File size: 3,749 Bytes
0c0f086
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
f0fe385
 
0c0f086
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
d0c0894
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
0c0f086
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
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
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
"""Module containing performance results for the Arcee-Lite model."""

results_arcee_lite = {
    "name": "Arcee-Lite",
    "modelType": "Qwen2 1.5B distilled from phi-3-medium 14B",
    "configurations": [
        {
            "instanceType": "c6i.xlarge",
            "quantization": "bitsandbytes-nf4",
            "container": "TGI 2.2.0",
            "status": "KO",
            "tokensPerSecond": "-",
            "notes": "OOM, might work with a prequantized model",
        },
        {
            "instanceType": "c6i.2xlarge",
            "quantization": "bitsandbytes-nf4",
            "container": "TGI 2.2.0",
            "status": "KO",
            "tokensPerSecond": "-",
            "notes": "OOM, might work with a prequantized model",
        },
        {
            "instanceType": "c6i.4xlarge",
            "configurations": [
                {
                    "quantization": "none",
                    "container": "TGI 2.2.0",
                    "status": "OK",
                    "tokensPerSecond": "10.7",
                },
                {
                    "quantization": "bitsandbytes (int8)",
                    "container": "TGI 2.2.0",
                    "status": "OK",
                    "tokensPerSecond": "10.5",
                },
                {
                    "quantization": "bitsandbytes-nf4",
                    "container": "TGI 2.2.0",
                    "status": "OK",
                    "tokensPerSecond": "10.6",
                },
            ],
        },
        {
            "instanceType": "c7i.4xlarge",
            "quantization": "none",
            "container": "TGI 2.2.0",
            "status": "OK",
            "tokensPerSecond": "11.4",
        },
        {
            "instanceType": "g5.xlarge",
            "configurations": [
                {
                    "quantization": "none",
                    "container": "TGI 2.2.0",
                    "status": "OK",
                    "tokensPerSecond": "110",
                },
                {
                    "quantization": "none",
                    "container": "DJL 0.28 vLLM",
                    "status": "OK",
                    "tokensPerSecond": "105",
                    "notes": '"OPTION_MAX_MODEL_LEN": "32768",',
                },
            ],
        },
        {
            "instanceType": "g6e.2xlarge",
            "configurations": [
                {
                    "container": "TGI 2.2.0",
                    "quantization": "none",
                    "status": "OK",
                    "tokensPerSecond": "160",
                },
                {
                    "container": "SGLang 0.2.13",
                    "quantization": "none",
                    "status": "OK",
                    "tokensPerSecond": "167",
                },
                {
                    "container": "vLLM 0.5.5",
                    "quantization": "none",
                    "status": "OK",
                    "tokensPerSecond": "150",
                },
            ],
        },
        {
            "instanceType": "c7g.2xlarge",
            "configurations": [
                {
                    "container": "PyTorch 2.4.1 on EC2",
                    "quantization": "none",
                    "status": "OK",
                    "tokensPerSecond": "16.5",
                },
            ],
        },
        {
            "instanceType": "c7g.8xlarge",
            "configurations": [
                {
                    "container": "PyTorch 2.4.1 on EC2",
                    "quantization": "none",
                    "status": "OK",
                    "tokensPerSecond": "26",
                },
            ],
        },
    ],
}