Hugging Face's logo

Datasets:
vnmoorthy
/
pavo-bench

Tasks:
Automatic Speech Recognition
Text Generation
Text-to-Speech
Modalities:
Document
Languages:
English
Size:
10K<n<100K
Tags:
pavo
benchmark
asr
llm
tts
pipeline-routing
License:
Dataset card Data Studio Files
xet
 Community
vnmoorthy commited on
Commit
41fc622
·
verified ·
1 Parent(s): c006cfc

Upload results/tier1_llm_latency_results.json with huggingface_hub

Browse files
Files changed (1) hide show
  1. results/tier1_llm_latency_results.json +182 -0
results/tier1_llm_latency_results.json ADDED
@@ -0,0 +1,182 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "llama3.1:8b_short": {
3
+ "model": "llama3.1:8b",
4
+ "context": "short",
5
+ "context_description": "Short (50 tokens)",
6
+ "num_predict": 50,
7
+ "n_measurements": 100,
8
+ "total_latency_ms": {
9
+ "mean": 551.72,
10
+ "std": 70.44,
11
+ "median": 576.87,
12
+ "p95": 605.78,
13
+ "p99": 612.48,
14
+ "min": 339.01,
15
+ "max": 614.75
16
+ },
17
+ "time_to_first_token_ms": {
18
+ "mean": 7.45,
19
+ "std": 2.55,
20
+ "median": 6.61
21
+ },
22
+ "tokens_per_second": {
23
+ "mean": 157.67,
24
+ "std": 1.74,
25
+ "median": 157.15
26
+ },
27
+ "output_tokens": {
28
+ "mean": 45.9,
29
+ "std": 9.68
30
+ }
31
+ },
32
+ "llama3.1:8b_medium": {
33
+ "model": "llama3.1:8b",
34
+ "context": "medium",
35
+ "context_description": "Medium (300 tokens)",
36
+ "num_predict": 300,
37
+ "n_measurements": 50,
38
+ "total_latency_ms": {
39
+ "mean": 2241.54,
40
+ "std": 300.39,
41
+ "median": 2389.78,
42
+ "p95": 2413.01,
43
+ "p99": 2426.19,
44
+ "min": 1611.02,
45
+ "max": 2428.77
46
+ },
47
+ "time_to_first_token_ms": {
48
+ "mean": 9.41,
49
+ "std": 3.75,
50
+ "median": 8.52
51
+ },
52
+ "tokens_per_second": {
53
+ "mean": 153.85,
54
+ "std": 0.55,
55
+ "median": 153.66
56
+ },
57
+ "output_tokens": {
58
+ "mean": 279.52,
59
+ "std": 41.11
60
+ }
61
+ },
62
+ "llama3.1:8b_long": {
63
+ "model": "llama3.1:8b",
64
+ "context": "long",
65
+ "context_description": "Long (800 tokens)",
66
+ "num_predict": 800,
67
+ "n_measurements": 30,
68
+ "total_latency_ms": {
69
+ "mean": 5604.46,
70
+ "std": 576.71,
71
+ "median": 5961.43,
72
+ "p95": 6018.53,
73
+ "p99": 6031.08,
74
+ "min": 4708.78,
75
+ "max": 6034.02
76
+ },
77
+ "time_to_first_token_ms": {
78
+ "mean": 12.04,
79
+ "std": 10.41,
80
+ "median": 8.53
81
+ },
82
+ "tokens_per_second": {
83
+ "mean": 151.74,
84
+ "std": 0.23,
85
+ "median": 151.66
86
+ },
87
+ "output_tokens": {
88
+ "mean": 747.5,
89
+ "std": 80.2
90
+ }
91
+ },
92
+ "gemma2:2b_short": {
93
+ "model": "gemma2:2b",
94
+ "context": "short",
95
+ "context_description": "Short (50 tokens)",
96
+ "num_predict": 50,
97
+ "n_measurements": 100,
98
+ "total_latency_ms": {
99
+ "mean": 452.28,
100
+ "std": 54.85,
101
+ "median": 480.17,
102
+ "p95": 498.86,
103
+ "p99": 511.55,
104
+ "min": 316.09,
105
+ "max": 516.29
106
+ },
107
+ "time_to_first_token_ms": {
108
+ "mean": 4.83,
109
+ "std": 1.66,
110
+ "median": 4.27
111
+ },
112
+ "tokens_per_second": {
113
+ "mean": 245.54,
114
+ "std": 4.98,
115
+ "median": 245.59
116
+ },
117
+ "output_tokens": {
118
+ "mean": 43.38,
119
+ "std": 10.09
120
+ }
121
+ },
122
+ "gemma2:2b_medium": {
123
+ "model": "gemma2:2b",
124
+ "context": "medium",
125
+ "context_description": "Medium (300 tokens)",
126
+ "num_predict": 300,
127
+ "n_measurements": 50,
128
+ "total_latency_ms": {
129
+ "mean": 1822.94,
130
+ "std": 120.22,
131
+ "median": 1875.08,
132
+ "p95": 1909.88,
133
+ "p99": 1921.75,
134
+ "min": 1476.14,
135
+ "max": 1923.77
136
+ },
137
+ "time_to_first_token_ms": {
138
+ "mean": 6.83,
139
+ "std": 1.73,
140
+ "median": 6.25
141
+ },
142
+ "tokens_per_second": {
143
+ "mean": 239.58,
144
+ "std": 0.43,
145
+ "median": 239.59
146
+ },
147
+ "output_tokens": {
148
+ "mean": 289.0,
149
+ "std": 22.16
150
+ }
151
+ },
152
+ "gemma2:2b_long": {
153
+ "model": "gemma2:2b",
154
+ "context": "long",
155
+ "context_description": "Long (800 tokens)",
156
+ "num_predict": 800,
157
+ "n_measurements": 30,
158
+ "total_latency_ms": {
159
+ "mean": 4430.18,
160
+ "std": 438.05,
161
+ "median": 4712.75,
162
+ "p95": 4745.05,
163
+ "p99": 4783.99,
164
+ "min": 3752.64,
165
+ "max": 4796.41
166
+ },
167
+ "time_to_first_token_ms": {
168
+ "mean": 7.56,
169
+ "std": 4.06,
170
+ "median": 6.19
171
+ },
172
+ "tokens_per_second": {
173
+ "mean": 234.56,
174
+ "std": 0.58,
175
+ "median": 234.37
176
+ },
177
+ "output_tokens": {
178
+ "mean": 747.97,
179
+ "std": 79.19
180
+ }
181
+ }
182
+ }