Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Caffe2: Open Source Cross-Platform Machine Learning Tools

caffe2.ai

Post reply on HN
Print view View original

11–20 of 57 posts

Re: Caffe2: Open Source Cross-Platform Machine Learning Tools

#11

Post by jiayq84 »

lern_too_spel wrote: post #4
lacker wrote: post #3

This is the library that powers the "style transfer" feature in Facebook Messenger: http://caffe2.ai/docs/AI-Camera-demo-android.html I think Caffe2 is especially suited to machine learning that runs on mobile devices, so I wouldn't be surprised to see it become more popular as that mode of machine learning becomes more popular.

What makes Caffe2 especially suited to mobile devices?

So we carefully made the core much small and also made the platform more modular, so that the dependencies can be minimal when you build on Android/iOS. With our build system (buck) we are having very small binary footprints, which helps delivering the runtime to the phones more easily.

We also did a lot of optimizations on the mobile side - like using NEON, mobile GPU and stuff for optimized speed.

Re: Caffe2: Open Source Cross-Platform Machine Learning Tools

#12

Post by jiayq84 »

danielmorozoff wrote: post #10
drej wrote: post #7

Earlier quoted context omitted.

A super inexperienced observer here, so just a very basic question - how does Caffe(2) differ from TF/Theano/Torch etc.? What are the obvious upsides and potential downsides? A tweet-sized answer will do, I'm just curious as to what the high level differences are. Cheers and congrats on the release!

Would also love some information on performance. How does Caffe2 stack up against TF1.0+? When is MKL support planned for ?

MKL is going to be merged very soon - a full MKL pull request is here if you'd like to track it:

https://github.com/caffe2/caffe2/pull/270

And thanks so much for the awesome work from Intel!

Re: Caffe2: Open Source Cross-Platform Machine Learning Tools

#13

Post by jiayq84 »

danielmorozoff wrote: post #10
drej wrote: post #7

Earlier quoted context omitted.

A super inexperienced observer here, so just a very basic question - how does Caffe(2) differ from TF/Theano/Torch etc.? What are the obvious upsides and potential downsides? A tweet-sized answer will do, I'm just curious as to what the high level differences are. Cheers and congrats on the release!

Would also love some information on performance. How does Caffe2 stack up against TF1.0+? When is MKL support planned for ?

RE the performance - many frameworks nowadays enjoy the high-performance libraries such as CuDNN for optimized runtime on different platforms. For example, we've been collaborating with NVidia on optimized distributed training on GPUs, and you can check out more details here: https://blogs.nvidia.com/blog/2017/04/18/caffe2/

Re: Caffe2: Open Source Cross-Platform Machine Learning Tools

#14

Post by danielmorozoff »

jiayq84 wrote: post #13
danielmorozoff wrote: post #10

Earlier quoted context omitted.

Would also love some information on performance. How does Caffe2 stack up against TF1.0+? When is MKL support planned for ?

RE the performance - many frameworks nowadays enjoy the high-performance libraries such as CuDNN for optimized runtime on different platforms. For example, we've been collaborating with NVidia on optimized distributed training on GPUs, and you can check out more details here: https://blogs.nvidia.com/blog/2017/04/18/caffe2/

Thanks for the link. Are there any public specifics comparing distributed tf to Caffe2's distributed implementation?

Re: Caffe2: Open Source Cross-Platform Machine Learning Tools

#15

Post by doubtfuluser »

This feels really strange almost like Facebook doesn't know wat they want. On the one hand they push torch, pytorch very hard, almost aggressively, now they throw out caffe2 and say this is the holy grail. Wonder if this is similar to the chaos that every app is trying to mimic Snapchat - but here it's like a group of teams want to dethrone TF

Re: Caffe2: Open Source Cross-Platform Machine Learning Tools

#17

Post by IshKebab »

jiayq84 wrote: post #2

Hey guys - Yangqing here, worked on Caffe and Caffe2 over the years, super excited to have continuous contribution to the OSS community. I am more than happy to answer questions if you are curious.

How difficult is it to set up custom recurrent networks? The only system I've ever seen that handles this well is CNTK - you can just say `x = PreviousValue(y)`.

I've tried to work out how to do similar things in Torch and Tensorflow but all they really offer is pre-packaged layers like LSTM. If you want to make your own it's difficult, undocumented and not at all ergonomic.

How does Caffe2 compare?

Re: Caffe2: Open Source Cross-Platform Machine Learning Tools

#18

Post by dzhulgakov »

polskibus wrote: post #16

Does Caffe allow training models on mobile devices or just running already trained models? I'm thinking something along the lines of https://research.googleblog.com/2017/04/federated-learning-c...

Right now we're training models in server-side environment and run highly optimized inference on pre-trained model in products (with focus on inference optimization).

The framework itself allows fine-tuning and training of the model on the mobile device too, but more work is required to enable particular use cases.

Re: Caffe2: Open Source Cross-Platform Machine Learning Tools

#19

Post by dzhulgakov »

IshKebab wrote: post #17
jiayq84 wrote: post #2

Hey guys - Yangqing here, worked on Caffe and Caffe2 over the years, super excited to have continuous contribution to the OSS community. I am more than happy to answer questions if you are curious.

How difficult is it to set up custom recurrent networks? The only system I've ever seen that handles this well is CNTK - you can just say `x = PreviousValue(y)`. I've tried to work out how to do similar things in Torch and Tensorflow but all they really offer is pre-packaged layers like LSTM. If you want to make your own it's difficult, undocumented and not at all ergonomic. How does Caffe2 compare?

The recurrent network work is under way - we are revisiting the designs to consciously balance between performance and API niceness. We'll share more tech details in the upcoming days. Right now, this is an example RNN:

https://github.com/caffe2/caffe2/blob/master/caffe2/python/e...

We use it for machine translation so the perf is nice.

If you have design feedbacks please let us know - we'll be really grateful.

Re: Caffe2: Open Source Cross-Platform Machine Learning Tools

#20

Post by cs702 »

jiayq84 wrote: post #2

Hey guys - Yangqing here, worked on Caffe and Caffe2 over the years, super excited to have continuous contribution to the OSS community. I am more than happy to answer questions if you are curious.

Yangqing, thank you. This is awesome.

Quick question: why did you develop and release this as a completely separate codebase from the original Caffe you created at UC Berkeley?

Post reply on HN