#datacomp search results

(1/n) Introduce negCLIPLoss & NormSim, simple, effective, and efficient #data selection methods for training #CLIP without utilizing external models. Outperform the OAI-CLIP-based-only methods by 5.3% on #DataComp-medium, and achieve new SOTA by combining with the existing best…

ypwang61's tweet image. (1/n) Introduce negCLIPLoss & NormSim, simple, effective, and efficient #data selection methods for training #CLIP without utilizing external models. Outperform the OAI-CLIP-based-only methods by 5.3% on #DataComp-medium, and achieve new SOTA by combining with the existing best…

Relatively under the radar benchmark #DATACOMP focuses on designing multimodal #AI datasets. Propose & evaluate custom training sets on 38 tasks. Initial experiments a 1B subset achieving 79.2% ImageNet accuracy, outperforming CLIP. Fixed data fixes models.

gzlin's tweet image. Relatively under the radar benchmark #DATACOMP focuses on designing multimodal #AI datasets. Propose & evaluate custom training sets on 38 tasks. Initial experiments a 1B subset achieving 79.2% ImageNet accuracy, outperforming CLIP. Fixed data fixes models.

Don’t forget - the 2023 Datacomp/JLT Market Reports Florida for 29 markets in Florida are now available! #Datacomp #MHIndustry #MHProfessional #MHAppraisal #MHMarket #WeHaveTheData

datacompusa's tweet image. Don’t forget - the 2023 Datacomp/JLT Market Reports Florida for 29 markets in Florida are now available!

#Datacomp #MHIndustry #MHProfessional #MHAppraisal #MHMarket #WeHaveTheData

Our comprehensive 2023 Datacomp/JLT Market Reports are now available for major markets in Florida! #Datacomp #MHIndustry #MHProfessional #MHAppraisal #MHMarket #WeHaveTheData

datacompusa's tweet image. Our comprehensive 2023 Datacomp/JLT Market Reports are now available for major markets in Florida!

#Datacomp #MHIndustry #MHProfessional #MHAppraisal #MHMarket #WeHaveTheData

My interest in computer vision started about 8 years ago, and here we are 🙂 🐱 PURRlab from IT University of Copenhagen today at #ICCV2023 ! @DrVeronikaCH @DovileJuodelyte Presenting at #DataComp Workshop how to augment chest X-ray datasets with non-expert annotations 🩻

ameliajimsan's tweet image. My interest in computer vision started about 8 years ago, and here we are 🙂

🐱 PURRlab from IT University of Copenhagen today at #ICCV2023 ! @DrVeronikaCH @DovileJuodelyte 

Presenting at #DataComp Workshop how to augment chest X-ray datasets with non-expert annotations 🩻
ameliajimsan's tweet image. My interest in computer vision started about 8 years ago, and here we are 🙂

🐱 PURRlab from IT University of Copenhagen today at #ICCV2023 ! @DrVeronikaCH @DovileJuodelyte 

Presenting at #DataComp Workshop how to augment chest X-ray datasets with non-expert annotations 🩻

Sleduj prebiehajúce #4kaMSR23 na twitch.tv/yzonesk; zbieraj Twitch pointy a vyber si jednu zo skvelých cien od #Datacomp 😍 #esport #csgo #esporttournament #msrvesportoch2023 #msrves2023

YGamesSlovakia's tweet image. Sleduj prebiehajúce #4kaMSR23 na twitch.tv/yzonesk; zbieraj Twitch pointy a vyber si jednu zo skvelých cien od #Datacomp 😍

#esport #csgo #esporttournament #msrvesportoch2023 #msrves2023

Check out our latest paper in data selection for #CLIP pretraining! by simply switching from CLIPScore filter to negCLIPLoss filter, we have achieved universal improvements. Our additional techniques have also led us to set a new SOTA on #Datacomp-medium.

(1/n) Introduce negCLIPLoss & NormSim, simple, effective, and efficient #data selection methods for training #CLIP without utilizing external models. Outperform the OAI-CLIP-based-only methods by 5.3% on #DataComp-medium, and achieve new SOTA by combining with the existing best…

ypwang61's tweet image. (1/n) Introduce negCLIPLoss & NormSim, simple, effective, and efficient #data selection methods for training #CLIP without utilizing external models. Outperform the OAI-CLIP-based-only methods by 5.3% on #DataComp-medium, and achieve new SOTA by combining with the existing best…


#ComputationalAntitrust - what is already out there? What have we learned? What new applications are possible? Find out in our literature review: papers.ssrn.com/sol3/papers.cf… @graz_wps #DataComp #WorkingPaper

VRobCompLaw's tweet image. #ComputationalAntitrust - what is already out there? What have we learned? What new applications are possible? Find out in our literature review: papers.ssrn.com/sol3/papers.cf… 
@graz_wps #DataComp #WorkingPaper

