diff --git a/config.pbtxt b/config.pbtxt index ef086eb..0fb9564 100644 --- a/config.pbtxt +++ b/config.pbtxt @@ -1,75 +1,105 @@ -# Triton backend to use -name: "stablelm-2-1_6b" +# Triton Backend for TransformerLLM. backend: "python" max_batch_size: 0 # Triton should expect as input a single string # input of variable length named 'text_input' input [ + { name: "text_input" data_type: TYPE_STRING - dims: [ -1 ] + dims: [ 1 ] + + }, { name: "max_length" data_type: TYPE_INT32 dims: [ 1 ] + optional: true + + }, { name: "max_new_tokens" data_type: TYPE_INT32 dims: [ 1 ] + optional: true + + }, { name: "do_sample" data_type: TYPE_BOOL dims: [ 1 ] + optional: true + + }, { name: "top_k" data_type: TYPE_INT32 dims: [ 1 ] + optional: true + + }, { name: "top_p" data_type: TYPE_FP32 dims: [ 1 ] + optional: true + + }, { name: "temperature" data_type: TYPE_FP32 dims: [ 1 ] + optional: true + + }, { name: "repetition_penalty" data_type: TYPE_FP32 dims: [ 1 ] + optional: true + + }, { name: "stream" data_type: TYPE_BOOL dims: [ 1 ] + optional: true + + } + ] # Triton should expect to respond with a single string # output of variable length named 'text_output' output [ + { name: "text_output" data_type: TYPE_STRING - dims: [ -1 ] + dims: [ 1 ] + } + ] parameters: [ @@ -78,15 +108,24 @@ parameters: [ value: {string_value: "/cheetah/input/model/groupuser/stablelm-2-1_6b"} }, { - key: "enable_inference_trace", - value: {string_value: "True"} + key: "is_adapter_model", + value: {string_value: "false"} + }, + { + key: "adapter_model_path", + value: {string_value: ""} + }, + + { + key: "quantization", + value: {string_value: "none"} } ] instance_group [ { - kind: KIND_GPU, + kind: KIND_AUTO, count: 1 } ]