Spaces:
Running
Running
File size: 6,213 Bytes
0c0f086 b63ff12 0c0f086 07482ee 633e287 07482ee 633e287 07482ee 633e287 07482ee 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 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 |
"""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": "g6e.12xlarge",
"configurations": [
{
"quantization": "none",
"container": "TGI 2.2.0",
"status": "OK",
"tokensPerSecond": "17",
},
{
"quantization": "none",
"container": "vLLM 0.5.5",
"status": "OK",
"tokensPerSecond": "17.8",
},
{
"quantization": "none",
"container": "SGLang 0.2.13",
"status": "OK",
"tokensPerSecond": "18.2",
},
],
},
{
"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": "p5.48xlarge",
"quantization": "none",
"container": "vLLM 0.6.4.post1",
"status": "OK",
"tokensPerSecond": "76",
"notes": "--tensor-parallel-size 8",
},
{
"instanceType": "p5.48xlarge (4 GPUs)",
"quantization": "none",
"container": "vLLM 0.6.4.post1",
"status": "OK",
"tokensPerSecond": "51",
"notes": "--tensor-parallel-size 4",
},
{
"instanceType": "p5.48xlarge (2 GPUs)",
"quantization": "none",
"container": "vLLM 0.6.4.post1",
"status": "OK",
"tokensPerSecond": "32",
"notes": "--tensor-parallel-size 2 --max-model-len 16384 --gpu_memory-utilization 0.95",
},
],
}
|