Would you recommend this product?
No reviews yet
Hey PH, it's great to be hunted! Deepgram builds hard AI tools used for speech processing. We're open sourcing our extremely useful deep learning tool, Kur, that lets users train their own state-of-the-art deep learning models using TensorFlow and Theano. You use Kur by describing deep neural network models layer by layer in a Kurfile, not by coding. These files can be shared for collaboration using www.KurHub.com (sign up for the beta :) ). Kur works on anything: images, speech, text; if it can be deep learned, we've got it covered. DG also open sourced a brand spanking new speech dataset called the DEEPGRAM10. It's 10 hours of two way conversation between Deepgrammers (listen to some of the DG10 files for juicy insider content) and this is the very dataset used in the speech recognition training example. Kur provides examples for: — speech recognition for DEEPGRAM10 (conversational speech, similar to Baidu's Deepspeech) — image recognition for MNIST (handwritten digits) — image recognition for CIFAR-10 (color images) Check out Kur at: — kur.deepgram.com — github.com/deepgram/kur — kurhub.com We're pumped to be releasing Kur and our team will be really glad to chime in for questions. Thanks from the Deepgram AI Research Team! @DeepgramAI
I have considered taking time this year to learn AI, taking cues from this article https://hackernoon.com/learning-... My question is, how advanced should my skills be before I jump into something like Kur?
@matsilva Hey Mat! If you are familiar with Python 3 and can use a command line terminal, then you've got what you need to use Kur. The install instructions are laid out in http://kur.deepgram.com/install...., but the main point is: If you have Python 3 installed, then just run this in the command line: pip install kur Simple!
@stephensonsco Thanks for that Scott. I have bookmarked this to dive into this weekend & see if I can make any sense of it all ;) Thanks for such a great contribution in this space!
Nice! Looks like Ansible for deep learning :D