Subscribe
Sign in
51
Replies
This is great @saip @narenst! Was trying to figure out what GPUs you guys are using (seems like K80). Definitely on par with all the cloud providers right now (GCP, AWS, Azure) but I'm frustrated that noone is offering the GTX 1080 or the new Titan X. Both of those cards are quite a bit faster than the K80s for Deep Learning. Would love it if you guys could incorporate one of them somehow :).
@samiur1204 @narenst Agreed! We're currently on AWS, so we're limited to Tesla K80s at the moment, unfortunately. But better GPUs and hosting our own infra is something we have been talking quite a bit recently. I'll keep you posted on that front!
This is pretty awesome... the whole field of Deep Learning is becoming more and more accessible and easy to use.
Here's the launch post on HN.
Great work! If I want to use you, but am completely new to DL, ML, etc., what are the best current resources to learn and get smart in this space?
@ianmikutel There are loads of fantastic machine learning courses available for free.
The first mention usually is the machine learning course by Andrew Ng on Coursera. Its highly recommended simply because Andrew is one of the pioneers of deep learning. be warned theres a lot of math inside, which makes sense because machine learning is 10% programming and 90% math. Udacity also has two incredible courses on machine and deep learning that i can highly recommend. another option is "machine learning a-z" on udemy (although id wait for another of their massive discount periods). closing up with my personal favorite course "practical deep learning for coders" by Jeremy Howard on YouTube.
why is it my favorite? he gets practical very fast and yet still explains every mathematical detail. others may start with very dry theoretical bits. the course was also live recorded with actual students sitting in the room, who often ask just the right questions. however you should know a little about non-deep machine learning before taking that one.
Id recommend starting with "Intro to Machine Learning" on Udacity and then continue with Jeremys course on YouTube. most importantly though: whatever you start with, stick with it. all of these are great. dont be stupid like me and lose time switching between courses to find your personal favorite.
@ianmikutel It's awesome that you're interested in learning!
@gopietz That's a fantastic reply - thanks Pietz!
Theory is obviously fundamental - and Pietz has mentioned some great resources. What works well, for me at least, is trying out some practical projects in parallel. Also helps keep me motivated. There are tons of open source repos on Github. Spin up a few iPython notebooks and play with code/tweak stuff and see how theory translates to results in practice.
Here's a few Tensorflow iPython notebooks: https://github.com/floydhub/tens.... You can run these locally on your laptop. If you want to run them on Floyd, here's some instructions: http://docs.floydhub.com/guides/....
@kleneway Thank you - appreciate the comment! We can't wait for you to try it too :)
So I'm probably a bit sensitive when it comes to Heroku comparisons 😉 But...
Nice DX, what appears at this stage to be pretty compelling pricing, and a marketplace in the works too?! 😱 Wow. This is shaping up to be something very impressive. Great job team.
I'm not really in the target market as I'm fine running infrastructure but a massive n00b when it comes to tensorflow. I really want to up my TF game so I can take this for a serious spin though.
@glenngillen Thanks! Driving down price is something we're serious about and that's a good part of our tech challenge. GPUs on the cloud are ridiculously expensive and a big barrier for people to experiment with DL.
Do take it for a spin and let us know what you think! We definitely want to make it easy to get started, even for beginners and enthusiasts like yourself. We have one full end-to-end tutorial (style transfer), but hoping to add more over the coming days - may be that's where the marketplace will help :)
Hey @saip @narenst -
Ok, this looks supercool! 👏👏 Particularly as I'm setting up DL environments for various projects -- computer vision, NLP, creative AI, and more.
I find creating/establishing proper working environments is roughly 10%-20% of the challenge.
I'd love to see and explore the following installed and working:
-- OpenCV (for computer vision / self-driving car projects)
-- MXNet
-- PyTorch
Datasets I'd love to see available locally:
-- Youtube-8M
-- Quora Q&A dataset
-- MusicNet
-- Million Song Dataset
How do I get started? Thanks!
@atulacharya @narenst Hey, thanks Atul! Glad you're finding it useful. Some details on what's supported and what's coming soon:
- OpenCV is available in all the environments by default. Will add MXNet, PyTorch and Chainer over the weekend.
- Datasets: We have Quora Q&A (http://docs.floydhub.com/guides/...). The rest are also great suggestions. I'll look at their licenses and load them into Floyd soon too!
To get started, check out: https://www.floydhub.com/welcome. We have a simple MNIST project to get you up and running and a more complete Neural Style Transfer guide here: http://docs.floydhub.com/guides/.... Ping me if there's anything at all I can do to help!
