Fluid distributed training performance of several models

	A	B	C	D	E	F	G	H	I	J
1
2	Model	CPU/GPU	Figure	Dataset	batch_size	Pservers	Trainers	GPUs per Trainer	Accuracy	Code
3	text_classification	GPU	figure-1	imdb	512	0	1	1	Pass id: 300, test_acc: 0.859640	https://github.com/typhoonzero/fluid_gpu_benchmark/blob/master/text_fluid.py
4	text_classification	GPU	figure-1	imdb	256x2	2	2	1	Pass id: 300, test_acc: 0.864200
5
6	sequence_tagging_ner	CPU	figure-2	imikolov	200	0	1		pass_id:200 pass_precision:[0.5405262] pass_recall:[0.51514643] pass_f1_score:[0.52753127]	https://github.com/typhoonzero/fluid_gpu_benchmark/tree/master/seq_tag_ner
7	sequence_tagging_ner	CPU	figure-2	imikolov	50x4	4	4		pass_id:200 pass_precision:[0.5758079] pass_recall:[0.52473915] pass_f1_score:[0.54908866]
8
9	language_model	GPU		imikolov	20	0	1	1	Pass 19, test avg ppl: 52.138	https://github.com/typhoonzero/fluid_gpu_benchmark/tree/master/lm
10	language_model	GPU		imikolov		4	4	1	Pass 19, test avg ppl: 52.209
11
12	transformer	GPU		nist06n		0	1	1	local BLEU: 0.4197
13		GPU		nist06n		4	8	1	dist BLEU:0.3910	https://github.com/gongweibao/transformer-nist/tree/add-data-util/transformer
14
15	mobilenet	CPU		flowers	40	0	1		pass 4, train_acc [0.00325256], test_acc [0.00980392]
16	mobilenet	CPU		flowers		4	4		pass 4, train_acc [0.00260247], test_acc [0.00980392]
17
18	se_resnext	CPU		flowers	16	0	1		pass 0 test_acc 0.00961538497359
19	se_resnext	CPU		flowers	16	4	4		pass 0 test_acc 0.0105769233778
20
21	vgg16	CPU		flowers	40	0	1	0	Pass = 3 Test accuracy = 0.009804	https://github.com/typhoonzero/fluid_gpu_benchmark/blob/master/vgg16_fluid.py
22	vgg16	CPU		flowers	40	4	4	0	Pass = 3 Test accuracy = 0.009804
23
24
25
26
27
28
29
30
31
32
33
34
35	Figure-1							Figure-2
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100