[HN Gopher] Diffusers: Modular Diffusion model library from Hugg...
       ___________________________________________________________________
        
       Diffusers: Modular Diffusion model library from HuggingFace
        
       Author : kashifr
       Score  : 27 points
       Date   : 2022-07-21 16:41 UTC (6 hours ago)
        
 (HTM) web link (github.com)
 (TXT) w3m dump (github.com)
        
       | kashifr wrote:
       | A library of pretrained diffusion models across multiple
       | modalities, such as vision and audio.
        
       | WhatsName wrote:
       | Can someone explain this to me? I've got a basic understanding of
       | ML, but seem out of the loop lately.
        
         | minimaxir wrote:
         | The meta tl;dr is that diffusion is generally used for modern
         | AI-image generation (with a lot of developments in that space),
         | but the code workflows/Colab Notebooks are a mess and filled
         | with technical debt, so a Hugging Face high-level approach for
         | the tool makes sense.
         | 
         | HF recently hired the creators of one of the most popular Colab
         | Notebooks:
         | https://twitter.com/multimodalart/status/1549720764588376064
        
         | swframe2 wrote:
         | Check out this video
         | https://www.youtube.com/watch?v=HoKDTa5jHvg of the math. A lot
         | of other sources skip over the math to give you the big
         | picture. After that you will find the video very helpful in
         | understanding the math.
        
           | uoaei wrote:
           | Holy shit, this is one of the most well-produced, succinct,
           | educational YouTube videos I've ever seen. Instantly
           | subscribed.
        
       ___________________________________________________________________
       (page generated 2022-07-21 23:01 UTC)