results_arcee_supernova.py

"""Module containing performance results for the Arcee-SuperNova model."""

results_arcee_supernova = {
    "name": "Arcee-SuperNova",
    "modelType": "Llama 3.1 70B",
    "configurations": [
        {
            "instanceType": "g5.12xlarge",
            "quantization": "awq",
            "container": "TGI 2.2.0",
            "status": "OK",
            "tokensPerSecond": "33",
            "notes": "MAX_INPUT_TOKENS: 8192, MAX_TOTAL_TOKENS: 16384",
        },
        {
            "instanceType": "p4d.24xlarge",
            "quantization": "awq",
            "container": "TGI 2.2.0",
            "status": "OK",
            "tokensPerSecond": "58",
            "notes": "MAX_INPUT_TOKENS: 16384, MAX_TOTAL_TOKENS: 32768",
        },
        {
            "instanceType": "p5.48xlarge",
            "quantization": "awq",
            "container": "TGI 2.2.0",
            "status": "OK",
            "tokensPerSecond": "73",
            "notes": "MAX_INPUT_TOKENS: 16384, MAX_TOTAL_TOKENS: 32768",
        },
        {
            "instanceType": "inf2.24xlarge",
            "configurations": [
                {
                    "quantization": "none",
                    "container": "LMI 0.29+transformers-neuronx 0.11.351",
                    "status": "KO",
                    "tokensPerSecond": "-",
                    "notes": "OOM bs=2,seqlen=4096 - SDK 2.19.1",
                },
                {
                    "quantization": "none",
                    "container": "LMI 0.29+transformers-neuronx 0.11.351",
                    "status": "KO",
                    "tokensPerSecond": "-",
                    "notes": "OOM bs=2,seqlen=2048 - SDK 2.19.1",
                },
            ],
        },
        {
            "instanceType": "inf2.48xlarge",
            "configurations": [
                {
                    "quantization": "none",
                    "container": "LMI 0.29+transformers-neuronx 0.11.351",
                    "status": "OK",
                    "tokensPerSecond": "28",
                    "notes": "bs=4,seqlen=4096 - SDK 2.19.1",
                },
                {
                    "quantization": "none",
                    "container": "LMI 0.29+transformers-neuronx 0.11.351",
                    "status": "OK",
                    "tokensPerSecond": "24",
                    "notes": "bs=2,seqlen=8192 - SDK 2.19.1",
                },
                {
                    "quantization": "none",
                    "container": "LMI 0.29+transformers-neuronx 0.11.351",
                    "status": "KO",
                    "tokensPerSecond": "-",
                    "notes": "OOM bs=2,seqlen=16384 - SDK 2.19.1",
                },
            ],
        },
        {
            "instanceType": "p4d.24xlarge",
            "quantization": "none",
            "container": "TGI 2.2.0",
            "status": "OK",
            "tokensPerSecond": "30",
            "notes": "",
        },
        {
            "instanceType": "p5.48xlarge",
            "quantization": "none",
            "container": "TGI 2.2.0",
            "status": "OK",
            "tokensPerSecond": "58",
            "notes": "MAX_INPUT_TOKENS: 16384, MAX_TOTAL_TOKENS: 32768",
        },
    ],
}