inference-cache-config/llama3.json

{
   "meta-llama/Meta-Llama-3-8B": [
    {
      "batch_size": 1,
      "sequence_length": 4096,
      "num_cores": 2,
      "auto_cast_type": "fp16"
    },
    {
      "batch_size": 4,
      "sequence_length": 4096,
      "num_cores": 2,
      "auto_cast_type": "fp16"
    },
    {
      "batch_size": 8,
      "sequence_length": 4096,
      "num_cores": 2,
      "auto_cast_type": "fp16"
    },
    {
      "batch_size": 4,
      "sequence_length": 4096,
      "num_cores": 8,
      "auto_cast_type": "fp16"
    },
    {
      "batch_size": 8,
      "sequence_length": 4096,
      "num_cores": 8,
      "auto_cast_type": "fp16"
    },
    {
      "batch_size": 16,
      "sequence_length": 4096,
      "num_cores": 8,
      "auto_cast_type": "fp16"
    },
    {
      "batch_size": 32,
      "sequence_length": 4096,
      "num_cores": 8,
      "auto_cast_type": "fp16"
    }
  ],
  "meta-llama/Meta-Llama-3-70B": [
    {
      "batch_size": 1,
      "sequence_length": 4096,
      "num_cores": 24,
      "auto_cast_type": "fp16"
    },
    {
      "batch_size": 4,
      "sequence_length": 4096,
      "num_cores": 24,
      "auto_cast_type": "fp16"
    }
  ]
}