 
            
              Kevin Musgrave
            
            @KevMusgrave
Followers
                1K
              Following
                72
              Media
                85
              Statuses
                617
              Research engineer @fujitsulabs. Co-creator of @settadev. Creator of PyTorch Metric Learning. ML PhD from Cornell.
              
              Joined June 2011
            
            
           Me when GPT codex finally admits it's wrong about how my code works: 
          
                
                0
              
              
                
                0
              
              
                
                0
              
             Why do recent tech announcement videos always start with a person walking into frame and sitting down? 
          
                
                0
              
              
                
                0
              
              
                
                0
              
             Version 2.9.0 of pytorch-metric-learning is out, and includes SmoothAPLoss and other improvements. Thanks to ir2718, lucamarini22, and marcpaga on GitHub! 
          
                
                1
              
              
                
                0
              
              
                
                1
              
             In the UK, you delete old family photos to receive your daily water ration. 
          
                
                0
              
              
                
                0
              
              
                
                0
              
             The lip syncing is amazing 
          
          
                
                0
              
              
                
                0
              
              
                
                1
              
             Curious about image retrieval and contrastive learning? We release the most comprehensive retrieval benchmark, with thousands of experiments on 4 diverse datasets, dozens of losses, investigating batch sizes, LRs, training techniques, data labeling, and much more! A summary ā¬ļø 
          
                
                1
              
              
                
                10
              
              
                
                59
              
             Star the repo on GitHub!  https://t.co/wEjXiJ58Js 
          
          
            
            github.com
              Streamline Python coding, configuration, UI creation, and onboarding. - settadev/setta
            
                
                0
              
              
                
                0
              
              
                
                0
              
             š Iām excited to announce Setta, a dev tool streamlining Python coding and UI creation. No more: - Config parsers - Yaml files - Frontend code @settadev achieves this by treating configuration and your application UI as two sides of the same coin. Tutorial videos coming soon! 
          
                
                1
              
              
                
                4
              
              
                
                9
              
             Pretty lifelike 
           š Can we make a humanoid move like Cristiano Ronaldo, LeBron James and Kobe Byrant? YES! š¤ Introducing ASAP: Aligning Simulation and Real-World Physics for Learning Agile Humanoid Whole-Body Skills Website:  https://t.co/XQga7tIfdw  Code:  https://t.co/NpEeJtVxpp 
            
            
                
                0
              
              
                
                1
              
              
                
                1
              
             I continue to be amazed by Veo 2 
           My feed is lacking silly animal videos. Here's an otter zooming across the water on a jet ski, holding on with both paws, while balancing a fish snack on its head. #veo2
            
            
                
                0
              
              
                
                0
              
              
                
                2
              
             Even billionaires look down at their phones to avoid conversation. 
          
                
                1
              
              
                
                0
              
              
                
                3
              
             Sora looked unbelievably good in February. Now it seems mediocre compared to Veo 2. Did they downgrade Sora's capabilities to enable faster inference, or were the videos in February really that cherry-picked? 
          
                
                0
              
              
                
                1
              
              
                
                3
              
             Gotta say, this one is the most surprising AI result for me in this entire thread (that and the high quality potato anime further up). 
           A flat shaded vector art motion graphics animation of the word 'potato!' in friendly, bright colors. There is another small text 'Powered by Veo!' @oliver_wang2
            
            
                
                0
              
              
                
                1
              
              
                
                0
              
             
               
               
             
               
               
            