Spaces:
Running
Running
File size: 4,865 Bytes
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 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 |
"""Module containing performance results for the Arcee-Nova model."""
results_arcee_nova = {
"name": "Arcee-Nova",
"modelType": "Qwen2 72B",
"notes": "",
"configurations": [
{
"instanceType": "g4dn.12xlarge",
"quantization": "bitsandbytes-nf4",
"container": "TGI 2.2.0",
"status": "KO",
"tokensPerSecond": "-",
"notes": "Flash Attention requires Ampere GPUs or newer",
},
{
"instanceType": "g5.12xlarge",
"configurations": [
{
"quantization": "bitsandbytes-nf4",
"container": "TGI 2.2.0",
"status": "OK",
"tokensPerSecond": "12",
},
{
"quantization": "bitsandbytes-fp4",
"container": "TGI 2.2.0",
"status": "OK",
"tokensPerSecond": "12",
},
{
"quantization": "bitsandbytes (int8)",
"container": "TGI 2.2.0",
"status": "KO",
"tokensPerSecond": "-",
"notes": "CUDA OOM",
},
{
"quantization": "eetq (int8)",
"container": "TGI 2.2.0",
"status": "KO",
"tokensPerSecond": "-",
"notes": "[FT Error] Heurisitc failed to find a valid config.",
},
],
},
{
"instanceType": "g5.48xlarge",
"configurations": [
{
"quantization": "none",
"container": "TGI 2.2.0",
"status": "KO",
"tokensPerSecond": "-",
"notes": "CUDA OOM (but g6.48xlarge works!)",
},
{
"quantization": "bitsandbytes-nf4",
"container": "TGI 2.2.0",
"status": "OK",
"tokensPerSecond": "12.3",
},
{
"quantization": "bitsandbytes-fp4",
"container": "TGI 2.2.0",
"status": "OK",
"tokensPerSecond": "12.5",
},
{
"quantization": "bitsandbytes (int8)",
"container": "TGI 2.2.0",
"status": "KO",
"tokensPerSecond": "-",
"notes": "The model deploys, but inference times out.",
},
],
},
{
"instanceType": "g6.12xlarge",
"configurations": [
{
"quantization": "bitsandbytes-nf4",
"container": "TGI 2.2.0",
"status": "OK",
"tokensPerSecond": "1.5-2",
"notes": "Too slow, timeouts are likely",
},
{
"quantization": "bitsandbytes-fp4",
"container": "TGI 2.2.0",
"status": "OK",
"tokensPerSecond": "2",
"notes": "Too slow, timeouts are likely",
},
{
"quantization": "bitsandbytes (int8)",
"container": "TGI 2.2.0",
"status": "KO",
"tokensPerSecond": "-",
"notes": "CUDA OOM",
},
],
},
{
"instanceType": "g6.48xlarge",
"quantization": "none",
"container": "TGI 2.2.0",
"status": "OK",
"tokensPerSecond": "12",
},
{
"instanceType": "p4d.24xlarge",
"quantization": "none",
"container": "TGI 2.2.0",
"status": "OK",
"tokensPerSecond": "40",
"notes": '"MAX_INPUT_LENGTH": "16384", "MAX_TOTAL_TOKENS": "32768",',
},
{
"instanceType": "p4de.24xlarge",
"quantization": "none",
"container": "TGI 2.2.0",
"status": "waiting for quota",
},
{
"instanceType": "p5.48xlarge",
"quantization": "none",
"container": "TGI 2.2.0",
"status": "OK",
"tokensPerSecond": "58",
"notes": '"MAX_INPUT_LENGTH": "16384", "MAX_TOTAL_TOKENS": "32768",',
},
{
"instanceType": "inf2.*",
"container": "TGI 2.2.0",
"status": "not supported",
"tokensPerSecond": "-",
"notes": "Qwen2: TGI OK, Neuron SDK KO, optimum-neuron KO",
},
],
}
|