| { |
| "model_type": "ONNX", |
| "npu_mode": "NPU3", |
| "quant": { |
| "input_configs": [ |
| { |
| "tensor_name": "DEFAULT", |
| "calibration_dataset": "./onnx-calibration-no-controlnet/transformer.tar", |
| "calibration_size": 4, |
| "calibration_format": "NumpyObject" |
| } |
| ], |
| "calibration_method": "MinMax", |
| "precision_analysis": true, |
| "precision_analysis_method": "EndToEnd", |
| "layer_configs": [ |
| { |
| "start_tensor_names": [ |
| "DEFAULT" |
| ], |
| "end_tensor_names": [ |
| "DEFAULT" |
| ], |
| "data_type": "U16" |
| } |
| ] |
| }, |
| "input_processors": [ |
| { |
| "tensor_name": "DEFAULT", |
| "tensor_format": "AutoColorSpace", |
| "tensor_layout": "NCHW" |
| } |
| ], |
| "compiler": { |
| "check": 0, |
| "sub_configs": [ |
| { |
| "start_tensor_names": ["timestep"], |
| "end_tensor_names": ["/model/t_embedder/mlp/mlp.2/Gemm_output_0"], |
| "check_mode": "CheckPerLayer" |
| }, |
| { |
| "start_tensor_names": ["prompt_embeds"], |
| "end_tensor_names": ["/model/Slice_1_output_0"], |
| "check_mode": "CheckPerLayer" |
| }, |
| { |
| "start_tensor_names": ["latent_model_input", "/model/t_embedder/mlp/mlp.2/Gemm_output_0"], |
| "end_tensor_names": ["/model/Slice_output_0"], |
| "check_mode": "CheckPerLayer" |
| }, |
| { |
| "start_tensor_names": ["/model/Slice_1_output_0", "/model/Slice_output_0", "/model/t_embedder/mlp/mlp.2/Gemm_output_0"], |
| "end_tensor_names": ["/model/layers.0/Add_4_output_0"], |
| "check_mode": "CheckPerLayer" |
| }, |
| { |
| "start_tensor_names": ["/model/layers.0/Add_4_output_0", "/model/t_embedder/mlp/mlp.2/Gemm_output_0"], |
| "end_tensor_names": ["/model/layers.1/Add_4_output_0"], |
| "check_mode": "CheckPerLayer" |
| }, |
| { |
| "start_tensor_names": ["/model/layers.1/Add_4_output_0", "/model/t_embedder/mlp/mlp.2/Gemm_output_0"], |
| "end_tensor_names": ["/model/layers.2/Add_4_output_0"], |
| "check_mode": "CheckPerLayer" |
| }, |
| { |
| "start_tensor_names": ["/model/layers.2/Add_4_output_0", "/model/t_embedder/mlp/mlp.2/Gemm_output_0"], |
| "end_tensor_names": ["/model/layers.3/Add_4_output_0"], |
| "check_mode": "CheckPerLayer" |
| }, |
| { |
| "start_tensor_names": ["/model/layers.3/Add_4_output_0", "/model/t_embedder/mlp/mlp.2/Gemm_output_0"], |
| "end_tensor_names": ["/model/layers.4/Add_4_output_0"], |
| "check_mode": "CheckPerLayer" |
| }, |
| { |
| "start_tensor_names": ["/model/layers.4/Add_4_output_0", "/model/t_embedder/mlp/mlp.2/Gemm_output_0"], |
| "end_tensor_names": ["/model/layers.5/Add_4_output_0"], |
| "check_mode": "CheckPerLayer" |
| }, |
| { |
| "start_tensor_names": ["/model/layers.5/Add_4_output_0", "/model/t_embedder/mlp/mlp.2/Gemm_output_0"], |
| "end_tensor_names": ["/model/layers.6/Add_4_output_0"], |
| "check_mode": "CheckPerLayer" |
| }, |
| { |
| "start_tensor_names": ["/model/layers.6/Add_4_output_0", "/model/t_embedder/mlp/mlp.2/Gemm_output_0"], |
| "end_tensor_names": ["/model/layers.7/Add_4_output_0"], |
| "check_mode": "CheckPerLayer" |
| }, |
| { |
| "start_tensor_names": ["/model/layers.7/Add_4_output_0", "/model/t_embedder/mlp/mlp.2/Gemm_output_0"], |
| "end_tensor_names": ["/model/layers.8/Add_4_output_0"], |
| "check_mode": "CheckPerLayer" |
| }, |
| { |
| "start_tensor_names": ["/model/layers.8/Add_4_output_0", "/model/t_embedder/mlp/mlp.2/Gemm_output_0"], |
| "end_tensor_names": ["/model/layers.9/Add_4_output_0"], |
| "check_mode": "CheckPerLayer" |
| }, |
| { |
| "start_tensor_names": ["/model/layers.9/Add_4_output_0", "/model/t_embedder/mlp/mlp.2/Gemm_output_0"], |
| "end_tensor_names": ["/model/layers.10/Add_4_output_0"], |
| "check_mode": "CheckPerLayer" |
| }, |
| { |
| "start_tensor_names": ["/model/layers.10/Add_4_output_0", "/model/t_embedder/mlp/mlp.2/Gemm_output_0"], |
| "end_tensor_names": ["/model/layers.11/Add_4_output_0"], |
| "check_mode": "CheckPerLayer" |
| }, |
| { |
