Spaces:
Running
Running
File size: 2,844 Bytes
0c0f086 9f46468 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 |
"""Module containing performance results for the Llama-Spark model."""
results_llama_spark = {
"name": "Llama-Spark",
"modelType": "Llama 3.1 8B",
"configurations": [
{
"instanceType": "g5.2xlarge",
"quantization": "none",
"container": "TGI 2.2.0",
"status": "OK",
"tokensPerSecond": "29",
"notes": "4K/8K fails",
},
{
"instanceType": "g5.12xlarge",
"quantization": "none",
"container": "TGI 2.2.0",
"status": "OK",
"tokensPerSecond": "85",
"notes": '"MAX_INPUT_TOKENS": "16384", "MAX_TOTAL_TOKENS": "32768",',
},
{
"instanceType": "g5.48xlarge",
"quantization": "none",
"container": "TGI 2.2.0",
"status": "OK",
"tokensPerSecond": "105",
"notes": '"MAX_INPUT_TOKENS": "20480", "MAX_TOTAL_TOKENS": "40960"\n\n32K/64K fails',
},
{
"instanceType": "g6.12xlarge",
"quantization": "none",
"container": "TGI 2.2.0",
"status": "OK",
"tokensPerSecond": "51",
"notes": "same as g5?",
},
{
"instanceType": "g6.48xlarge",
"quantization": "none",
"container": "TGI 2.2.0",
"status": "OK",
"tokensPerSecond": "81",
"notes": "same as g5?",
},
{
"instanceType": "g6e.2xlarge",
"quantization": "none",
"status": "OK",
"configurations": [
{"container": "TGI 2.2.0", "tokensPerSecond": "42.1"},
{"container": "SGLang 0.2.13", "tokensPerSecond": "45"},
{"container": "vLLM 0.5.5", "tokensPerSecond": "43.4"},
],
},
{
"instanceType": "g6e.12xlarge",
"quantization": "none",
"status": "OK",
"configurations": [
{"container": "TGI 2.2.0", "tokensPerSecond": "112"},
{"container": "SGLang 0.2.13", "tokensPerSecond": "123"},
{"container": "vLLM 0.5.5", "tokensPerSecond": "106"},
],
},
{
"instanceType": "p4d.24xlarge",
"quantization": "none",
"container": "TGI 2.2.0",
"status": "OK",
"tokensPerSecond": "145",
"notes": '"MAX_INPUT_TOKENS": "40960", "MAX_TOTAL_TOKENS": "81920"\n\n64K/128K fails (even with 4-bit)',
},
{
"instanceType": "inf2.*",
"container": "TGI 2.2.0",
"status": "not supported",
"tokensPerSecond": "-",
"notes": "Llama-3.1: TGI OK, Neuron SDK OK, optimum-neuron KO",
},
],
}
|