Beyond just deploying OpenVLA models, I'm excited about all the future directions. Data: Do we need all 970K trajectories? What does filtering and curation look like for robotics? #DataComp for Open X? What tools can we use from the #NLProc & #ComputerVision communities? (5/8)


Why Appraisers Are Terrified of Getting It Wrong. You know that feeling when you’re driving on the interstate, when suddenly, police lights flash and sirens blare behind you? realwired.com/why-appraisers… #Realwired #Youconnect #Datacomp #Glances

RealWired's tweet image. Why Appraisers Are Terrified of Getting It Wrong.
You know that feeling when you’re driving on the interstate, when suddenly, police lights flash and sirens blare behind you?
realwired.com/why-appraisers…
#Realwired #Youconnect #Datacomp #Glances

IFML Director Adam Klivans and Co-director @AlexGDimakis discuss #DataComp, UT Austin's new Center for Generative AI, and the importance of open-source #LLMs on UT's new AI for the Rest of Us podcast. #TexasScience #YearofAI

AI is transforming the way scientists design cancer drugs and vaccines, study how stars form, translate brain activity into speech and turn pollution into useful products. 🎧 Listen to AI for the Rest of Us: txsci.net/AI-discovery #TexasAI #YearofAI #AIForTheRestOfUs

TexasScience's tweet image. AI is transforming the way scientists design cancer drugs and vaccines, study how stars form, translate brain activity into speech and turn pollution into useful products.

🎧 Listen to AI for the Rest of Us: txsci.net/AI-discovery

#TexasAI #YearofAI #AIForTheRestOfUs


AI-enhanced compression algorithms tailored for scientific data reduce bandwidth use for high-resolution imagery, seismic streams, and genomic datasets while preserving fidelity. #DataComp #AI


March 2023 Datacomp/JLT Market Reports are now available for major markets in MD, NH, and NY! ➡️ reports.datacompusa.com/collections/fr… #Datacomp #MHIndustry #MHProfessional #MHAppraisal #MHMarket #WeHaveTheData


AI-enhanced compression algorithms tailored for scientific data reduce bandwidth use for high-resolution imagery, seismic streams, and genomic datasets while preserving fidelity. #DataComp #AI


detector than a non-cynic. They’re wrong. So most people have faith in people who don't have faith in people. In the appraisal industry, a cynical mindset may be more damaging than we think. Is cynicism really helping? realwired.com/cynical-apprai… #Youconnect #Datacomp #Glances


Why Appraisers Are Terrified of Getting It Wrong. You know that feeling when you’re driving on the interstate, when suddenly, police lights flash and sirens blare behind you? realwired.com/why-appraisers… #Realwired #Youconnect #Datacomp #Glances

RealWired's tweet image. Why Appraisers Are Terrified of Getting It Wrong.
You know that feeling when you’re driving on the interstate, when suddenly, police lights flash and sirens blare behind you?
realwired.com/why-appraisers…
#Realwired #Youconnect #Datacomp #Glances

new tech and process. Our world moves at a glacier pace, while the rest of the economy hurtles forward like a McLaren. Many feel appraisers feel we’re just a legislative fiat, a mandatory requirement. realwired.com/were-way-more-… #Realwired #YouConnect #Datacomp


(1/n) Introduce negCLIPLoss & NormSim, simple, effective, and efficient #data selection methods for training #CLIP without utilizing external models. Outperform the OAI-CLIP-based-only methods by 5.3% on #DataComp-medium, and achieve new SOTA by combining with the existing best…

ypwang61's tweet image. (1/n) Introduce negCLIPLoss & NormSim, simple, effective, and efficient #data selection methods for training #CLIP without utilizing external models. Outperform the OAI-CLIP-based-only methods by 5.3% on #DataComp-medium, and achieve new SOTA by combining with the existing best…

The #Datacomp talent show is not disappointing! Makes me wish I had a talent.

SavIsSavvy's tweet image. The #Datacomp talent show is not disappointing! Makes me wish I had a talent.

A massive thanks to @Datacom for inviting us along to their 2022 Datacomp hackathon. We had the opportunity to listen to innovative and diverse solutions to unique challenges facing the community, including online child sexual exploitation. #Datacomp

ThinkUKnow_Aus's tweet image. A massive thanks to @Datacom for inviting us along to their 2022 Datacomp hackathon. We had the opportunity to listen to innovative and diverse solutions to unique challenges facing the community, including online child sexual exploitation. 

#Datacomp

I knew #Datacomp would be a blast, but I didn’t expect to become part of the furniture so quickly!

SavIsSavvy's tweet image. I knew #Datacomp would be a blast, but I didn’t expect to become part of the furniture so quickly!

