Executable tutorial #38

paraynaud · 2022-07-06T20:50:42Z

environnement @example
add required dependencies

dpo · 2022-07-07T13:58:57Z

docs/src/tutorial.md

+  w :: Param{Matrix{Float32}} # parameters of the layers
+  b :: Param{Vector{Float32}} # bias of the layer
+  # w :: Param{CuArray{Float32, 2, CUDA.Mem.DeviceBuffer}} # for GPU
+  # b :: Param{CuArray{Float32, 1, CUDA.Mem.DeviceBuffer}} # for GPU


How about parameterizing so that the user can use the struct on both CPU and GPU?

We know that

Dense(i :: Int, o :: Int, f=sigm) = Dense(param(o, i), param0(o), f)

define the suitable Param structure.

You want to type it like

struct Dense{ T <: Union{Matrix{Float32}, CuArray{Float32, 2, CUDA.Mem.DeviceBuffer}} Y <: Union{Vector{Float32}, CuArray{Float32, 1, CUDA.Mem.DeviceBuffer}} } w :: Param{T} b :: Param{Y}

?

Yes, for example.

Does

struct Dense{T, Y} w :: Param{T} b :: Param{Y}

work?

It should work.
I typed it as accurate as I can.
I don't know about the performance (of Union), but I think it is more clear as it is than just T and Y.

docs/src/tutorial.md

paraynaud · 2022-07-15T20:02:57Z

I can't test it on GPU right now, all GPUs are used.

paraynaud · 2022-07-27T18:03:50Z

I tested it on Atlas, and it worked.

dpo · 2022-07-28T11:41:51Z

README.md

-set_size_minibatch!(DenseNetNLPModel, p) # p::Int > 1
-```
+## How to use
+Check the [tutorial](https://juliasmoothoptimizers.github.io/KnetNLPModels.jl/dev/tutorial/).


Better point to stable.

dpo · 2022-07-28T11:43:22Z

docs/Project.toml

@@ -1,5 +1,10 @@
 [deps]
+CUDA = "052768ef-5323-5732-b1bb-66c8b64840ba"


VMs don't always have a GPU. Why not run the docs on CPU?

I added it to run the documentation (mainly the tutorial) on Atlas.
It will be remove by the next commit.

(update) If we remove CUDA, we can't type the Dense structure like this:

struct Dense{ T <: Union{Matrix{Float32}, CuArray{Float32, 2, CUDA.Mem.DeviceBuffer}} Y <: Union{Vector{Float32}, CuArray{Float32, 1, CUDA.Mem.DeviceBuffer}} } w :: Param{T} b :: Param{Y}

which makes documentation run fail.

Use the more general typing I suggested above. I don't think the Union adds anything.

It is done, I think we are ready to go.

dpo · 2022-07-28T11:43:58Z

docs/src/index.md

@@ -18,3 +22,6 @@ In addition, it provides tools to:
 - switch the minibatch used to evaluate the neural network;
 - change the minibatch size;
 - measure the neural network's accuracy at the current `w`.
+
+## How to use
+Check the [tutorial](https://juliasmoothoptimizers.github.io/KnetNLPModels.jl/dev/tutorial/).


Why does this point to dev?

dpo · 2022-07-28T11:44:25Z

docs/src/tutorial.md

@@ -1,6 +1,7 @@
 # KnetNLPModels.jl Tutorial

 ## Synopsis
+KnetNLPModels is an interface between [Knet.jl](https://github.com/denizyuret/Knet.jl.git)'s classification neural networks and [NLPModels.jl](https://github.com/JuliaSmoothOptimizers/NLPModels.jl.git).


too much repetition

…NLPModel

Co-authored-by: Dominique <dominique.orban@gmail.com>

…to index.md

…ect.toml

dpo

Great, thank you!

dpo requested changes Jul 7, 2022

View reviewed changes

paraynaud force-pushed the executable_tutorial branch from ed41066 to 0a9c485 Compare July 15, 2022 16:30

dpo requested changes Jul 28, 2022

View reviewed changes

paraynaud force-pushed the executable_tutorial branch from aa5d311 to dba6154 Compare August 2, 2022 12:09

RAYNAUD Paul (raynaudp) and others added 9 commits August 10, 2022 10:55

executive tutorial + add required dependencies + fix README _initKnet…

f00092b

…NLPModel

fix manual downloading of the dataset by @example

fd16d28

parametrize Dense-layer struct to support CPU/GPU evaluation

bd9018c

Update docs/src/tutorial.md

ff1e85b

Co-authored-by: Dominique <dominique.orban@gmail.com>

Update docs/src/tutorial.md

4ccfe2c

Co-authored-by: Dominique <dominique.orban@gmail.com>

remove tutorial from README + adjust tutorial's text + copy README in…

7b415d5

…to index.md

dev/ -> stable/ + remove synopsis tutorial.md + remove CUDA docs/Proj…

9546375

…ect.toml

add CUDA to docs/Project.toml

b2ae378

modify Dense parametrization

ff91f7c

paraynaud force-pushed the executable_tutorial branch from dba6154 to b2ae378 Compare August 10, 2022 14:48

dpo approved these changes Aug 11, 2022

View reviewed changes

dpo merged commit eaad4c7 into main Aug 11, 2022

dpo deleted the executable_tutorial branch August 11, 2022 19:47

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Executable tutorial #38

Executable tutorial #38

paraynaud commented Jul 6, 2022

dpo Jul 7, 2022

paraynaud Jul 7, 2022 •

edited

Loading

dpo Jul 15, 2022

dpo Jul 15, 2022

paraynaud Jul 15, 2022

paraynaud commented Jul 15, 2022 •

edited

Loading

paraynaud commented Jul 27, 2022

dpo Jul 28, 2022

dpo Jul 28, 2022

paraynaud Jul 28, 2022 •

edited

Loading

paraynaud Jul 28, 2022

dpo Aug 2, 2022

paraynaud Aug 11, 2022

dpo Jul 28, 2022

dpo Jul 28, 2022

dpo left a comment

		@@ -1,5 +1,10 @@
		[deps]
		CUDA = "052768ef-5323-5732-b1bb-66c8b64840ba"

Executable tutorial #38

Executable tutorial #38

Conversation

paraynaud commented Jul 6, 2022

Choose a reason for hiding this comment

paraynaud Jul 7, 2022 • edited Loading

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

paraynaud commented Jul 15, 2022 • edited Loading

paraynaud commented Jul 27, 2022

Choose a reason for hiding this comment

Choose a reason for hiding this comment

paraynaud Jul 28, 2022 • edited Loading

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

dpo left a comment

Choose a reason for hiding this comment

paraynaud Jul 7, 2022 •

edited

Loading

paraynaud commented Jul 15, 2022 •

edited

Loading

paraynaud Jul 28, 2022 •

edited

Loading