| name: "paraphrase-multilingual-MiniLM-L12-v2_model_onnx" | |
| max_batch_size: 0 | |
| platform: "onnxruntime_onnx" | |
| default_model_filename: "model.bin" | |
| input [ | |
| { | |
| name: "input_ids" | |
| data_type: TYPE_INT32 | |
| dims: [-1, -1] | |
| }, | |
| { | |
| name: "token_type_ids" | |
| data_type: TYPE_INT32 | |
| dims: [-1, -1] | |
| }, | |
| { | |
| name: "attention_mask" | |
| data_type: TYPE_INT32 | |
| dims: [-1, -1] | |
| } | |
| ] | |
| output { | |
| name: "output" | |
| data_type: TYPE_FP32 | |
| dims: [-1, 384] | |
| } | |
| instance_group [ | |
| { | |
| count: 1 | |
| kind: KIND_GPU | |
| } | |
| ] |