Comparison Between TreeValue and Tianshou Batch¶
In this section, we will take a look at the feature and performance of the Tianshou Batch library, which is developed by Tsinghua Machine Learning Group.
Before starting the comparison, let us define some thing.
[1]:
import torch
_TREE_DATA_1 = {'a': 1, 'b': 2, 'x': {'c': 3, 'd': 4}}
_TREE_DATA_2 = {
'a': torch.randn(2, 3),
'x': {
'c': torch.randn(3, 4)
},
}
_TREE_DATA_3 = {
'obs': torch.randn(4, 84, 84),
'action': torch.randint(0, 6, size=(1,)),
'reward': torch.rand(1),
}
Read and Write Operation¶
Reading and writing are the two most common operations in the tree data structure based on the data model (TreeValue and Tianshou Batch both belong to this type), so this section will compare the reading and writing performance of these two libraries.
TreeValue’s Get and Set¶
[2]:
from treevalue import FastTreeValue
t = FastTreeValue(_TREE_DATA_2)
[3]:
t
[3]:
<FastTreeValue 0x7fd004ba3e20>
├── 'a' --> tensor([[0.0160, 0.9517, 0.7109],
│ [0.9249, 0.6371, 0.0753]])
└── 'x' --> <FastTreeValue 0x7fd004bf1b20>
└── 'c' --> tensor([[-0.1180, 1.1987, 0.7311, -0.3844],
[-1.3379, 0.8012, -1.3079, -0.1670],
[-0.5804, 0.1720, 1.1919, -0.9883]])
[4]:
t.a
[4]:
tensor([[0.0160, 0.9517, 0.7109],
[0.9249, 0.6371, 0.0753]])
[5]:
%timeit t.a
49.3 ns ± 0.166 ns per loop (mean ± std. dev. of 7 runs, 10,000,000 loops each)
[6]:
new_value = torch.randn(2, 3)
t.a = new_value
t
[6]:
<FastTreeValue 0x7fd004ba3e20>
├── 'a' --> tensor([[-0.3580, 0.5381, 0.3574],
│ [-2.2296, 0.5389, -0.3477]])
└── 'x' --> <FastTreeValue 0x7fd004bf1b20>
└── 'c' --> tensor([[-0.1180, 1.1987, 0.7311, -0.3844],
[-1.3379, 0.8012, -1.3079, -0.1670],
[-0.5804, 0.1720, 1.1919, -0.9883]])
[7]:
%timeit t.a = new_value
54.6 ns ± 0.491 ns per loop (mean ± std. dev. of 7 runs, 10,000,000 loops each)
Tianshou Batch’s Get and Set¶
[8]:
from tianshou.data import Batch
b = Batch(**_TREE_DATA_2)
[9]:
b
[9]:
Batch(
a: tensor([[0.0160, 0.9517, 0.7109],
[0.9249, 0.6371, 0.0753]]),
x: Batch(
c: tensor([[-0.1180, 1.1987, 0.7311, -0.3844],
[-1.3379, 0.8012, -1.3079, -0.1670],
[-0.5804, 0.1720, 1.1919, -0.9883]]),
),
)
[10]:
b.a
[10]:
tensor([[0.0160, 0.9517, 0.7109],
[0.9249, 0.6371, 0.0753]])
[11]:
%timeit b.a
40.9 ns ± 0.305 ns per loop (mean ± std. dev. of 7 runs, 10,000,000 loops each)
[12]:
new_value = torch.randn(2, 3)
b.a = new_value
b
[12]:
Batch(
a: tensor([[ 1.6719, -2.5417, -0.6172],
[-1.0708, -0.9851, -0.1983]]),
x: Batch(
c: tensor([[-0.1180, 1.1987, 0.7311, -0.3844],
[-1.3379, 0.8012, -1.3079, -0.1670],
[-0.5804, 0.1720, 1.1919, -0.9883]]),
),
)
[13]:
%timeit b.a = new_value
