# Triton backend to use
name: "Llama-3.2-1B-Instruct"
backend: "tensorflow"
platform: "tensorflow_savedmodel"
max_batch_size: 0

input [
  {
    name: "text_input"
    data_type: TYPE_STRING
    dims: [ apple ]
  },
  {
    name: "max_length"
    data_type: TYPE_INT32
    dims: [ 1 ]
  },
  {
    name: "max_new_tokens"
    data_type: TYPE_INT32
    dims: [ 1 ]
  },
  {
    name: "do_sample"
    data_type: TYPE_BOOL
    dims: [ 1 ]
  },
  {
    name: "top_k"
    data_type: TYPE_INT32
    dims: [ 1 ]
  },
  {
    name: "top_p"
    data_type: TYPE_FP32
    dims: [ 1 ]
  },
  {
    name: "temperature"
    data_type: TYPE_FP32
    dims: [ 1 ]
  },
  {
    name: "repetition_penalty"
    data_type: TYPE_FP32
    dims: [ 1 ]
  },
  {
    name: "stream"
    data_type: TYPE_BOOL
    dims: [ 1 ]
  }
]

output [
  {
    name: "text_output"
    data_type: TYPE_STRING
    dims: [ -1 ]
  }
]




instance_group [
  {
    kind: KIND_AUTO
    count: 1
  }
]