Skip to content

Commit e66c4e0

Browse files
committed
overview (#36)
* overview * fix
1 parent ca99c10 commit e66c4e0

File tree

3 files changed

+212
-12
lines changed

3 files changed

+212
-12
lines changed

nnvm/README.md

Lines changed: 14 additions & 12 deletions
Original file line numberDiff line numberDiff line change
@@ -1,13 +1,13 @@
11
# NNVM: Build deep learning system by parts
22

3-
NNVM is not a deep learning library. It is a modular, decentralized and lightweight part to
4-
help build deep learning libraries.
3+
NNVM is not a deep learning library. It is a modular,
4+
decentralized and lightweight part to help build deep learning libraries.
55

66
## What is it
77

88
While most deep learning systems offer end to end solutions,
9-
it is interesting to ask if we can actually assemble a deep learning system by parts.
10-
The goal is to enable hackers can customize optimizations, target platforms and set of operators they care about.
9+
it is interesting to assemble a deep learning system by parts.
10+
The goal is to enable user to customize optimizations, target platforms and set of operators they care about.
1111
We believe that the decentralized modular system is an interesting direction.
1212

1313
The hope is that effective parts can be assembled together just like you assemble your own desktops.
@@ -17,20 +17,22 @@ while maxiziming the users' need.
1717
NNVM offers one such part, it provides a generic way to do
1818
computation graph optimization such as memory reduction, device allocation and more
1919
while being agnostic to the operator interface defintion and how operators are executed.
20-
NNVM is inspired by LLVM, aiming to be an intermediate representation library
20+
NNVM is inspired by LLVM, aiming to be a high level intermediate representation library
2121
for neural nets and computation graphs generation and optimizations.
2222

23-
## Why build deep learning system by parts
23+
See [Overview](docs/overview.md) for an introduction on what it provides.
24+
25+
## Why build learning system by parts
26+
27+
This is essentially ***Unix philosophy*** applied to machine learning system.
2428

2529
- Essential parts can be assembled in minimum way for embedding systems.
26-
- Hackers can hack the parts they need and compose with other well defined parts.
30+
- Developers can hack the parts they need and compose with other well defined parts.
2731
- Decentralized modules enable new extensions creators to own their project
2832
without creating a monothilic version.
2933

30-
## Deep learning system by parts
31-
32-
This is one way to divide the deep learning system into common parts.
33-
Each can be isolated to a modular part.
34+
Deep learning system itself is not necessary one part, for example
35+
here are some relative independent parts that can
3436

3537
- Computation graph definition, manipulation.
3638
- Computation graph intermediate optimization.
@@ -43,5 +45,5 @@ so system building can be fun and rewarding.
4345

4446
## Links
4547

46-
[MXNet](https://github.com/dmlc/mxnet) will be using NNVM as its intermediate
48+
[MXNet](https://github.com/dmlc/mxnet) is moving to NNVM as its intermediate
4749
representation layer for symbolic graphs.

