results_arcee_scribe.py

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

results_arcee_scribe = {
    "name": "Arcee-Scribe",
    "modelType": "InternLM2.5 8B",
    "configurations": [
        {
            "instanceType": "g5.2xlarge",
            "quantization": "none",
            "container": "DJL 0.28 vLLM",
            "status": "OK",
            "tokensPerSecond": 29,
            "notes": '"OPTION_MAX_MODEL_LEN": "32768",\n"TENSOR_PARALLEL_DEGREE": "max",',
        },
        {
            "instanceType": "g5.12xlarge",
            "quantization": "none",
            "container": "DJL 0.28 vLLM",
            "status": "OK",
            "tokensPerSecond": 65,
            "notes": '"OPTION_MAX_MODEL_LEN": "32768",\n"TENSOR_PARALLEL_DEGREE": "max",\nNot supported by AutoAWQ and AutoGPTQ',
        },
        {
            "instanceType": "g5.48xlarge",
            "quantization": "none",
            "container": "DJL 0.28 vLLM",
            "status": "OK",
            "tokensPerSecond": 80,
            "notes": '"OPTION_MAX_MODEL_LEN": "32768",\n"TENSOR_PARALLEL_DEGREE": "max",',
        },
        {
            "instanceType": "g6.2xlarge",
            "quantization": "none",
            "container": "DJL 0.28 vLLM",
            "status": "OK",
            "tokensPerSecond": 16,
            "notes": '"OPTION_MAX_MODEL_LEN": "4096"',
        },
        {
            "instanceType": "g6.12xlarge",
            "quantization": "none",
            "container": "DJL 0.28 vLLM",
            "status": "OK",
            "tokensPerSecond": 50,
            "notes": '"OPTION_MAX_MODEL_LEN": "32768",\n"TENSOR_PARALLEL_DEGREE": "max",',
        },
        {
            "instanceType": "g6.48xlarge",
            "quantization": "none",
            "container": "DJL 0.28 vLLM",
            "status": "OK",
            "tokensPerSecond": 69,
            "notes": '"OPTION_MAX_MODEL_LEN": "32768",\n"TENSOR_PARALLEL_DEGREE": "max",',
        },
        {
            "instanceType": "g6e.2xlarge",
            "quantization": "none",
            "container": "SGLang 0.2.13",
            "status": "OK",
            "tokensPerSecond": 46,
        },
        {
            "instanceType": "p4d.24xlarge",
            "quantization": "none",
            "container": "DJL 0.28 vLLM",
            "status": "OK",
            "tokensPerSecond": 82,
            "notes": '"OPTION_MAX_MODEL_LEN": "32768",\n"TENSOR_PARALLEL_DEGREE": "max",',
        },
    ],
}