Update config.pbtxt

This commit is contained in:
cheetahadmin 2025-11-13 05:05:02 +00:00
parent 091f1485e6
commit 7437bf7190

@ -5,97 +5,20 @@ max_batch_size: 0
# Triton should expect as input a single string # Triton should expect as input a single string
# input of variable length named 'text_input' # input of variable length named 'text_input'
input [ input [
{ {
name: "text_input" name: "text_input"
data_type: TYPE_STRING 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 'embedding_outputt'
# output of variable length named 'text_output'
output [ output [
{ {
name: "embedding_output" name: "embedding_output"
data_type: TYPE_FP32 # 또는 TYPE_FP16 data_type: TYPE_FP32 # 또는 TYPE_FP16
dims: [ -1, -1 ] # [seq_len, hidden_size] dims: [ -1, -1 ] # [seq_len, hidden_size]
} }
] ]
@ -103,19 +26,6 @@ parameters: [
{ {
key: "base_model_path", key: "base_model_path",
value: {string_value: "/cheetah/input/model/groupuser/codesage-large-v2"} value: {string_value: "/cheetah/input/model/groupuser/codesage-large-v2"}
},
{
key: "is_adapter_model",
value: {string_value: "false"}
},
{
key: "adapter_model_path",
value: {string_value: ""}
},
{
key: "quantization",
value: {string_value: "none"}
} }
] ]
@ -125,4 +35,3 @@ instance_group [
count: 1 count: 1
} }
] ]