363 ns ± 1.47 ns per loop (mean ± std. dev. of 7 runs, 1,000,000 loops each)
Initialization¶
TreeValue’s Initialization¶
[14]:
%timeit FastTreeValue(_TREE_DATA_1)
619 ns ± 4.31 ns per loop (mean ± std. dev. of 7 runs, 1,000,000 loops each)
Tianshou Batch’s Initialization¶
[15]:
%timeit Batch(**_TREE_DATA_1)
8.55 µs ± 42 ns per loop (mean ± std. dev. of 7 runs, 100,000 loops each)
Deep Copy Operation¶
[16]:
import copy
Deep Copy of TreeValue¶
[17]:
t3 = FastTreeValue(_TREE_DATA_3)
%timeit copy.deepcopy(t3)
127 µs ± 712 ns per loop (mean ± std. dev. of 7 runs, 10,000 loops each)
Deep Copy of Tianshou Batch¶
[18]:
b3 = Batch(**_TREE_DATA_3)
%timeit copy.deepcopy(b3)
124 µs ± 549 ns per loop (mean ± std. dev. of 7 runs, 10,000 loops each)
Stack, Concat and Split Operation¶
Performance of TreeValue¶
[19]:
trees = [FastTreeValue(_TREE_DATA_2) for _ in range(8)]
[20]:
t_stack = FastTreeValue.func(subside=True)(torch.stack)
t_stack(trees)
[20]:
<FastTreeValue 0x7fcf27f20250>
├── 'a' --> tensor([[[0.0160, 0.9517, 0.7109],
│ [0.9249, 0.6371, 0.0753]],
│
│ [[0.0160, 0.9517, 0.7109],
│ [0.9249, 0.6371, 0.0753]],
│
│ [[0.0160, 0.9517, 0.7109],
│ [0.9249, 0.6371, 0.0753]],
│
│ [[0.0160, 0.9517, 0.7109],
│ [0.9249, 0.6371, 0.0753]],
│
│ [[0.0160, 0.9517, 0.7109],
│ [0.9249, 0.6371, 0.0753]],
│
│ [[0.0160, 0.9517, 0.7109],
│ [0.9249, 0.6371, 0.0753]],
│
│ [[0.0160, 0.9517, 0.7109],
│ [0.9249, 0.6371, 0.0753]],
│
│ [[0.0160, 0.9517, 0.7109],
│ [0.9249, 0.6371, 0.0753]]])
└── 'x' --> <FastTreeValue 0x7fcf1ec6ff40>
└── 'c' --> tensor([[[-0.1180, 1.1987, 0.7311, -0.3844],
[-1.3379, 0.8012, -1.3079, -0.1670],
[-0.5804, 0.1720, 1.1919, -0.9883]],
[[-0.1180, 1.1987, 0.7311, -0.3844],
[-1.3379, 0.8012, -1.3079, -0.1670],
[-0.5804, 0.1720, 1.1919, -0.9883]],
[[-0.1180, 1.1987, 0.7311, -0.3844],
[-1.3379, 0.8012, -1.3079, -0.1670],
[-0.5804, 0.1720, 1.1919, -0.9883]],
[[-0.1180, 1.1987, 0.7311, -0.3844],
[-1.3379, 0.8012, -1.3079, -0.1670],
[-0.5804, 0.1720, 1.1919, -0.9883]],
[[-0.1180, 1.1987, 0.7311, -0.3844],
[-1.3379, 0.8012, -1.3079, -0.1670],
[-0.5804, 0.1720, 1.1919, -0.9883]],
[[-0.1180, 1.1987, 0.7311, -0.3844],
[-1.3379, 0.8012, -1.3079, -0.1670],
[-0.5804, 0.1720, 1.1919, -0.9883]],
[[-0.1180, 1.1987, 0.7311, -0.3844],
[-1.3379, 0.8012, -1.3079, -0.1670],
[-0.5804, 0.1720, 1.1919, -0.9883]],
[[-0.1180, 1.1987, 0.7311, -0.3844],
[-1.3379, 0.8012, -1.3079, -0.1670],
[-0.5804, 0.1720, 1.1919, -0.9883]]])
[21]:
%timeit t_stack(trees)
24.3 µs ± 83.6 ns per loop (mean ± std. dev. of 7 runs, 10,000 loops each)
[22]:
t_cat = FastTreeValue.func(subside=True)(torch.cat)