nnvm/docs/overview.md

Lines changed: 196 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,196 @@
1+
# NNVM Overview
2+
3+
NNVM is a C++ library to help developers to build deep learning system.
4+
It provides ways to construct, represent and transform computation graphs
5+
invariant of how it is executed.
6+
7+
To begin with, let us start with a few stories to tell the design goals.
8+
9+
## Stories and Design Goals
10+
11+
X has built a new deep learning framework for image classification for fun,
12+
with the modular tools like CuDNN and CUDA, it is not hard to assemble a C++ API.
13+
However, most users like to use python/R/scala or other languages.
14+
By registering the operators to NNVM, X can now get the graph composition
15+
language front-end on these languages quickly without coding it up for
16+
each type of langugage.
17+
18+
Y want to build a deep learning serving system on embedded devices.
19+
To do that, we need to cut things off, as opposed to add new parts,
20+
because codes such as gradient calculation multi-GPU scheduling is NOT relevant.
21+
It is hard to build things from scratch as well, because we want to
22+
reuse components such as memory optimization and kernel execution.
23+
It is hard to do so in current frameworks because all these information
24+
are tied to the operator interface. We want to be able to keep
25+
certain part of the system we need and throw away other parts
26+
to get the minimum system we need.
27+
28+
Z want to extend an existing deep learning system by adding a new feature,
29+
say FPGA execution of some operators. To do so Z need to add a interface like ```FPGAKernel```
30+
to the operators. E want to do another new feature that generate code for
31+
certain subset of operations. Then interface like ```GenLLVMCode``` need to be added
32+
to the operator. Eventually the system end up with a fat operator interface
33+
in order to support everything (while everyone only want some part of it).
34+
35+
We can think more stories, as the deep learning landscape shifts to more devices
36+
applications and scenarios. It is desirable to have different specialized
37+
learning system to solve some problem well,
38+
39+
Here is a list of things we want:
40+
- Minimum dependency
41+
- Being able to assemble some part together while discarding some other parts
42+
- No centralized operator interface but still allow user to provide various information about operators.
43+
44+
## Minimum Registration for a Symbolic Front-End
45+
To use NNVM to build language front end, developer only need to register
46+
minimum information about each operators.
47+
48+
```c++
49+
NNVM_REGISTER_OP(add)
50+
.describe("add two data together")
51+
.set_num_inputs(2);
52+
53+
NNVM_REGISTER_OP(conv2d)
54+
.describe("take 2d convolution of input")
55+
.set_num_inputs(2);
56+
57+
NNVM_REGISTER_OP(assign)
58+
.describe("assign second input argument to the first one")
59+
.set_num_inputs(2);
60+
```
61+
62+
Compiling the code with nnvm library. User can use the following interface
63+
to compose the computation graph in python, like the following code.
64+
65+
```python
66+
import nnvm.symbol as nn
67+
68+
# symbolic variable
69+
x = nn.Variable('x')
70+
y = nn.Variable('y')
71+
w = nn.Variable('w')
72+
73+
z = nn.conv2d(nn.add(x, y), w, filter_size=(2,2), name='conv1')
74+
```
75+
76+
The graph structure can be accessed in the backend. Currently python interface is supported.
77+
But as NNVM follows the same C bridge API design as [MXNet](https://github.com/dmlc/mxnet),
78+
which support many languages such as R/Julia/Scala/C++, more language support can be easily
79+
moved in in the future.
80+
81+
## Operator Attribute for More Extensions
82+
83+
While the minimum information provided by the operator is enough to get a front-end.
84+
In order to do transformations and executing the graph. We need more information from each operator.
85+
A typical difference between neural nets' computation graph and traditional LLVM IR is that
86+
there are a lot more high level operators. We cannot fix the set of operators in the graph.
87+
88+
Instead developers are allowed to register attributes of operator. The attributes can include shape
89+
inference function, whether the operator can be carried in-place etc.
90+
91+
This design to having an operator attribute registry is not uncommon in deep learning systems.
92+
For example, MXNet has a ```OpProperty``` class, Tensorflow has a ```OpDef``` and Caffe2 have a ```OperatorSchema``` class.
93+
However, the operator attribute interface listed in these frameworks only support a number of defined attributes of interest to the system.
94+
For example, MXNet support inplace optimization decision, shape and type inference function.
95+
If we want to extend the framework to add new type of attributes in each operator, we need to change the operator registry.
96+
Eventually the operator interface become big and have to evolve in the centralized repo.
97+
98+
In NNVM, we decided to change the design and support arbitrary type of operator attributes,
99+
without need to change the operator registry. This also echos the need of minimum interface
100+
so that the code can be easier to share accross multiple projects