One of the most exiting time of the year at @Datacom is less than 3 weeks away. Bring on the 7th Datacomp!! #datacomp #datacomp2018

thivy's tweet image. One of the most exiting time of the year at @Datacom is less than 3 weeks away.  Bring on the 7th Datacomp!!  

#datacomp #datacomp2018

Not only did multiple teams hack all weekend at #Datacomp but we had to hack a tripod to kick things off! Best wishes to all the teams pitching this afternoon @Datacom 💪🧠

SavIsSavvy's tweet image. Not only did multiple teams hack all weekend at #Datacomp but we had to hack a tripod to kick things off! Best wishes to all the teams pitching this afternoon @Datacom 💪🧠

Day two of #Datacomp off to a smashing start - incredible to see a company host a fun internal hack that actually celebrates the innovation journey!

SavIsSavvy's tweet image. Day two of #Datacomp off to a smashing start - incredible to see a company host a fun internal hack that actually celebrates the innovation journey!
SavIsSavvy's tweet image. Day two of #Datacomp off to a smashing start - incredible to see a company host a fun internal hack that actually celebrates the innovation journey!
SavIsSavvy's tweet image. Day two of #Datacomp off to a smashing start - incredible to see a company host a fun internal hack that actually celebrates the innovation journey!
SavIsSavvy's tweet image. Day two of #Datacomp off to a smashing start - incredible to see a company host a fun internal hack that actually celebrates the innovation journey!

“You have to dream a little bit to arrive at aspirations” - Greg Davidson, @Datacom CEO #Datacomp #Circular #Hackathon

Datacom's tweet image. “You have to dream a little bit to arrive at aspirations” - Greg Davidson, @Datacom CEO #Datacomp #Circular #Hackathon

Datacomp 2018 is “EmotionalX” With the increase of AI and ambient devices around us, this year we are asking the question of “how can we bridge human emotions and technology?” #Datacomp2018 #Datacomp

thivy's tweet image. Datacomp 2018 is “EmotionalX”

With the increase of AI and ambient devices around us, this year we are asking the question of “how can we bridge human emotions and technology?”

#Datacomp2018 #Datacomp

Relatively under the radar benchmark #DATACOMP focuses on designing multimodal #AI datasets. Propose & evaluate custom training sets on 38 tasks. Initial experiments a 1B subset achieving 79.2% ImageNet accuracy, outperforming CLIP. Fixed data fixes models.

gzlin's tweet image. Relatively under the radar benchmark #DATACOMP focuses on designing multimodal #AI datasets. Propose & evaluate custom training sets on 38 tasks. Initial experiments a 1B subset achieving 79.2% ImageNet accuracy, outperforming CLIP. Fixed data fixes models.

After a late but very fun night, it's the final day of Datacomp and our teams are practicing their pitches ahead of the presentation and judging this afternoon. Good luck! #Datacomp #ThriveTogether #ItsThePeople #Hackathon

Datacom's tweet image. After a late but very fun night, it's the final day of Datacomp and our teams are practicing their pitches ahead of the presentation and judging this afternoon. Good luck! #Datacomp #ThriveTogether #ItsThePeople #Hackathon
Datacom's tweet image. After a late but very fun night, it's the final day of Datacomp and our teams are practicing their pitches ahead of the presentation and judging this afternoon. Good luck! #Datacomp #ThriveTogether #ItsThePeople #Hackathon
Datacom's tweet image. After a late but very fun night, it's the final day of Datacomp and our teams are practicing their pitches ahead of the presentation and judging this afternoon. Good luck! #Datacomp #ThriveTogether #ItsThePeople #Hackathon
Datacom's tweet image. After a late but very fun night, it's the final day of Datacomp and our teams are practicing their pitches ahead of the presentation and judging this afternoon. Good luck! #Datacomp #ThriveTogether #ItsThePeople #Hackathon

My interest in computer vision started about 8 years ago, and here we are 🙂 🐱 PURRlab from IT University of Copenhagen today at #ICCV2023 ! @DrVeronikaCH @DovileJuodelyte Presenting at #DataComp Workshop how to augment chest X-ray datasets with non-expert annotations 🩻

ameliajimsan's tweet image. My interest in computer vision started about 8 years ago, and here we are 🙂

🐱 PURRlab from IT University of Copenhagen today at #ICCV2023 ! @DrVeronikaCH @DovileJuodelyte 

Presenting at #DataComp Workshop how to augment chest X-ray datasets with non-expert annotations 🩻
ameliajimsan's tweet image. My interest in computer vision started about 8 years ago, and here we are 🙂

🐱 PURRlab from IT University of Copenhagen today at #ICCV2023 ! @DrVeronikaCH @DovileJuodelyte 

Presenting at #DataComp Workshop how to augment chest X-ray datasets with non-expert annotations 🩻

Loading...

Something went wrong.


Something went wrong.


United States Trends