inference-cache-config/mistral-variants.json

{
    "HuggingFaceH4/zephyr-7b-beta": [
    {
      "batch_size": 1,
      "sequence_length": 4096,
      "num_cores": 2,
      "auto_cast_type": "bf16"
    },
    {
      "batch_size": 1,
      "sequence_length": 4096,
      "num_cores": 8,
      "auto_cast_type": "bf16"
    },
    {
      "batch_size": 4,
      "sequence_length": 4096,
      "num_cores": 2,
      "auto_cast_type": "bf16"
    },
    {
      "batch_size": 4,
      "sequence_length": 4096,
      "num_cores": 8,
      "auto_cast_type": "bf16"
    },
    {
      "batch_size": 8,
      "sequence_length": 4096,
      "num_cores": 8,
      "auto_cast_type": "bf16"
    },
    {
      "batch_size": 16,
      "sequence_length": 4096,
      "num_cores": 8,
      "auto_cast_type": "bf16"
    }
  ],
  "Intel/neural-chat-7b-v3-3": [
    {
      "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": "bf16"
    }
  ]
}