sdnext/configs/olive/sd/text_encoder.json

{
  "input_model": {
    "type": "PyTorchModel",
    "config": {
      "model_path": "",
      "model_loader": "text_encoder_load",
      "model_script": "modules/olive_script.py",
      "io_config": {
        "input_names": ["input_ids"],
        "output_names": ["last_hidden_state", "pooler_output"],
        "dynamic_axes": { "input_ids": { "0": "batch", "1": "sequence" } }
      },
      "dummy_inputs_func": "text_encoder_conversion_inputs"
    }
  },
  "systems": {
    "local_system": {
      "type": "LocalSystem",
      "config": {
        "accelerators": [
          {
            "device": "gpu",
            "execution_providers": ["DmlExecutionProvider"]
          }
        ]
      }
    }
  },
  "evaluators": {
    "common_evaluator": {
      "metrics": [
        {
          "name": "latency",
          "type": "latency",
          "sub_types": [{ "name": "avg" }],
          "user_config": {
            "user_script": "modules/olive_script.py",
            "dataloader_func": "text_encoder_data_loader",
            "batch_size": 1
          }
        }
      ]
    }
  },
  "passes": {
    "optimize_CPUExecutionProvider": {
      "type": "OrtTransformersOptimization",
      "config": {
        "model_type": "clip",
        "opt_level": 0,
        "float16": false,
        "use_gpu": false,
        "keep_io_types": false
      }
    },
    "optimize_DmlExecutionProvider": {
      "type": "OrtTransformersOptimization",
      "config": {
        "model_type": "clip",
        "opt_level": 0,
        "float16": true,
        "use_gpu": true,
        "keep_io_types": false,
        "optimization_options": {
          "enable_gelu": true,
          "enable_layer_norm": true,
          "enable_attention": true,
          "use_multi_head_attention": true,
          "enable_skip_layer_norm": false,
          "enable_embed_layer_norm": true,
          "enable_bias_skip_layer_norm": false,
          "enable_bias_gelu": true,
          "enable_gelu_approximation": false,
          "enable_qordered_matmul": false,
          "enable_shape_inference": true,
          "enable_gemm_fast_gelu": false,
          "enable_nhwc_conv": false,
          "enable_group_norm": true,
          "enable_bias_splitgelu": false,
          "enable_packed_qkv": true,
          "enable_packed_kv": true,
          "enable_bias_add": false,
          "group_norm_channels_last": false
        },
        "force_fp32_ops": ["RandomNormalLike"],
        "force_fp16_inputs": {
          "GroupNorm": [0, 1, 2]
        }
      }
    },
    "optimize_CUDAExecutionProvider": {
      "type": "OrtTransformersOptimization",
      "config": {
        "model_type": "clip",
        "opt_level": 0,
        "float16": true,
        "use_gpu": true,
        "keep_io_types": false
      }
    },
    "optimize_ROCMExecutionProvider": {
      "type": "OrtTransformersOptimization",
      "config": {
        "model_type": "clip",
        "opt_level": 0,
        "float16": true,
        "use_gpu": true,
        "keep_io_types": false
      }
    },
    "quantization": {
      "type": "OnnxDynamicQuantization",
      "disable_search": true,
      "config": {
        "save_as_external_data": false,
        "all_tensors_to_one_file": true,
        "per_channel": false,
        "reduce_range": false,
        "MatMulConstBOnly": true
      }
    }
  },
  "pass_flows": [["optimize_AutoExecutionProvider"]],
  "engine": {
    "log_severity_level": 0,
    "evaluator": "common_evaluator",
    "evaluate_input_model": false,
    "host": "local_system",
    "target": "local_system",
    "cache_dir": "cache",
    "output_name": "text_encoder",
    "output_dir": "footprints"
  }
}