npew

11 ms·

1. 1.▲
   by npew 3y ago

   (I work at OpenAI) We're planning to cover the cost for fine-tuning replacement models. We're still working through the exact mechanics that will work best for customers, and will be reaching out to customers to get feedback on di
2. 2.▲
   by npew 4y ago

   (author here) It depends a bit on how you define "retain". Most GPT-3 applications use custom "prompts" to train it for their specific use case. So in that way, the prompt is retained with every request. You can also fin
3. 3.▲
   by npew 5y ago

   I work at OpenAI. This was unfortunately a bug, and we've fixed it since.
4. 4.▲
   by npew 7y ago

   We use W&B a lot for our research projects, and it's been super useful and has pretty much replaced TensorBoard (TB) for us. The killer features on top of TB is the ability to track your experiments (without keeping track of your l
5. 5.▲
   by npew 9y ago

   The vision network is trained before-hand on lots of different configurations in simulation and then used to infer the block locations in the image from the camera. So it’s not learning continously. The imitation network takes the block loc
6. 6.▲
   by npew 9y ago

   We actually use both. The data used to train the neural networks is all generated through a scripted policy simulation. We then use a single human demonstration in VR to show the robot how to carry out a particular task. VR lets a human do
7. 7.▲
   Using computer vision and crowdsourcing to find taxis in Manhattan(anchovi.posterous.com)

   2 points by npew 15y ago|0 comments