satorii

6 ms·

1. 1.▲
   by satorii 5y ago

   Transformer optimization and acceleration
2. 2.▲
   by satorii 5y ago

   Interesting findings! Not sure about whether it is within certain research communities or a broader trend. But Clip cloud be a good plug-in for nowadays writings/design then, something like Clip empowered unsplash.
3. 3.▲
   On the Opportunities and Risks of Foundation Models(arxiv.org)

   20 points by satorii 5y ago|6 comments
4. 4.▲
   Transformer, the model architecture of the year comes to Computer Vision(chengh.medium.com)

   2 points by satorii 6y ago|0 comments
5. 5.▲
   by satorii 6y ago

   Cool! I will drop you a message or catch up with you on openai-api slack. There are a couple of features we see GPT3 stands out and potentially benefits the product features. Currently we are evaluating the end to end quality and cost versu
6. 6.▲
   by satorii 6y ago

   assume each prediction use 1000 tokens (including both input prompt and output text), then 100k tokens -> 100 predictions
7. 7.▲
   by satorii 6y ago

   Thanks for pointing out the errors. I got them fixed in the blog. This is also a good use case of GPT3, checking grammar and spelling. I might build an application for this.
8. 8.▲
   Low-code, no-model recommendation system with GPT3(medium.com)

   95 points by satorii 6y ago|62 comments
9. 9.▲
   by satorii 10y ago

   Amazed by how much work he has done in 100 hours while running one of the world's largest internet company. If anyone interested in trying out some of the "Jarvis" without hacking, checking out https://saymosaic.co