t_cat(trees)
[22]:
<FastTreeValue 0x7fcf1ec71c40>
├── 'a' --> tensor([[0.0160, 0.9517, 0.7109],
│ [0.9249, 0.6371, 0.0753],
│ [0.0160, 0.9517, 0.7109],
│ [0.9249, 0.6371, 0.0753],
│ [0.0160, 0.9517, 0.7109],
│ [0.9249, 0.6371, 0.0753],
│ [0.0160, 0.9517, 0.7109],
│ [0.9249, 0.6371, 0.0753],
│ [0.0160, 0.9517, 0.7109],
│ [0.9249, 0.6371, 0.0753],
│ [0.0160, 0.9517, 0.7109],
│ [0.9249, 0.6371, 0.0753],
│ [0.0160, 0.9517, 0.7109],
│ [0.9249, 0.6371, 0.0753],
│ [0.0160, 0.9517, 0.7109],
│ [0.9249, 0.6371, 0.0753]])
└── 'x' --> <FastTreeValue 0x7fcf1fcda040>
└── 'c' --> tensor([[-0.1180, 1.1987, 0.7311, -0.3844],
[-1.3379, 0.8012, -1.3079, -0.1670],
[-0.5804, 0.1720, 1.1919, -0.9883],
[-0.1180, 1.1987, 0.7311, -0.3844],
[-1.3379, 0.8012, -1.3079, -0.1670],
[-0.5804, 0.1720, 1.1919, -0.9883],
[-0.1180, 1.1987, 0.7311, -0.3844],
[-1.3379, 0.8012, -1.3079, -0.1670],
[-0.5804, 0.1720, 1.1919, -0.9883],
[-0.1180, 1.1987, 0.7311, -0.3844],
[-1.3379, 0.8012, -1.3079, -0.1670],
[-0.5804, 0.1720, 1.1919, -0.9883],
[-0.1180, 1.1987, 0.7311, -0.3844],
[-1.3379, 0.8012, -1.3079, -0.1670],
[-0.5804, 0.1720, 1.1919, -0.9883],
[-0.1180, 1.1987, 0.7311, -0.3844],
[-1.3379, 0.8012, -1.3079, -0.1670],
[-0.5804, 0.1720, 1.1919, -0.9883],
[-0.1180, 1.1987, 0.7311, -0.3844],
[-1.3379, 0.8012, -1.3079, -0.1670],
[-0.5804, 0.1720, 1.1919, -0.9883],
[-0.1180, 1.1987, 0.7311, -0.3844],
[-1.3379, 0.8012, -1.3079, -0.1670],
[-0.5804, 0.1720, 1.1919, -0.9883]])
[23]:
%timeit t_cat(trees)
22.6 µs ± 219 ns per loop (mean ± std. dev. of 7 runs, 10,000 loops each)
[24]:
t_split = FastTreeValue.func(rise=True)(torch.split)
tree = FastTreeValue({
'obs': torch.randn(8, 4, 84, 84),
'action': torch.randint(0, 6, size=(8, 1,)),
'reward': torch.rand(8, 1),
})
%timeit t_split(tree, 1)
49.7 µs ± 813 ns per loop (mean ± std. dev. of 7 runs, 10,000 loops each)
Performance of Tianshou Batch¶
[25]:
batches = [Batch(**_TREE_DATA_2) for _ in range(8)]
Batch.stack(batches)
[25]:
Batch(
a: tensor([[[0.0160, 0.9517, 0.7109],
[0.9249, 0.6371, 0.0753]],
[[0.0160, 0.9517, 0.7109],
[0.9249, 0.6371, 0.0753]],
[[0.0160, 0.9517, 0.7109],
[0.9249, 0.6371, 0.0753]],
[[0.0160, 0.9517, 0.7109],
[0.9249, 0.6371, 0.0753]],
[[0.0160, 0.9517, 0.7109],
[0.9249, 0.6371, 0.0753]],
[[0.0160, 0.9517, 0.7109],
[0.9249, 0.6371, 0.0753]],
[[0.0160, 0.9517, 0.7109],
[0.9249, 0.6371, 0.0753]],
[[0.0160, 0.9517, 0.7109],
[0.9249, 0.6371, 0.0753]]]),
x: Batch(
c: tensor([[[-0.1180, 1.1987, 0.7311, -0.3844],
[-1.3379, 0.8012, -1.3079, -0.1670],
[-0.5804, 0.1720, 1.1919, -0.9883]],
[[-0.1180, 1.1987, 0.7311, -0.3844],