101+
102+
User can register new attribute, such as inplace property checking function as follows.
103+
```c++
104+
using FInplaceOption = std::function<
105+
std::vector<std::pair<int, int> > (const NodeAttrs& attrs)>;
106+
107+
// attributes can be registered from multiple places.
108+
NNVM_REGISTER_OP(add)
109+
.set_num_inputs(1);
110+
111+
// register to tell first input can be calculate inplace with first output
112+
NNVM_REGISTER_OP(add)
113+
.attr<FInplaceOption>("FInplaceOption", [](const NodeAttrs& attrs) {
114+
return std::vector<std::pair<int, int> >{{0, 0}};
115+
});
116+
117+
NNVM_REGISTER_OP(exp)
118+
.set_num_inputs(1)
119+
.attr<FInplaceOption>("FInplaceOption", [](const NodeAttrs& attrs) {
120+
return std::vector<std::pair<int, int> >{{0, 0}};
121+
});
122+
```
123+
124+
These attributes can be queried at arbitrary parts of the code, like the following parts.
125+
Under the hood, each attributes are stored in a any type columar store,
126+
that can easily be retrieved and cast back to typed table and do quick lookups.
127+
128+
```c++
129+
void MyFunction() {
130+
const Op* add = Op::Get("add");
131+
// if we need quick query, we can use static variable
132+
// attribute map contains attributes of all operators.
133+
static auto& finplace_option_map = Op::GetAttr<FInplaceOption>("FInplaceOption");
134+
135+
// quick look up attribute of add, O(1) time, vector index lookup internally.
136+
auto add_inplace = finplace_option_tbl[add];
137+
}
138+
```
139+
Besides making the code minimum, this attribute store enables decentralization of projects.
140+
Before, all the attributes of operator have to sit on a centralized interface class.
141+
Now, everyone can register their own attribute, take some other attribute they need from another project
142+
without need to change the operator interface.
143+
144+
See [example code](../example/src/operator.cc) on how operators can be registered.
145+
146+
## Graph and Pass
147+
148+
When we get more information about the operators.
149+
We can use them to do optimizations and get more information about the graph.
150+
Graph is the unit we manipulate in these steps. A Graph in NNVM contains
151+
two parts:
152+
- The computation graph structure
153+
- A attribute map from string to any type ```map<string, shared_ptr<any> >```
154+
155+
The second attribute map is quite important, as we may need different kinds
156+
of information about the graph during the transformation process. Let it be
157+
shapes of each tensor, types of each tensor or the storage allocation plans.
158+
159+
A ```Pass``` can take a graph with existing attribute information,
160+
and transform it to the same graph with more attributes, or another graph.
161+
162+
We have bunch of pass implemented in NNVM, including symbolic differentiation,
163+
memory planning, shape/type inference and we can support more.
164+
165+
## Executing the Graph
166+
167+
Currently the library defined nothing on how the graph can be executed.
168+
Execution is intentionally excluded from this module because we believe
169+
that can be another module, and there can be many ways to execute one graph.
170+
We can target different runtime platforms, or even write your own ones.
171+
172+
More importantly, the information such as memory allocation plan,
173+
shape and type of each tensor can be used during execution phase
174+
to enhance.
175+
176+
We can also register more runtime related information to the operator registry,
177+
and define pass function to do runtime related optimization of the graph.
178+
179+
## Relation to LLVM
180+
181+
NNVM is inspired by LLVM. It is at a more high level, in a sense that there are a lot of optimization
182+
chance we can have by knowing the high level information about the operator.
183+
On the other hand, we do believe that code generation to LLVM can be a natural extension and can benefit some of the usecases.
184+
185+
## Unix Philosophy in Learning Systems
186+
187+
There are a few existing computation graph based deep learning frameworks (e.g. Theano, Tensorflow, Caffe2, MXNet etc.).
188+
NNVM do not intend to become another one. Instead, NNVM summarizes a module that contains
189+
190+
- The graph representation is minimum, with no code dependency
191+
- Operator attribute allow arbitrary information registered in unified way
192+
- Invariant of execution layer to be re-targetable to multiple frontend and backend.
193+
194+
We believe this is the correct way for learning system.
195+
By having more such modules, we can pick one ones we need, and remove the ones we do not want in our use cases.
196+
Hopefully these effort can make deep learning system research and building easy, fun and rewarding.

nnvm/example/src/operator.cc

Lines changed: 2 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -1,5 +1,7 @@
11
// Copyright (c) 2016 by Contributors
22
// This is an example on how we can register operator information to NNVM
3+
// these operator information are used to support various graph building and optimizations
4+
// see tests/python/ folder for the test-cases that uses these information.
35

46
#include <nnvm/base.h>
57
#include <nnvm/op.h>

0 commit comments

Comments
 (0)