search

dmlc / minerva

Minerva: a fast and flexible tool for deep learning on multi-GPU. It provides ndarray programming interface, just like Numpy. Python bindings and C++ bindings are both available. The resulting code can be run on CPU or GPU. Multi-GPU support is very easy.
Other
698 stars 172 forks source link

Is it easy(possible) to implement RNN? #8

Closed byzhang closed 9 years ago

byzhang commented 9 years ago

Could you provide a simple example?

hotpxl commented 9 years ago

I think we already have a running example of RNN. @SerailHydra @sneakerkg Could you guys provide a frontend file?

byzhang commented 9 years ago

a kind ping...

hotpxl commented 9 years ago

sorry.. let me ping @SerailHydra again.

We have a simple LSTM/RNN example with fake data, would that suffice?

byzhang commented 9 years ago

Yes, it sounds helpful! On Mar 9, 2015 11:36 PM, "Hot.PxL" notifications@github.com wrote:

sorry.. let me ping @SerailHydra https://github.com/SerailHydra again.

We have a simple LSTM/RNN example with fake data, would that suffice?

— Reply to this email directly or view it on GitHub https://github.com/minerva-developers/minerva/issues/8#issuecomment-78002978 .

SerailHydra commented 9 years ago

Sorry for my late reply. Have been busy for several days. I will send you code later. Is it simple RNN example you want? For now we only have LSTM at hand.

2015-03-10 2:37 GMT-04:00 byzhang notifications@github.com:

Yes, it sounds helpful! On Mar 9, 2015 11:36 PM, "Hot.PxL" notifications@github.com wrote:

sorry.. let me ping @SerailHydra https://github.com/SerailHydra again.

We have a simple LSTM/RNN example with fake data, would that suffice?

— Reply to this email directly or view it on GitHub < https://github.com/minerva-developers/minerva/issues/8#issuecomment-78002978

.

— Reply to this email directly or view it on GitHub https://github.com/minerva-developers/minerva/issues/8#issuecomment-78003052 .

Hongyu Zhu Embedded and Pervasive Computing Center (EPCC Lab) ACM class, Shanghai Jiao Tong University, Shanghai, 200240, China

SerailHydra commented 9 years ago

FYI, here is our current code and data for LSTM architecture. Put them in minerva/owl/apps and do the following: The code is not perfect but it is sufficient to show how to implement one with Minerva.

@minerva/: 1, cd release/owl 2, make 3, cd ../../owl/apps 4, ln -s ../../release/owl/libowl.sp 5, ln -s ../owl/ 6, python lstm-example.py

Thanks

2015-03-10 9:38 GMT-04:00 Hongyu Zhu serailhydra@gmail.com:

Sorry for my late reply. Have been busy for several days. I will send you code later. Is it simple RNN example you want? For now we only have LSTM at hand.

2015-03-10 2:37 GMT-04:00 byzhang notifications@github.com:

Yes, it sounds helpful!

On Mar 9, 2015 11:36 PM, "Hot.PxL" notifications@github.com wrote:

sorry.. let me ping @SerailHydra https://github.com/SerailHydra again.

We have a simple LSTM/RNN example with fake data, would that suffice?

— Reply to this email directly or view it on GitHub < https://github.com/minerva-developers/minerva/issues/8#issuecomment-78002978

.

— Reply to this email directly or view it on GitHub https://github.com/minerva-developers/minerva/issues/8#issuecomment-78003052 .

Hongyu Zhu Embedded and Pervasive Computing Center (EPCC Lab) ACM class, Shanghai Jiao Tong University, Shanghai, 200240, China

Hongyu Zhu Embedded and Pervasive Computing Center (EPCC Lab) ACM class, Shanghai Jiao Tong University, Shanghai, 200240, China

byzhang commented 9 years ago

@SerailHydra , did you try to share the code and data as attachment? They are probably filtered by github. Can you checkin them?

SerailHydra commented 9 years ago

give me your email and I will send them to you directly

2015-03-10 18:23 GMT-04:00 byzhang notifications@github.com:

@SerailHydra https://github.com/SerailHydra , did you try to share the code and data as attachment? They are probably filtered by github. Can you checkin them?

— Reply to this email directly or view it on GitHub https://github.com/minerva-developers/minerva/issues/8#issuecomment-78161709 .

Hongyu Zhu Embedded and Pervasive Computing Center (EPCC Lab) ACM class, Shanghai Jiao Tong University, Shanghai, 200240, China

byzhang commented 9 years ago

Thanks

gujunli commented 9 years ago

Hi :

Will the LSTM code be public? I am interested in taking a look.

This is Junli Gu from AMD research, Beijing lab.

Thanks! junli

On Wed, Mar 11, 2015 at 7:51 AM, SerailHydra notifications@github.com wrote:

give me your email and I will send them to you directly

2015-03-10 18:23 GMT-04:00 byzhang notifications@github.com:

@SerailHydra https://github.com/SerailHydra , did you try to share the code and data as attachment? They are probably filtered by github. Can you checkin them?

— Reply to this email directly or view it on GitHub < https://github.com/minerva-developers/minerva/issues/8#issuecomment-78161709

.

Hongyu Zhu Embedded and Pervasive Computing Center (EPCC Lab) ACM class, Shanghai Jiao Tong University, Shanghai, 200240, China

— Reply to this email directly or view it on GitHub https://github.com/minerva-developers/minerva/issues/8#issuecomment-78173927 .

Junli Gu--谷俊丽 Coordinated Science Lab University of Illinois at Urbana-Champaign

SerailHydra commented 9 years ago

Yes. We have an initiative version now, and we are still working on improving it.

2015-03-11 3:06 GMT-04:00 gujunli notifications@github.com:

Hi :

Will the LSTM code be public? I am interested in taking a look.

This is Junli Gu from AMD research, Beijing lab.

Thanks! junli

On Wed, Mar 11, 2015 at 7:51 AM, SerailHydra notifications@github.com wrote:

give me your email and I will send them to you directly

2015-03-10 18:23 GMT-04:00 byzhang notifications@github.com:

@SerailHydra https://github.com/SerailHydra , did you try to share the code and data as attachment? They are probably filtered by github. Can you checkin them?

— Reply to this email directly or view it on GitHub <

https://github.com/minerva-developers/minerva/issues/8#issuecomment-78161709

.

Hongyu Zhu Embedded and Pervasive Computing Center (EPCC Lab) ACM class, Shanghai Jiao Tong University, Shanghai, 200240, China

— Reply to this email directly or view it on GitHub < https://github.com/minerva-developers/minerva/issues/8#issuecomment-78173927

.

Junli Gu--谷俊丽 Coordinated Science Lab University of Illinois at Urbana-Champaign

— Reply to this email directly or view it on GitHub https://github.com/minerva-developers/minerva/issues/8#issuecomment-78214766 .

Hongyu Zhu Embedded and Pervasive Computing Center (EPCC Lab) ACM class, Shanghai Jiao Tong University, Shanghai, 200240, China