sharing the power of our RDUs! As #AI continues to mature, the industry is starting to move beyond traditional #GPUs. RDUs are an excellent alternative to power the most demanding generative & agentic AI workloads.