| "start_tensor_names": ["/model/layers.11/Add_4_output_0", "/model/t_embedder/mlp/mlp.2/Gemm_output_0"], |
| "end_tensor_names": ["/model/layers.12/Add_4_output_0"], |
| "check_mode": "CheckPerLayer" |
| }, |
| { |
| "start_tensor_names": ["/model/layers.12/Add_4_output_0", "/model/t_embedder/mlp/mlp.2/Gemm_output_0"], |
| "end_tensor_names": ["/model/layers.13/Add_4_output_0"], |
| "check_mode": "CheckPerLayer" |
| }, |
| { |
| "start_tensor_names": ["/model/layers.13/Add_4_output_0", "/model/t_embedder/mlp/mlp.2/Gemm_output_0"], |
| "end_tensor_names": ["/model/layers.14/Add_4_output_0"], |
| "check_mode": "CheckPerLayer" |
| }, |
| { |
| "start_tensor_names": ["/model/layers.14/Add_4_output_0", "/model/t_embedder/mlp/mlp.2/Gemm_output_0"], |
| "end_tensor_names": ["/model/layers.15/Add_4_output_0"], |
| "check_mode": "CheckPerLayer" |
| }, |
| { |
| "start_tensor_names": ["/model/layers.15/Add_4_output_0", "/model/t_embedder/mlp/mlp.2/Gemm_output_0"], |
| "end_tensor_names": ["/model/layers.16/Add_4_output_0"], |
| "check_mode": "CheckPerLayer" |
| }, |
| { |
| "start_tensor_names": ["/model/layers.16/Add_4_output_0", "/model/t_embedder/mlp/mlp.2/Gemm_output_0"], |
| "end_tensor_names": ["/model/layers.17/Add_4_output_0"], |
| "check_mode": "CheckPerLayer" |
| }, |
| { |
| "start_tensor_names": ["/model/layers.17/Add_4_output_0", "/model/t_embedder/mlp/mlp.2/Gemm_output_0"], |
| "end_tensor_names": ["/model/layers.18/Add_4_output_0"], |
| "check_mode": "CheckPerLayer" |
| }, |
| { |
| "start_tensor_names": ["/model/layers.18/Add_4_output_0", "/model/t_embedder/mlp/mlp.2/Gemm_output_0"], |
| "end_tensor_names": ["/model/layers.19/Add_4_output_0"], |
| "check_mode": "CheckPerLayer" |
| }, |
| { |
| "start_tensor_names": ["/model/layers.19/Add_4_output_0", "/model/t_embedder/mlp/mlp.2/Gemm_output_0"], |
| "end_tensor_names": ["/model/layers.20/Add_4_output_0"], |
| "check_mode": "CheckPerLayer" |
| }, |
| { |
| "start_tensor_names": ["/model/layers.20/Add_4_output_0", "/model/t_embedder/mlp/mlp.2/Gemm_output_0"], |
| "end_tensor_names": ["/model/layers.21/Add_4_output_0"], |
| "check_mode": "CheckPerLayer" |
| }, |
| { |
| "start_tensor_names": ["/model/layers.21/Add_4_output_0", "/model/t_embedder/mlp/mlp.2/Gemm_output_0"], |
| "end_tensor_names": ["/model/layers.22/Add_4_output_0"], |
| "check_mode": "CheckPerLayer" |
| }, |
| { |
| "start_tensor_names": ["/model/layers.22/Add_4_output_0", "/model/t_embedder/mlp/mlp.2/Gemm_output_0"], |
| "end_tensor_names": ["/model/layers.23/Add_4_output_0"], |
| "check_mode": "CheckPerLayer" |
| }, |
| { |
| "start_tensor_names": ["/model/layers.23/Add_4_output_0", "/model/t_embedder/mlp/mlp.2/Gemm_output_0"], |
| "end_tensor_names": ["/model/layers.24/Add_4_output_0"], |
| "check_mode": "CheckPerLayer" |
| }, |
| { |
| "start_tensor_names": ["/model/layers.24/Add_4_output_0", "/model/t_embedder/mlp/mlp.2/Gemm_output_0"], |
| "end_tensor_names": ["/model/layers.25/Add_4_output_0"], |
| "check_mode": "CheckPerLayer" |
| }, |
| { |
| "start_tensor_names": ["/model/layers.25/Add_4_output_0", "/model/t_embedder/mlp/mlp.2/Gemm_output_0"], |
| "end_tensor_names": ["/model/layers.26/Add_4_output_0"], |
| "check_mode": "CheckPerLayer" |
| }, |
| { |
| "start_tensor_names": ["/model/layers.26/Add_4_output_0", "/model/t_embedder/mlp/mlp.2/Gemm_output_0"], |
| "end_tensor_names": ["/model/layers.27/Add_4_output_0"], |
| "check_mode": "CheckPerLayer" |
| }, |
| { |
| "start_tensor_names": ["/model/layers.27/Add_4_output_0", "/model/t_embedder/mlp/mlp.2/Gemm_output_0"], |
| "end_tensor_names": ["/model/layers.28/Add_4_output_0"], |
| "check_mode": "CheckPerLayer" |
| }, |
| { |
| "start_tensor_names": ["/model/layers.28/Add_4_output_0", "/model/t_embedder/mlp/mlp.2/Gemm_output_0"], |
| "end_tensor_names": ["/model/layers.29/Add_4_output_0"], |
| "check_mode": "CheckPerLayer" |
| } |
| ] |
| } |
| } |
|
|
|
|
|
|