@salilnavgire Glad you like it, thanks! Are you still at NYU? Looking to talk to some folks from academia.
@salilnavgire Ah, ok :) I'll ping you offline over the weekend anyway. I know NYU has some really cool DL projects/talent. Would love to get some info or some introductions. Cheers!
Having just been through the pain of setting up DL on AWS, I totally appreciate your proposition. The thing is, you can do DL stuff pretty easily, once you get the infrastructure set up. But getting the infrastructure set up is harder than the DL. Well, not any more :-) Lovely proposition and site. Congrats.
@prattarazzi Thanks! Glad you're finding it useful. We're working on some features the make the DL stuff easier/more productive too, so do keep an eye out :)
Heroku for deep learning with cheaper-than-AWS GPU hardware and Jupyter notebook support? I am intrigued…
@metakermit Thanks! Take it for a spin; let us know if you meet your expectations :) Always happy to hear feedback and/or feature requests!
@saip what are these modules I see in the web app? Are they like models that people have published? I see the examples I ran in there as well – are they public by default?
@metakermit Ah, I see you dug deeper into our hidden features! That's awesome!
First off - don't worry - none of your experiments are public unless you explicitly make it public (this feature is disabled now, so you can't make your stuff public ATM :-)
Some deets:
Our current version lets you run single script jobs only (e.g. floyd run python train.py). We will soon be supporting data workflows that allow you to chain jobs to arbitrarily complex pipelines, like the one in the screenshot above :) More info upcoming!
@verkalets Thanks! Quick question - we want to add more frameworks and DL algorithms to Floyd. Is there anything specific you'd like to see? Sorta doing audience polling :)
@bluemonk482 Sweet! Very interested in getting an AI researcher's perspective. How does this fit with your existing workflow? If there's anything we can do to make it frictionless, let me know - sai@floydhub.com. All ears!
Great effort. Would love to see more compelling demos here that also make deep learning seem simpler.
@vishnugopal Thanks! We will be porting over a few popular DL algorithms and writing in depth guides over the next few days. Anything you'd like to see in particular?
@vishnugopal Working on something for sentiment analysis! Will send out an update soon.
I just gave it a try. I had to wait a couple of minutes in the queue, but otherwise that was the most straightforward experience I've ever had in ML or Data Science. Great work!
@brennenlb Thanks! Sorry about the wait. Lots more jobs than usual running right now and we're spinning up more machines to keep up with the traffic :)
@taha_tweets Thanks! Definitely do want to lower the barrier to entry into deep learning.
Hello Hunters!
I'm Naren, the other co-founder. We are very excited to see all the feedback so far, please keep them coming!
One question for all:
What Deep Learning frameworks / algorithms do you want to see on Floyd?
Maybe something that you already use or always wanted to try but it has been difficult to setup / get started. And we will port them to Floyd :)
@narenst Keras, Chainer and PyTorch are what I use. Since Keras is part of TensorFlow now, I'm sure that's already supported, but Chainer and PyTorch handle dynamic networks much better.
@samiur1204 @narenst Feedback taken. Chainer and PyTorch coming up very shortly!
Hi Hunters,
I’m Sai, one of the cofounders at FloydHub. We're building FloydHub to be a “Heroku for deep learning”. We are in the current batch (W17) at YC. Thanks to @chrismessina for hunting us!
10 months ago, I was working at Microsoft and doing a lot of deep learning (DL) there. While the DL community is terrific, I was often frustrated by how difficult it was to get started and build upon others’ work. For example, running any popular Github project often started with an exercise in dependency hell. As I untangled these for myself, I wrote up some notes on setting up the top DL frameworks, which became super popular on Github (https://github.com/saiprashanths...). That's when I realized that engineering was a huge bottleneck in deep learning and a problem worth solving after all.
I’ve since quit my job and have been working fulltime for the last 9 months on building FloydHub to make deep learning easier. Our goal is to let the data scientists focus on the science, while we handle the engineering grunt work (provisioning and scaling infra, running reproducible experiments, enabling sharing and collaboration, supporting DL frameworks with zero setup, shipping trained models to production easily, etc.) Lots of interesting challenges - happy to talk about them!
We have a lot of work ahead, but we’re excited to share with you what we have so far! We have a neat Neural Style Transfer demo you can try out now (https://www.floydhub.com/#examples). Let us know what you think! Looking forward to your feedback and answering questions.
@gopietz Not yet, but we've been thinking about setting up an academic discount. I'd like to learn more - shoot us an email: founders@floydhub.com
Kitchenbowl