[-1.3379, 0.8012, -1.3079, -0.1670],
[-0.5804, 0.1720, 1.1919, -0.9883]],
[[-0.1180, 1.1987, 0.7311, -0.3844],
[-1.3379, 0.8012, -1.3079, -0.1670],
[-0.5804, 0.1720, 1.1919, -0.9883]],
[[-0.1180, 1.1987, 0.7311, -0.3844],
[-1.3379, 0.8012, -1.3079, -0.1670],
[-0.5804, 0.1720, 1.1919, -0.9883]],
[[-0.1180, 1.1987, 0.7311, -0.3844],
[-1.3379, 0.8012, -1.3079, -0.1670],
[-0.5804, 0.1720, 1.1919, -0.9883]],
[[-0.1180, 1.1987, 0.7311, -0.3844],
[-1.3379, 0.8012, -1.3079, -0.1670],
[-0.5804, 0.1720, 1.1919, -0.9883]],
[[-0.1180, 1.1987, 0.7311, -0.3844],
[-1.3379, 0.8012, -1.3079, -0.1670],
[-0.5804, 0.1720, 1.1919, -0.9883]],
[[-0.1180, 1.1987, 0.7311, -0.3844],
[-1.3379, 0.8012, -1.3079, -0.1670],
[-0.5804, 0.1720, 1.1919, -0.9883]]]),
),
)
[26]:
%timeit Batch.stack(batches)
63.1 µs ± 534 ns per loop (mean ± std. dev. of 7 runs, 10,000 loops each)
[27]:
Batch.cat(batches)
[27]:
Batch(
a: tensor([[0.0160, 0.9517, 0.7109],
[0.9249, 0.6371, 0.0753],
[0.0160, 0.9517, 0.7109],
[0.9249, 0.6371, 0.0753],
[0.0160, 0.9517, 0.7109],
[0.9249, 0.6371, 0.0753],
[0.0160, 0.9517, 0.7109],
[0.9249, 0.6371, 0.0753],
[0.0160, 0.9517, 0.7109],
[0.9249, 0.6371, 0.0753],
[0.0160, 0.9517, 0.7109],
[0.9249, 0.6371, 0.0753],
[0.0160, 0.9517, 0.7109],
[0.9249, 0.6371, 0.0753],
[0.0160, 0.9517, 0.7109],
[0.9249, 0.6371, 0.0753]]),
x: Batch(
c: tensor([[-0.1180, 1.1987, 0.7311, -0.3844],
[-1.3379, 0.8012, -1.3079, -0.1670],
[-0.5804, 0.1720, 1.1919, -0.9883],
[-0.1180, 1.1987, 0.7311, -0.3844],
[-1.3379, 0.8012, -1.3079, -0.1670],
[-0.5804, 0.1720, 1.1919, -0.9883],
[-0.1180, 1.1987, 0.7311, -0.3844],
[-1.3379, 0.8012, -1.3079, -0.1670],
[-0.5804, 0.1720, 1.1919, -0.9883],
[-0.1180, 1.1987, 0.7311, -0.3844],
[-1.3379, 0.8012, -1.3079, -0.1670],
[-0.5804, 0.1720, 1.1919, -0.9883],
[-0.1180, 1.1987, 0.7311, -0.3844],
[-1.3379, 0.8012, -1.3079, -0.1670],
[-0.5804, 0.1720, 1.1919, -0.9883],
[-0.1180, 1.1987, 0.7311, -0.3844],
[-1.3379, 0.8012, -1.3079, -0.1670],
[-0.5804, 0.1720, 1.1919, -0.9883],
[-0.1180, 1.1987, 0.7311, -0.3844],
[-1.3379, 0.8012, -1.3079, -0.1670],
[-0.5804, 0.1720, 1.1919, -0.9883],
[-0.1180, 1.1987, 0.7311, -0.3844],
[-1.3379, 0.8012, -1.3079, -0.1670],
[-0.5804, 0.1720, 1.1919, -0.9883]]),
),
)
[28]:
%timeit Batch.cat(batches)
119 µs ± 486 ns per loop (mean ± std. dev. of 7 runs, 10,000 loops each)
[29]:
batch = Batch({
'obs': torch.randn(8, 4, 84, 84),
'action': torch.randint(0, 6, size=(8, 1,)),
'reward': torch.rand(8, 1)}
)
%timeit list(Batch.split(batch, 1, shuffle=False, merge_last=True))
281 µs ± 2.63 µs per loop (mean ± std. dev. of 7 runs, 1,000 loops each)
[ ]: