| { |
| "created_utc": "2026-09-19T19:43:44.834712+00:00", |
| "purpose": "Bounded real-input latency and scheduler placement; not a general accuracy benchmark", |
| "model_revision": "f54adbf447f4ca6ec259f529ee3f2e3e09f8cc71", |
| "package_files": { |
| "Data/com.apple.CoreML/model.mlmodel": "70485fc18cbb21785df833cbddddc0b5b59acb00d22394b76e55307e2c135dd0", |
| "Data/com.apple.CoreML/weights/weight.bin": "4da9259f798e44f5a1b50769ee1916fd3747c4d723dd9997b516c7fe238c7895", |
| "Manifest.json": "2bc0f5f62337b27fb6b0ecde248f1e3dc269e1ba4b65516aaeede2a60e293dcc" |
| }, |
| "dataset_revision": "8273f34778b99ac2e12d9f6e7d57dad99ae20845", |
| "dataset_sha256": "4f43b442e79ba2e2ce731e27e9b8e340c2b5dfcaffc92d8ff564c34f115ff1ca", |
| "environment": { |
| "chip": "Apple M5 Pro", |
| "memory_bytes": 25769803776, |
| "macos": "27.0", |
| "os_build": "26A428", |
| "python": "3.11.11", |
| "coremltools": "9.0" |
| }, |
| "protocol": { |
| "rows": [ |
| 0, |
| 68, |
| 130 |
| ], |
| "pages": [ |
| "patient-registration", |
| "job-application", |
| "auto-claim" |
| ], |
| "option_counts": [ |
| 27, |
| 21, |
| 19 |
| ], |
| "first_call_row": 0, |
| "warmup_passes": 2, |
| "timed_passes": 10, |
| "backend_order": [ |
| "CPU_ONLY", |
| "CPU_AND_GPU", |
| "CPU_AND_NE", |
| "ALL" |
| ], |
| "timing_scope": "Synchronous Python MLModel.predict; pre-encoded inputs; output checks outside timer", |
| "load_scope": "MLModel construction; system caches retained, not first-install cold start", |
| "first_call_scope": "First predict after this load; may use existing system/ANE caches", |
| "plan_order": "Compute plan loaded after timing each backend" |
| }, |
| "backends": { |
| "CPU_ONLY": { |
| "load_ms": 71.21945900144055, |
| "first_prediction_ms": 15.096832998096943, |
| "warm_prediction": { |
| "count": 30, |
| "median_ms": 1.5270625008270144, |
| "p95_ms": 1.601899994420819, |
| "min_ms": 1.3285410241223872, |
| "max_ms": 1.6374590049963444 |
| }, |
| "timed_predictions_correct": 30, |
| "calls": [ |
| { |
| "row": 0, |
| "milliseconds": 1.578499999595806 |
| }, |
| { |
| "row": 68, |
| "milliseconds": 1.561041979584843 |
| }, |
| { |
| "row": 130, |
| "milliseconds": 1.5606670058332384 |
| }, |
| { |
| "row": 0, |
| "milliseconds": 1.4908749726600945 |
| }, |
| { |
| "row": 68, |
| "milliseconds": 1.5069579822011292 |
| }, |
| { |
| "row": 130, |
| "milliseconds": 1.412415993399918 |
| }, |
| { |
| "row": 0, |
| "milliseconds": 1.5242919907905161 |
| }, |
| { |
| "row": 68, |
| "milliseconds": 1.5393749927170575 |
| }, |
| { |
| "row": 130, |
| "milliseconds": 1.3988339924253523 |
| }, |
| { |
| "row": 0, |
| "milliseconds": 1.5052919916342944 |
| }, |
| { |
| "row": 68, |
| "milliseconds": 1.4890410238876939 |
| }, |
| { |
| "row": 130, |
| "milliseconds": 1.5105829806998372 |
| }, |
| { |
| "row": 0, |
| "milliseconds": 1.6374590049963444 |
| }, |
| { |
| "row": 68, |
| "milliseconds": 1.6122499946504831 |
| }, |
| { |
| "row": 130, |
| "milliseconds": 1.5023329760879278 |
| }, |
| { |
| "row": 0, |
| "milliseconds": 1.5473749954253435 |
| }, |
| { |
| "row": 68, |
| "milliseconds": 1.3900829944759607 |
| }, |
| { |
| "row": 130, |
| "milliseconds": 1.5310419839806855 |
| }, |
| { |
| "row": 0, |
| "milliseconds": 1.3285410241223872 |
| }, |
| { |
| "row": 68, |
| "milliseconds": 1.5892499941401184 |
| }, |
| { |
| "row": 130, |
| "milliseconds": 1.5656249888706952 |
| }, |
| { |
| "row": 0, |
| "milliseconds": 1.546334009617567 |
| }, |
| { |
| "row": 68, |
| "milliseconds": 1.5298330108635128 |
| }, |
| { |
| "row": 130, |
| "milliseconds": 1.543457998195663 |
| }, |
| { |
| "row": 0, |
| "milliseconds": 1.5623749932274222 |
| }, |
| { |
| "row": 68, |
| "milliseconds": 1.4965840091463178 |
| }, |
| { |
| "row": 130, |
| "milliseconds": 1.5304579865187407 |
| }, |
| { |
| "row": 0, |
| "milliseconds": 1.465249981265515 |
| }, |
| { |
| "row": 68, |
| "milliseconds": 1.4851669839117676 |
| }, |
| { |
| "row": 130, |
| "milliseconds": 1.4999579871073365 |
| } |
| ], |
| "compute_plan": { |
| "kind": "MLComputePlan preferred device; not a runtime utilization or power trace", |
| "assigned_operation_count": 173, |
| "device_counts": { |
| "cpu": 173, |
| "gpu": 0, |
| "ane": 0, |
| "unknown": 0 |
| }, |
| "unassigned_operation_types": { |
| "const": 239 |
| }, |
| "operations": [ |
| { |
| "name": "option_mask_1", |
| "type": "ios17.not_equal", |
| "device": "cpu" |
| }, |
| { |
| "name": "context_mask", |
| "type": "ios17.not_equal", |
| "device": "cpu" |
| }, |
| { |
| "name": "var_27", |
| "type": "ios17.slice_by_index", |
| "device": "cpu" |
| }, |
| { |
| "name": "fill_like_0_cast_fp16", |
| "type": "ios16.fill_like", |
| "device": "cpu" |
| }, |
| { |
| "name": "var_43", |
| "type": "ios17.slice_by_index", |
| "device": "cpu" |
| }, |
| { |
| "name": "var_43_to_fp16", |
| "type": "ios17.cast", |
| "device": "cpu" |
| }, |
| { |
| "name": "safe_context_ids_cast_fp16", |
| "type": "ios17.concat", |
| "device": "cpu" |
| }, |
| { |
| "name": "context_ids_to_int16", |
| "type": "ios17.cast", |
| "device": "cpu" |
| }, |
| { |
| "name": "cast_54", |
| "type": "ios17.cast", |
| "device": "cpu" |
| }, |
| { |
| "name": "greater_equal_0", |
| "type": "ios17.greater_equal", |
| "device": "cpu" |
| }, |
| { |
| "name": "add_3", |
| "type": "ios17.add", |
| "device": "cpu" |
| }, |
| { |
| "name": "select_0", |
| "type": "select", |
| "device": "cpu" |
| }, |
| { |
| "name": "select_0_to_int16", |
| "type": "ios17.cast", |
| "device": "cpu" |
| }, |
| { |
| "name": "var_59_cast_fp16_cast_uint16_cast_uint16", |
| "type": "ios17.gather", |
| "device": "cpu" |
| }, |
| { |
| "name": "src_1_cast_fp16", |
| "type": "ios17.add", |
| "device": "cpu" |
| }, |
| { |
| "name": "mask_1_cast_fp16", |
| "type": "ios17.equal", |
| "device": "cpu" |
| }, |
| { |
| "name": "key_padding_mask_1_cast_fp16", |
| "type": "select", |
| "device": "cpu" |
| }, |
| { |
| "name": "query_1_cast_fp16", |
| "type": "ios17.layer_norm", |
| "device": "cpu" |
| }, |
| { |
| "name": "query_3_cast_fp16", |
| "type": "ios17.transpose", |
| "device": "cpu" |
| }, |
| { |
| "name": "linear_0_cast_fp16", |
| "type": "ios17.linear", |
| "device": "cpu" |
| }, |
| { |
| "name": "var_124_cast_fp16", |
| "type": "ios17.reshape", |
| "device": "cpu" |
| }, |
| { |
| "name": "var_125_cast_fp16", |
| "type": "ios17.expand_dims", |
| "device": "cpu" |
| }, |
| { |
| "name": "var_126_cast_fp16", |
| "type": "ios17.transpose", |
| "device": "cpu" |
| }, |
| { |
| "name": "var_127_cast_fp16", |
| "type": "ios17.squeeze", |
| "device": "cpu" |
| }, |
| { |
| "name": "q_1_cast_fp16", |
| "type": "ios17.slice_by_index", |
| "device": "cpu" |
| }, |
| { |
| "name": "k_1_cast_fp16", |
| "type": "ios17.slice_by_index", |
| "device": "cpu" |
| }, |
| { |
| "name": "v_1_cast_fp16", |
| "type": "ios17.slice_by_index", |
| "device": "cpu" |
| }, |
| { |
| "name": "var_136_cast_fp16", |
| "type": "ios17.reshape", |
| "device": "cpu" |
| }, |
| { |
| "name": "var_143_cast_fp16", |
| "type": "ios17.reshape", |
| "device": "cpu" |
| }, |
| { |
| "name": "var_150_cast_fp16", |
| "type": "ios17.reshape", |
| "device": "cpu" |
| }, |
| { |
| "name": "var_154_cast_fp16", |
| "type": "ios17.reshape", |
| "device": "cpu" |
| }, |
| { |
| "name": "var_156_cast_fp16", |
| "type": "tile", |
| "device": "cpu" |
| }, |
| { |
| "name": "q_3_cast_fp16", |
| "type": "ios17.transpose", |
| "device": "cpu" |
| }, |
| { |
| "name": "q_5_cast_fp16", |
| "type": "ios17.reshape", |
| "device": "cpu" |
| }, |
| { |
| "name": "k_3_cast_fp16", |
| "type": "ios17.transpose", |
| "device": "cpu" |
| }, |
| { |
| "name": "k_5_cast_fp16", |
| "type": "ios17.reshape", |
| "device": "cpu" |
| }, |
| { |
| "name": "v_3_cast_fp16", |
| "type": "ios17.transpose", |
| "device": "cpu" |
| }, |
| { |
| "name": "v_5_cast_fp16", |
| "type": "ios17.reshape", |
| "device": "cpu" |
| }, |
| { |
| "name": "mul_1_cast_fp16", |
| "type": "ios17.mul", |
| "device": "cpu" |
| }, |
| { |
| "name": "matmul_0_cast_fp16", |
| "type": "ios17.matmul", |
| "device": "cpu" |
| }, |
| { |
| "name": "add_0_cast_fp16", |
| "type": "ios17.add", |
| "device": "cpu" |
| }, |
| { |
| "name": "softmax_0_cast_fp16", |
| "type": "ios16.softmax", |
| "device": "cpu" |
| }, |
| { |
| "name": "attn_output_1_cast_fp16", |
| "type": "ios17.matmul", |
| "device": "cpu" |
| }, |
| { |
| "name": "var_172_cast_fp16", |
| "type": "ios17.transpose", |
| "device": "cpu" |
| }, |
| { |
| "name": "attn_output_3_cast_fp16", |
| "type": "ios17.reshape", |
| "device": "cpu" |
| }, |
| { |
| "name": "linear_1_cast_fp16", |
| "type": "ios17.linear", |
| "device": "cpu" |
| }, |
| { |
| "name": "attn_output_7_cast_fp16", |
| "type": "ios17.reshape", |
| "device": "cpu" |
| }, |
| { |
| "name": "input_3_cast_fp16", |
| "type": "ios17.transpose", |
| "device": "cpu" |
| }, |
| { |
| "name": "input_5_cast_fp16", |
| "type": "ios17.add", |
| "device": "cpu" |
| }, |
| { |
| "name": "input_7_cast_fp16", |
| "type": "ios17.layer_norm", |
| "device": "cpu" |
| }, |
| { |
| "name": "linear_2_cast_fp16", |
| "type": "ios17.linear", |
| "device": "cpu" |
| }, |
| { |
| "name": "input_11_cast_fp16", |
| "type": "ios16.relu", |
| "device": "cpu" |
| }, |
| { |
| "name": "linear_3_cast_fp16", |
| "type": "ios17.linear", |
| "device": "cpu" |
| }, |
| { |
| "name": "input_17_cast_fp16", |
| "type": "ios17.add", |
| "device": "cpu" |
| }, |
| { |
| "name": "query_5_cast_fp16", |
| "type": "ios17.layer_norm", |
| "device": "cpu" |
| }, |
| { |
| "name": "query_7_cast_fp16", |
| "type": "ios17.transpose", |
| "device": "cpu" |
| }, |
| { |
| "name": "linear_4_cast_fp16", |
| "type": "ios17.linear", |
| "device": "cpu" |
| }, |
| { |
| "name": "var_231_cast_fp16", |
| "type": "ios17.reshape", |
| "device": "cpu" |
| }, |
| { |
| "name": "var_232_cast_fp16", |
| "type": "ios17.expand_dims", |
| "device": "cpu" |
| }, |
| { |
| "name": "var_233_cast_fp16", |
| "type": "ios17.transpose", |
| "device": "cpu" |
| }, |
| { |
| "name": "var_234_cast_fp16", |
| "type": "ios17.squeeze", |
| "device": "cpu" |
| }, |
| { |
| "name": "q_7_cast_fp16", |
| "type": "ios17.slice_by_index", |
| "device": "cpu" |
| }, |
| { |
| "name": "k_7_cast_fp16", |
| "type": "ios17.slice_by_index", |
| "device": "cpu" |
| }, |
| { |
| "name": "v_7_cast_fp16", |
| "type": "ios17.slice_by_index", |
| "device": "cpu" |
| }, |
| { |
| "name": "var_243_cast_fp16", |
| "type": "ios17.reshape", |
| "device": "cpu" |
| }, |
| { |
| "name": "var_250_cast_fp16", |
| "type": "ios17.reshape", |
| "device": "cpu" |
| }, |
| { |
| "name": "var_257_cast_fp16", |
| "type": "ios17.reshape", |
| "device": "cpu" |
| }, |
| { |
| "name": "q_9_cast_fp16", |
| "type": "ios17.transpose", |
| "device": "cpu" |
| }, |
| { |
| "name": "q_11_cast_fp16", |
| "type": "ios17.reshape", |
| "device": "cpu" |
| }, |
| { |
| "name": "k_9_cast_fp16", |
| "type": "ios17.transpose", |
| "device": "cpu" |
| }, |
| { |
| "name": "k_11_cast_fp16", |
| "type": "ios17.reshape", |
| "device": "cpu" |
| }, |
| { |
| "name": "v_9_cast_fp16", |
| "type": "ios17.transpose", |
| "device": "cpu" |
| }, |
| { |
| "name": "v_11_cast_fp16", |
| "type": "ios17.reshape", |
| "device": "cpu" |
| }, |
| { |
| "name": "mul_3_cast_fp16", |
| "type": "ios17.mul", |
| "device": "cpu" |
| }, |
| { |
| "name": "matmul_1_cast_fp16", |
| "type": "ios17.matmul", |
| "device": "cpu" |
| }, |
| { |
| "name": "add_1_cast_fp16", |
| "type": "ios17.add", |
| "device": "cpu" |
| }, |
| { |
| "name": "softmax_1_cast_fp16", |
| "type": "ios16.softmax", |
| "device": "cpu" |
| }, |
| { |
| "name": "attn_output_9_cast_fp16", |
| "type": "ios17.matmul", |
| "device": "cpu" |
| }, |
| { |
| "name": "var_279_cast_fp16", |
| "type": "ios17.transpose", |
| "device": "cpu" |
| }, |
| { |
| "name": "attn_output_11_cast_fp16", |
| "type": "ios17.reshape", |
| "device": "cpu" |
| }, |
| { |
| "name": "linear_5_cast_fp16", |
| "type": "ios17.linear", |
| "device": "cpu" |
| }, |
| { |
| "name": "attn_output_15_cast_fp16", |
| "type": "ios17.reshape", |
| "device": "cpu" |
| }, |
| { |
| "name": "input_19_cast_fp16", |
| "type": "ios17.transpose", |
| "device": "cpu" |
| }, |
| { |
| "name": "input_21_cast_fp16", |
| "type": "ios17.add", |
| "device": "cpu" |
| }, |
| { |
| "name": "input_23_cast_fp16", |
| "type": "ios17.layer_norm", |
| "device": "cpu" |
| }, |
| { |
| "name": "linear_6_cast_fp16", |
| "type": "ios17.linear", |
| "device": "cpu" |
| }, |
| { |
| "name": "input_27_cast_fp16", |
| "type": "ios16.relu", |
| "device": "cpu" |
| }, |
| { |
| "name": "linear_7_cast_fp16", |
| "type": "ios17.linear", |
| "device": "cpu" |
| }, |
| { |
| "name": "context_cast_fp16", |
| "type": "ios17.add", |
| "device": "cpu" |
| }, |
| { |
| "name": "flat_ids", |
| "type": "ios17.reshape", |
| "device": "cpu" |
| }, |
| { |
| "name": "flat_mask", |
| "type": "ios17.not_equal", |
| "device": "cpu" |
| }, |
| { |
| "name": "var_333", |
| "type": "ios17.slice_by_index", |
| "device": "cpu" |
| }, |
| { |
| "name": "fill_like_1_cast_fp16", |
| "type": "ios16.fill_like", |
| "device": "cpu" |
| }, |
| { |
| "name": "var_349", |
| "type": "ios17.slice_by_index", |
| "device": "cpu" |
| }, |
| { |
| "name": "var_349_to_fp16", |
| "type": "ios17.cast", |
| "device": "cpu" |
| }, |
| { |
| "name": "safe_ids_cast_fp16", |
| "type": "ios17.concat", |
| "device": "cpu" |
| }, |
| { |
| "name": "flat_ids_to_int16", |
| "type": "ios17.cast", |
| "device": "cpu" |
| }, |
| { |
| "name": "cast_55", |
| "type": "ios17.cast", |
| "device": "cpu" |
| }, |
| { |
| "name": "greater_equal_1", |
| "type": "ios17.greater_equal", |
| "device": "cpu" |
| }, |
| { |
| "name": "add_4", |
| "type": "ios17.add", |
| "device": "cpu" |
| }, |
| { |
| "name": "select_1", |
| "type": "select", |
| "device": "cpu" |
| }, |
| { |
| "name": "select_1_to_int16", |
| "type": "ios17.cast", |
| "device": "cpu" |
| }, |
| { |
| "name": "var_365_cast_fp16_cast_uint16_cast_uint16", |
| "type": "ios17.gather", |
| "device": "cpu" |
| }, |
| { |
| "name": "src_cast_fp16", |
| "type": "ios17.add", |
| "device": "cpu" |
| }, |
| { |
| "name": "mask_cast_fp16", |
| "type": "ios17.equal", |
| "device": "cpu" |
| }, |
| { |
| "name": "key_padding_mask_7_cast_fp16", |
| "type": "select", |
| "device": "cpu" |
| }, |
| { |
| "name": "query_9_cast_fp16", |
| "type": "ios17.layer_norm", |
| "device": "cpu" |
| }, |
| { |
| "name": "query_11_cast_fp16", |
| "type": "ios17.transpose", |
| "device": "cpu" |
| }, |
| { |
| "name": "linear_8_cast_fp16", |
| "type": "ios17.linear", |
| "device": "cpu" |
| }, |
| { |
| "name": "var_428_cast_fp16", |
| "type": "ios17.reshape", |
| "device": "cpu" |
| }, |
| { |
| "name": "var_429_cast_fp16", |
| "type": "ios17.expand_dims", |
| "device": "cpu" |
| }, |
| { |
| "name": "var_430_cast_fp16", |
| "type": "ios17.transpose", |
| "device": "cpu" |
| }, |
| { |
| "name": "var_431_cast_fp16", |
| "type": "ios17.squeeze", |
| "device": "cpu" |
| }, |
| { |
| "name": "q_13_cast_fp16", |
| "type": "ios17.slice_by_index", |
| "device": "cpu" |
| }, |
| { |
| "name": "k_13_cast_fp16", |
| "type": "ios17.slice_by_index", |
| "device": "cpu" |
| }, |
| { |
| "name": "v_13_cast_fp16", |
| "type": "ios17.slice_by_index", |
| "device": "cpu" |
| }, |
| { |
| "name": "var_440_cast_fp16", |
| "type": "ios17.reshape", |
| "device": "cpu" |
| }, |
| { |
| "name": "var_447_cast_fp16", |
| "type": "ios17.reshape", |
| "device": "cpu" |
| }, |
| { |
| "name": "var_454_cast_fp16", |
| "type": "ios17.reshape", |
| "device": "cpu" |
| }, |
| { |
| "name": "var_458_cast_fp16", |
| "type": "ios17.reshape", |
| "device": "cpu" |
| }, |
| { |
| "name": "var_460_cast_fp16", |
| "type": "tile", |
| "device": "cpu" |
| }, |
| { |
| "name": "q_15_cast_fp16", |
| "type": "ios17.transpose", |
| "device": "cpu" |
| }, |
| { |
| "name": "q_cast_fp16", |
| "type": "ios17.reshape", |
| "device": "cpu" |
| }, |
| { |
| "name": "k_15_cast_fp16", |
| "type": "ios17.transpose", |
| "device": "cpu" |
| }, |
| { |
| "name": "k_cast_fp16", |
| "type": "ios17.reshape", |
| "device": "cpu" |
| }, |
| { |
| "name": "v_15_cast_fp16", |
| "type": "ios17.transpose", |
| "device": "cpu" |
| }, |
| { |
| "name": "v_cast_fp16", |
| "type": "ios17.reshape", |
| "device": "cpu" |
| }, |
| { |
| "name": "mul_5_cast_fp16", |
| "type": "ios17.mul", |
| "device": "cpu" |
| }, |
| { |
| "name": "matmul_2_cast_fp16", |
| "type": "ios17.matmul", |
| "device": "cpu" |
| }, |
| { |
| "name": "add_2_cast_fp16", |
| "type": "ios17.add", |
| "device": "cpu" |
| }, |
| { |
| "name": "softmax_2_cast_fp16", |
| "type": "ios16.softmax", |
| "device": "cpu" |
| }, |
| { |
| "name": "attn_output_17_cast_fp16", |
| "type": "ios17.matmul", |
| "device": "cpu" |
| }, |
| { |
| "name": "var_476_cast_fp16", |
| "type": "ios17.transpose", |
| "device": "cpu" |
| }, |
| { |
| "name": "attn_output_19_cast_fp16", |
| "type": "ios17.reshape", |
| "device": "cpu" |
| }, |
| { |
| "name": "linear_9_cast_fp16", |
| "type": "ios17.linear", |
| "device": "cpu" |
| }, |
| { |
| "name": "attn_output_cast_fp16", |
| "type": "ios17.reshape", |
| "device": "cpu" |
| }, |
| { |
| "name": "input_35_cast_fp16", |
| "type": "ios17.transpose", |
| "device": "cpu" |
| }, |
| { |
| "name": "input_37_cast_fp16", |
| "type": "ios17.add", |
| "device": "cpu" |
| }, |
| { |
| "name": "input_39_cast_fp16", |
| "type": "ios17.layer_norm", |
| "device": "cpu" |
| }, |
| { |
| "name": "linear_10_cast_fp16", |
| "type": "ios17.linear", |
| "device": "cpu" |
| }, |
| { |
| "name": "input_43_cast_fp16", |
| "type": "ios16.relu", |
| "device": "cpu" |
| }, |
| { |
| "name": "linear_11_cast_fp16", |
| "type": "ios17.linear", |
| "device": "cpu" |
| }, |
| { |
| "name": "hidden_cast_fp16", |
| "type": "ios17.add", |
| "device": "cpu" |
| }, |
| { |
| "name": "var_504", |
| "type": "ios17.expand_dims", |
| "device": "cpu" |
| }, |
| { |
| "name": "var_504_to_fp16", |
| "type": "ios17.cast", |
| "device": "cpu" |
| }, |
| { |
| "name": "var_510_cast_fp16", |
| "type": "ios17.mul", |
| "device": "cpu" |
| }, |
| { |
| "name": "var_515_cast_fp16", |
| "type": "ios16.reduce_sum", |
| "device": "cpu" |
| }, |
| { |
| "name": "var_520_cast_fp16", |
| "type": "ios16.reduce_sum", |
| "device": "cpu" |
| }, |
| { |
| "name": "var_522_cast_fp16", |
| "type": "ios17.maximum", |
| "device": "cpu" |
| }, |
| { |
| "name": "pooled_cast_fp16", |
| "type": "ios17.real_div", |
| "device": "cpu" |
| }, |
| { |
| "name": "options_cast_fp16", |
| "type": "ios17.reshape", |
| "device": "cpu" |
| }, |
| { |
| "name": "input_cast_fp16", |
| "type": "ios17.layer_norm", |
| "device": "cpu" |
| }, |
| { |
| "name": "input_53_cast_fp16", |
| "type": "ios17.layer_norm", |
| "device": "cpu" |
| }, |
| { |
| "name": "linear_12_cast_fp16", |
| "type": "ios17.linear", |
| "device": "cpu" |
| }, |
| { |
| "name": "linear_13_cast_fp16", |
| "type": "ios17.linear", |
| "device": "cpu" |
| }, |
| { |
| "name": "linear_14_cast_fp16", |
| "type": "ios17.linear", |
| "device": "cpu" |
| }, |
| { |
| "name": "matmul_3_cast_fp16", |
| "type": "ios17.matmul", |
| "device": "cpu" |
| }, |
| { |
| "name": "_inversed_scores_1_cast_fp16", |
| "type": "ios17.mul", |
| "device": "cpu" |
| }, |
| { |
| "name": "var_567", |
| "type": "ios17.expand_dims", |
| "device": "cpu" |
| }, |
| { |
| "name": "var_569", |
| "type": "ios17.logical_not", |
| "device": "cpu" |
| }, |
| { |
| "name": "scores_cast_fp16", |
| "type": "select", |
| "device": "cpu" |
| }, |
| { |
| "name": "var_571_cast_fp16", |
| "type": "ios16.softmax", |
| "device": "cpu" |
| }, |
| { |
| "name": "matmul_4_cast_fp16", |
| "type": "ios17.matmul", |
| "device": "cpu" |
| }, |
| { |
| "name": "var_574_cast_fp16", |
| "type": "ios17.mul", |
| "device": "cpu" |
| }, |
| { |
| "name": "var_576_cast_fp16", |
| "type": "ios16.reduce_sum", |
| "device": "cpu" |
| }, |
| { |
| "name": "_inversed_logits_1_cast_fp16", |
| "type": "ios17.mul", |
| "device": "cpu" |
| }, |
| { |
| "name": "var_579", |
| "type": "ios17.logical_not", |
| "device": "cpu" |
| }, |
| { |
| "name": "logits_3_cast_fp16", |
| "type": "select", |
| "device": "cpu" |
| }, |
| { |
| "name": "var_587_cast_fp16", |
| "type": "ios16.fill_like", |
| "device": "cpu" |
| }, |
| { |
| "name": "logits_cast_fp16", |
| "type": "select", |
| "device": "cpu" |
| }, |
| { |
| "name": "var_591_cast_fp16", |
| "type": "ios16.softmax", |
| "device": "cpu" |
| }, |
| { |
| "name": "probabilities", |
| "type": "ios17.cast", |
| "device": "cpu" |
| }, |
| { |
| "name": "logits", |
| "type": "ios17.cast", |
| "device": "cpu" |
| } |
| ] |
| } |
| }, |
| "CPU_AND_GPU": { |
| "load_ms": 70.07474999409169, |
| "first_prediction_ms": 66.33504197816364, |
| "warm_prediction": { |
| "count": 30, |
| "median_ms": 0.9288130095228553, |
| "p95_ms": 2.3798043403076, |
| "min_ms": 0.8875419734977186, |
| "max_ms": 2.6384999800939113 |
| }, |
| "timed_predictions_correct": 30, |
| "calls": [ |
| { |
| "row": 0, |
| "milliseconds": 2.5134169845841825 |
| }, |
| { |
| "row": 68, |
| "milliseconds": 2.6384999800939113 |
| }, |
| { |
| "row": 130, |
| "milliseconds": 1.9895419827662408 |
| }, |
| { |
| "row": 0, |
| "milliseconds": 2.21649999730289 |
| }, |
| { |
| "row": 68, |
| "milliseconds": 1.2297500215936452 |
| }, |
| { |
| "row": 130, |
| "milliseconds": 1.1009590234607458 |
| }, |
| { |
| "row": 0, |
| "milliseconds": 0.9331250039394945 |
| }, |
| { |
| "row": 68, |
| "milliseconds": 0.9082920150831342 |
| }, |
| { |
| "row": 130, |
| "milliseconds": 0.9552080009598285 |
| }, |
| { |
| "row": 0, |
| "milliseconds": 1.4009579899720848 |
| }, |
| { |
| "row": 68, |
| "milliseconds": 1.207541994517669 |
| }, |
| { |
| "row": 130, |
| "milliseconds": 0.9119999886024743 |
| }, |
| { |
| "row": 0, |
| "milliseconds": 0.9255839977413416 |
| }, |
| { |
| "row": 68, |
| "milliseconds": 1.0290830105077475 |
| }, |
| { |
| "row": 130, |
| "milliseconds": 0.932042021304369 |
| }, |
| { |
| "row": 0, |
| "milliseconds": 0.9105000062845647 |
| }, |
| { |
| "row": 68, |
| "milliseconds": 0.9429579949937761 |
| }, |
| { |
| "row": 130, |
| "milliseconds": 0.9207910043187439 |
| }, |
| { |
| "row": 0, |
| "milliseconds": 0.8967499888967723 |
| }, |
| { |
| "row": 68, |
| "milliseconds": 1.2767080042976886 |
| }, |
| { |
| "row": 130, |
| "milliseconds": 0.9552080009598285 |
| }, |
| { |
| "row": 0, |
| "milliseconds": 0.9043329919222742 |
| }, |
| { |
| "row": 68, |
| "milliseconds": 0.9005409956444055 |
| }, |
| { |
| "row": 130, |
| "milliseconds": 0.8939160034060478 |
| }, |
| { |
| "row": 0, |
| "milliseconds": 0.8917910163290799 |
| }, |
| { |
| "row": 68, |
| "milliseconds": 0.8989170019049197 |
| }, |
| { |
| "row": 130, |
| "milliseconds": 0.8875419734977186 |
| }, |
| { |
| "row": 0, |
| "milliseconds": 0.88900001719594 |
| }, |
| { |
| "row": 68, |
| "milliseconds": 0.8909160096663982 |
| }, |
| { |
| "row": 130, |
| "milliseconds": 0.8875839994288981 |
| } |
| ], |
| "compute_plan": { |
| "kind": "MLComputePlan preferred device; not a runtime utilization or power trace", |
| "assigned_operation_count": 173, |
| "device_counts": { |
| "cpu": 0, |
| "gpu": 173, |
| "ane": 0, |
| "unknown": 0 |
| }, |
| "unassigned_operation_types": { |
| "const": 239 |
| }, |
| "operations": [ |
| { |
| "name": "option_mask_1", |
| "type": "ios17.not_equal", |
| "device": "gpu" |
| }, |
| { |
| "name": "context_mask", |
| "type": "ios17.not_equal", |
| "device": "gpu" |
| }, |
| { |
| "name": "var_27", |
| "type": "ios17.slice_by_index", |
| "device": "gpu" |
| }, |
| { |
| "name": "fill_like_0_cast_fp16", |
| "type": "ios16.fill_like", |
| "device": "gpu" |
| }, |
| { |
| "name": "var_43", |
| "type": "ios17.slice_by_index", |
| "device": "gpu" |
| }, |
| { |
| "name": "var_43_to_fp16", |
| "type": "ios17.cast", |
| "device": "gpu" |
| }, |
| { |
| "name": "safe_context_ids_cast_fp16", |
| "type": "ios17.concat", |
| "device": "gpu" |
| }, |
| { |
| "name": "context_ids_to_int16", |
| "type": "ios17.cast", |
| "device": "gpu" |
| }, |
| { |
| "name": "cast_54", |
| "type": "ios17.cast", |
| "device": "gpu" |
| }, |
| { |
| "name": "greater_equal_0", |
| "type": "ios17.greater_equal", |
| "device": "gpu" |
| }, |
| { |
| "name": "add_3", |
| "type": "ios17.add", |
| "device": "gpu" |
| }, |
| { |
| "name": "select_0", |
| "type": "select", |
| "device": "gpu" |
| }, |
| { |
| "name": "select_0_to_int16", |
| "type": "ios17.cast", |
| "device": "gpu" |
| }, |
| { |
| "name": "var_59_cast_fp16_cast_uint16_cast_uint16", |
| "type": "ios17.gather", |
| "device": "gpu" |
| }, |
| { |
| "name": "src_1_cast_fp16", |
| "type": "ios17.add", |
| "device": "gpu" |
| }, |
| { |
| "name": "mask_1_cast_fp16", |
| "type": "ios17.equal", |
| "device": "gpu" |
| }, |
| { |
| "name": "key_padding_mask_1_cast_fp16", |
| "type": "select", |
| "device": "gpu" |
| }, |
| { |
| "name": "query_1_cast_fp16", |
| "type": "ios17.layer_norm", |
| "device": "gpu" |
| }, |
| { |
| "name": "query_3_cast_fp16", |
| "type": "ios17.transpose", |
| "device": "gpu" |
| }, |
| { |
| "name": "linear_0_cast_fp16", |
| "type": "ios17.linear", |
| "device": "gpu" |
| }, |
| { |
| "name": "var_124_cast_fp16", |
| "type": "ios17.reshape", |
| "device": "gpu" |
| }, |
| { |
| "name": "var_125_cast_fp16", |
| "type": "ios17.expand_dims", |
| "device": "gpu" |
| }, |
| { |
| "name": "var_126_cast_fp16", |
| "type": "ios17.transpose", |
| "device": "gpu" |
| }, |
| { |
| "name": "var_127_cast_fp16", |
| "type": "ios17.squeeze", |
| "device": "gpu" |
| }, |
| { |
| "name": "q_1_cast_fp16", |
| "type": "ios17.slice_by_index", |
| "device": "gpu" |
| }, |
| { |
| "name": "k_1_cast_fp16", |
| "type": "ios17.slice_by_index", |
| "device": "gpu" |
| }, |
| { |
| "name": "v_1_cast_fp16", |
| "type": "ios17.slice_by_index", |
| "device": "gpu" |
| }, |
| { |
| "name": "var_136_cast_fp16", |
| "type": "ios17.reshape", |
| "device": "gpu" |
| }, |
| { |
| "name": "var_143_cast_fp16", |
| "type": "ios17.reshape", |
| "device": "gpu" |
| }, |
| { |
| "name": "var_150_cast_fp16", |
| "type": "ios17.reshape", |
| "device": "gpu" |
| }, |
| { |
| "name": "var_154_cast_fp16", |
| "type": "ios17.reshape", |
| "device": "gpu" |
| }, |
| { |
| "name": "var_156_cast_fp16", |
| "type": "tile", |
| "device": "gpu" |
| }, |
| { |
| "name": "q_3_cast_fp16", |
| "type": "ios17.transpose", |
| "device": "gpu" |
| }, |
| { |
| "name": "q_5_cast_fp16", |
| "type": "ios17.reshape", |
| "device": "gpu" |
| }, |
| { |
| "name": "k_3_cast_fp16", |
| "type": "ios17.transpose", |
| "device": "gpu" |
| }, |
| { |
| "name": "k_5_cast_fp16", |
| "type": "ios17.reshape", |
| "device": "gpu" |
| }, |
| { |
| "name": "v_3_cast_fp16", |
| "type": "ios17.transpose", |
| "device": "gpu" |
| }, |
| { |
| "name": "v_5_cast_fp16", |
| "type": "ios17.reshape", |
| "device": "gpu" |
| }, |
| { |
| "name": "mul_1_cast_fp16", |
| "type": "ios17.mul", |
| "device": "gpu" |
| }, |
| { |
| "name": "matmul_0_cast_fp16", |
| "type": "ios17.matmul", |
| "device": "gpu" |
| }, |
| { |
| "name": "add_0_cast_fp16", |
| "type": "ios17.add", |
| "device": "gpu" |
| }, |
| { |
| "name": "softmax_0_cast_fp16", |
| "type": "ios16.softmax", |
| "device": "gpu" |
| }, |
| { |
| "name": "attn_output_1_cast_fp16", |
| "type": "ios17.matmul", |
| "device": "gpu" |
| }, |
| { |
| "name": "var_172_cast_fp16", |
| "type": "ios17.transpose", |
| "device": "gpu" |
| }, |
| { |
| "name": "attn_output_3_cast_fp16", |
| "type": "ios17.reshape", |
| "device": "gpu" |
| }, |
| { |
| "name": "linear_1_cast_fp16", |
| "type": "ios17.linear", |
| "device": "gpu" |
| }, |
| { |
| "name": "attn_output_7_cast_fp16", |
| "type": "ios17.reshape", |
| "device": "gpu" |
| }, |
| { |
| "name": "input_3_cast_fp16", |
| "type": "ios17.transpose", |
| "device": "gpu" |
| }, |
| { |
| "name": "input_5_cast_fp16", |
| "type": "ios17.add", |
| "device": "gpu" |
| }, |
| { |
| "name": "input_7_cast_fp16", |
| "type": "ios17.layer_norm", |
| "device": "gpu" |
| }, |
| { |
| "name": "linear_2_cast_fp16", |
| "type": "ios17.linear", |
| "device": "gpu" |
| }, |
| { |
| "name": "input_11_cast_fp16", |
| "type": "ios16.relu", |
| "device": "gpu" |
| }, |
| { |
| "name": "linear_3_cast_fp16", |
| "type": "ios17.linear", |
| "device": "gpu" |
| }, |
| { |
| "name": "input_17_cast_fp16", |
| "type": "ios17.add", |
| "device": "gpu" |
| }, |
| { |
| "name": "query_5_cast_fp16", |
| "type": "ios17.layer_norm", |
| "device": "gpu" |
| }, |
| { |
| "name": "query_7_cast_fp16", |
| "type": "ios17.transpose", |
| "device": "gpu" |
| }, |
| { |
| "name": "linear_4_cast_fp16", |
| "type": "ios17.linear", |
| "device": "gpu" |
| }, |
| { |
| "name": "var_231_cast_fp16", |
| "type": "ios17.reshape", |
| "device": "gpu" |
| }, |
| { |
| "name": "var_232_cast_fp16", |
| "type": "ios17.expand_dims", |
| "device": "gpu" |
| }, |
| { |
| "name": "var_233_cast_fp16", |
| "type": "ios17.transpose", |
| "device": "gpu" |
| }, |
| { |
| "name": "var_234_cast_fp16", |
| "type": "ios17.squeeze", |
| "device": "gpu" |
| }, |
| { |
| "name": "q_7_cast_fp16", |
| "type": "ios17.slice_by_index", |
| "device": "gpu" |
| }, |
| { |
| "name": "k_7_cast_fp16", |
| "type": "ios17.slice_by_index", |
| "device": "gpu" |
| }, |
| { |
| "name": "v_7_cast_fp16", |
| "type": "ios17.slice_by_index", |
| "device": "gpu" |
| }, |
| { |
| "name": "var_243_cast_fp16", |
| "type": "ios17.reshape", |
| "device": "gpu" |
| }, |
| { |
| "name": "var_250_cast_fp16", |
| "type": "ios17.reshape", |
| "device": "gpu" |
| }, |
| { |
| "name": "var_257_cast_fp16", |
| "type": "ios17.reshape", |
| "device": "gpu" |
| }, |
| { |
| "name": "q_9_cast_fp16", |
| "type": "ios17.transpose", |
| "device": "gpu" |
| }, |
| { |
| "name": "q_11_cast_fp16", |
| "type": "ios17.reshape", |
| "device": "gpu" |
| }, |
| { |
| "name": "k_9_cast_fp16", |
| "type": "ios17.transpose", |
| "device": "gpu" |
| }, |
| { |
| "name": "k_11_cast_fp16", |
| "type": "ios17.reshape", |
| "device": "gpu" |
| }, |
| { |
| "name": "v_9_cast_fp16", |
| "type": "ios17.transpose", |
| "device": "gpu" |
| }, |
| { |
| "name": "v_11_cast_fp16", |
| "type": "ios17.reshape", |
| "device": "gpu" |
| }, |
| { |
| "name": "mul_3_cast_fp16", |
| "type": "ios17.mul", |
| "device": "gpu" |
| }, |
| { |
| "name": "matmul_1_cast_fp16", |
| "type": "ios17.matmul", |
| "device": "gpu" |
| }, |
| { |
| "name": "add_1_cast_fp16", |
| "type": "ios17.add", |
| "device": "gpu" |
| }, |
| { |
| "name": "softmax_1_cast_fp16", |
| "type": "ios16.softmax", |
| "device": "gpu" |
| }, |
| { |
| "name": "attn_output_9_cast_fp16", |
| "type": "ios17.matmul", |
| "device": "gpu" |
| }, |
| { |
| "name": "var_279_cast_fp16", |
| "type": "ios17.transpose", |
| "device": "gpu" |
| }, |
| { |
| "name": "attn_output_11_cast_fp16", |
| "type": "ios17.reshape", |
| "device": "gpu" |
| }, |
| { |
| "name": "linear_5_cast_fp16", |
| "type": "ios17.linear", |
| "device": "gpu" |
| }, |
| { |
| "name": "attn_output_15_cast_fp16", |
| "type": "ios17.reshape", |
| "device": "gpu" |
| }, |
| { |
| "name": "input_19_cast_fp16", |
| "type": "ios17.transpose", |
| "device": "gpu" |
| }, |
| { |
| "name": "input_21_cast_fp16", |
| "type": "ios17.add", |
| "device": "gpu" |
| }, |
| { |
| "name": "input_23_cast_fp16", |
| "type": "ios17.layer_norm", |
| "device": "gpu" |
| }, |
| { |
| "name": "linear_6_cast_fp16", |
| "type": "ios17.linear", |
| "device": "gpu" |
| }, |
| { |
| "name": "input_27_cast_fp16", |
| "type": "ios16.relu", |
| "device": "gpu" |
| }, |
| { |
| "name": "linear_7_cast_fp16", |
| "type": "ios17.linear", |
| "device": "gpu" |
| }, |
| { |
| "name": "context_cast_fp16", |
| "type": "ios17.add", |
| "device": "gpu" |
| }, |
| { |
| "name": "flat_ids", |
| "type": "ios17.reshape", |
| "device": "gpu" |
| }, |
| { |
| "name": "flat_mask", |
| "type": "ios17.not_equal", |
| "device": "gpu" |
| }, |
| { |
| "name": "var_333", |
| "type": "ios17.slice_by_index", |
| "device": "gpu" |
| }, |
| { |
| "name": "fill_like_1_cast_fp16", |
| "type": "ios16.fill_like", |
| "device": "gpu" |
| }, |
| { |
| "name": "var_349", |
| "type": "ios17.slice_by_index", |
| "device": "gpu" |
| }, |
| { |
| "name": "var_349_to_fp16", |
| "type": "ios17.cast", |
| "device": "gpu" |
| }, |
| { |
| "name": "safe_ids_cast_fp16", |
| "type": "ios17.concat", |
| "device": "gpu" |
| }, |
| { |
| "name": "flat_ids_to_int16", |
| "type": "ios17.cast", |
| "device": "gpu" |
| }, |
| { |
| "name": "cast_55", |
| "type": "ios17.cast", |
| "device": "gpu" |
| }, |
| { |
| "name": "greater_equal_1", |
| "type": "ios17.greater_equal", |
| "device": "gpu" |
| }, |
| { |
| "name": "add_4", |
| "type": "ios17.add", |
| "device": "gpu" |
| }, |
| { |
| "name": "select_1", |
| "type": "select", |
| "device": "gpu" |
| }, |
| { |
| "name": "select_1_to_int16", |
| "type": "ios17.cast", |
| "device": "gpu" |
| }, |
| { |
| "name": "var_365_cast_fp16_cast_uint16_cast_uint16", |
| "type": "ios17.gather", |
| "device": "gpu" |
| }, |
| { |
| "name": "src_cast_fp16", |
| "type": "ios17.add", |
| "device": "gpu" |
| }, |
| { |
| "name": "mask_cast_fp16", |
| "type": "ios17.equal", |
| "device": "gpu" |
| }, |
| { |
| "name": "key_padding_mask_7_cast_fp16", |
| "type": "select", |
| "device": "gpu" |
| }, |
| { |
| "name": "query_9_cast_fp16", |
| "type": "ios17.layer_norm", |
| "device": "gpu" |
| }, |
| { |
| "name": "query_11_cast_fp16", |
| "type": "ios17.transpose", |
| "device": "gpu" |
| }, |
| { |
| "name": "linear_8_cast_fp16", |
| "type": "ios17.linear", |
| "device": "gpu" |
| }, |
| { |
| "name": "var_428_cast_fp16", |
| "type": "ios17.reshape", |
| "device": "gpu" |
| }, |
| { |
| "name": "var_429_cast_fp16", |
| "type": "ios17.expand_dims", |
| "device": "gpu" |
| }, |
| { |
| "name": "var_430_cast_fp16", |
| "type": "ios17.transpose", |
| "device": "gpu" |
| }, |
| { |
| "name": "var_431_cast_fp16", |
| "type": "ios17.squeeze", |
| "device": "gpu" |
| }, |
| { |
| "name": "q_13_cast_fp16", |
| "type": "ios17.slice_by_index", |
| "device": "gpu" |
| }, |
| { |
| "name": "k_13_cast_fp16", |
| "type": "ios17.slice_by_index", |
| "device": "gpu" |
| }, |
| { |
| "name": "v_13_cast_fp16", |
| "type": "ios17.slice_by_index", |
| "device": "gpu" |
| }, |
| { |
| "name": "var_440_cast_fp16", |
| "type": "ios17.reshape", |
| "device": "gpu" |
| }, |
| { |
| "name": "var_447_cast_fp16", |
| "type": "ios17.reshape", |
| "device": "gpu" |
| }, |
| { |
| "name": "var_454_cast_fp16", |
| "type": "ios17.reshape", |
| "device": "gpu" |
| }, |
| { |
| "name": "var_458_cast_fp16", |
| "type": "ios17.reshape", |
| "device": "gpu" |
| }, |
| { |
| "name": "var_460_cast_fp16", |
| "type": "tile", |
| "device": "gpu" |
| }, |
| { |
| "name": "q_15_cast_fp16", |
| "type": "ios17.transpose", |
| "device": "gpu" |
| }, |
| { |
| "name": "q_cast_fp16", |
| "type": "ios17.reshape", |
| "device": "gpu" |
| }, |
| { |
| "name": "k_15_cast_fp16", |
| "type": "ios17.transpose", |
| "device": "gpu" |
| }, |
| { |
| "name": "k_cast_fp16", |
| "type": "ios17.reshape", |
| "device": "gpu" |
| }, |
| { |
| "name": "v_15_cast_fp16", |
| "type": "ios17.transpose", |
| "device": "gpu" |
| }, |
| { |
| "name": "v_cast_fp16", |
| "type": "ios17.reshape", |
| "device": "gpu" |
| }, |
| { |
| "name": "mul_5_cast_fp16", |
| "type": "ios17.mul", |
| "device": "gpu" |
| }, |
| { |
| "name": "matmul_2_cast_fp16", |
| "type": "ios17.matmul", |
| "device": "gpu" |
| }, |
| { |
| "name": "add_2_cast_fp16", |
| "type": "ios17.add", |
| "device": "gpu" |
| }, |
| { |
| "name": "softmax_2_cast_fp16", |
| "type": "ios16.softmax", |
| "device": "gpu" |
| }, |
| { |
| "name": "attn_output_17_cast_fp16", |
| "type": "ios17.matmul", |
| "device": "gpu" |
| }, |
| { |
| "name": "var_476_cast_fp16", |
| "type": "ios17.transpose", |
| "device": "gpu" |
| }, |
| { |
| "name": "attn_output_19_cast_fp16", |
| "type": "ios17.reshape", |
| "device": "gpu" |
| }, |
| { |
| "name": "linear_9_cast_fp16", |
| "type": "ios17.linear", |
| "device": "gpu" |
| }, |
| { |
| "name": "attn_output_cast_fp16", |
| "type": "ios17.reshape", |
| "device": "gpu" |
| }, |
| { |
| "name": "input_35_cast_fp16", |
| "type": "ios17.transpose", |
| "device": "gpu" |
| }, |
| { |
| "name": "input_37_cast_fp16", |
| "type": "ios17.add", |
| "device": "gpu" |
| }, |
| { |
| "name": "input_39_cast_fp16", |
| "type": "ios17.layer_norm", |
| "device": "gpu" |
| }, |
| { |
| "name": "linear_10_cast_fp16", |
| "type": "ios17.linear", |
| "device": "gpu" |
| }, |
| { |
| "name": "input_43_cast_fp16", |
| "type": "ios16.relu", |
| "device": "gpu" |
| }, |
| { |
| "name": "linear_11_cast_fp16", |
| "type": "ios17.linear", |
| "device": "gpu" |
| }, |
| { |
| "name": "hidden_cast_fp16", |
| "type": "ios17.add", |
| "device": "gpu" |
| }, |
| { |
| "name": "var_504", |
| "type": "ios17.expand_dims", |
| "device": "gpu" |
| }, |
| { |
| "name": "var_504_to_fp16", |
| "type": "ios17.cast", |
| "device": "gpu" |
| }, |
| { |
| "name": "var_510_cast_fp16", |
| "type": "ios17.mul", |
| "device": "gpu" |
| }, |
| { |
| "name": "var_515_cast_fp16", |
| "type": "ios16.reduce_sum", |
| "device": "gpu" |
| }, |
| { |
| "name": "var_520_cast_fp16", |
| "type": "ios16.reduce_sum", |
| "device": "gpu" |
| }, |
| { |
| "name": "var_522_cast_fp16", |
| "type": "ios17.maximum", |
| "device": "gpu" |
| }, |
| { |
| "name": "pooled_cast_fp16", |
| "type": "ios17.real_div", |
| "device": "gpu" |
| }, |
| { |
| "name": "options_cast_fp16", |
| "type": "ios17.reshape", |
| "device": "gpu" |
| }, |
| { |
| "name": "input_cast_fp16", |
| "type": "ios17.layer_norm", |
| "device": "gpu" |
| }, |
| { |
| "name": "input_53_cast_fp16", |
| "type": "ios17.layer_norm", |
| "device": "gpu" |
| }, |
| { |
| "name": "linear_12_cast_fp16", |
| "type": "ios17.linear", |
| "device": "gpu" |
| }, |
| { |
| "name": "linear_13_cast_fp16", |
| "type": "ios17.linear", |
| "device": "gpu" |
| }, |
| { |
| "name": "linear_14_cast_fp16", |
| "type": "ios17.linear", |
| "device": "gpu" |
| }, |
| { |
| "name": "matmul_3_cast_fp16", |
| "type": "ios17.matmul", |
| "device": "gpu" |
| }, |
| { |
| "name": "_inversed_scores_1_cast_fp16", |
| "type": "ios17.mul", |
| "device": "gpu" |
| }, |
| { |
| "name": "var_567", |
| "type": "ios17.expand_dims", |
| "device": "gpu" |
| }, |
| { |
| "name": "var_569", |
| "type": "ios17.logical_not", |
| "device": "gpu" |
| }, |
| { |
| "name": "scores_cast_fp16", |
| "type": "select", |
| "device": "gpu" |
| }, |
| { |
| "name": "var_571_cast_fp16", |
| "type": "ios16.softmax", |
| "device": "gpu" |
| }, |
| { |
| "name": "matmul_4_cast_fp16", |
| "type": "ios17.matmul", |
| "device": "gpu" |
| }, |
| { |
| "name": "var_574_cast_fp16", |
| "type": "ios17.mul", |
| "device": "gpu" |
| }, |
| { |
| "name": "var_576_cast_fp16", |
| "type": "ios16.reduce_sum", |
| "device": "gpu" |
| }, |
| { |
| "name": "_inversed_logits_1_cast_fp16", |
| "type": "ios17.mul", |
| "device": "gpu" |
| }, |
| { |
| "name": "var_579", |
| "type": "ios17.logical_not", |
| "device": "gpu" |
| }, |
| { |
| "name": "logits_3_cast_fp16", |
| "type": "select", |
| "device": "gpu" |
| }, |
| { |
| "name": "var_587_cast_fp16", |
| "type": "ios16.fill_like", |
| "device": "gpu" |
| }, |
| { |
| "name": "logits_cast_fp16", |
| "type": "select", |
| "device": "gpu" |
| }, |
| { |
| "name": "var_591_cast_fp16", |
| "type": "ios16.softmax", |
| "device": "gpu" |
| }, |
| { |
| "name": "probabilities", |
| "type": "ios17.cast", |
| "device": "gpu" |
| }, |
| { |
| "name": "logits", |
| "type": "ios17.cast", |
| "device": "gpu" |
| } |
| ] |
| } |
| }, |
| "CPU_AND_NE": { |
| "load_ms": 566.8206670088693, |
| "first_prediction_ms": 1.6522080113645643, |
| "warm_prediction": { |
| "count": 30, |
| "median_ms": 0.9285830019507557, |
| "p95_ms": 0.9725897951284423, |
| "min_ms": 0.9064170008059591, |
| "max_ms": 1.2301660026423633 |
| }, |
| "timed_predictions_correct": 30, |
| "calls": [ |
| { |
| "row": 0, |
| "milliseconds": 0.9575410222169012 |
| }, |
| { |
| "row": 68, |
| "milliseconds": 0.9647910192143172 |
| }, |
| { |
| "row": 130, |
| "milliseconds": 0.9704580006655306 |
| }, |
| { |
| "row": 0, |
| "milliseconds": 0.9583340142853558 |
| }, |
| { |
| "row": 68, |
| "milliseconds": 0.9467090130783617 |
| }, |
| { |
| "row": 130, |
| "milliseconds": 1.2301660026423633 |
| }, |
| { |
| "row": 0, |
| "milliseconds": 0.9743339905980974 |
| }, |
| { |
| "row": 68, |
| "milliseconds": 0.9361249976791441 |
| }, |
| { |
| "row": 130, |
| "milliseconds": 0.9265419794246554 |
| }, |
| { |
| "row": 0, |
| "milliseconds": 0.9583329956512898 |
| }, |
| { |
| "row": 68, |
| "milliseconds": 0.9652920125517994 |
| }, |
| { |
| "row": 130, |
| "milliseconds": 0.9257500059902668 |
| }, |
| { |
| "row": 0, |
| "milliseconds": 0.9283330000471324 |
| }, |
| { |
| "row": 68, |
| "milliseconds": 0.9066659840755165 |
| }, |
| { |
| "row": 130, |
| "milliseconds": 0.9307079890277237 |
| }, |
| { |
| "row": 0, |
| "milliseconds": 0.9274999902117997 |
| }, |
| { |
| "row": 68, |
| "milliseconds": 0.9069579828064889 |
| }, |
| { |
| "row": 130, |
| "milliseconds": 0.9290000016335398 |
| }, |
| { |
| "row": 0, |
| "milliseconds": 0.927332992432639 |
| }, |
| { |
| "row": 68, |
| "milliseconds": 0.917499972274527 |
| }, |
| { |
| "row": 130, |
| "milliseconds": 0.9173330035991967 |
| }, |
| { |
| "row": 0, |
| "milliseconds": 0.9288330038543791 |
| }, |
| { |
| "row": 68, |
| "milliseconds": 0.9251659794244915 |
| }, |
| { |
| "row": 130, |
| "milliseconds": 0.9205000242218375 |
| }, |
| { |
| "row": 0, |
| "milliseconds": 0.9272919851355255 |
| }, |
| { |
| "row": 68, |
| "milliseconds": 0.9192500147037208 |
| }, |
| { |
| "row": 130, |
| "milliseconds": 0.933834002353251 |
| }, |
| { |
| "row": 0, |
| "milliseconds": 0.9269170113839209 |
| }, |
| { |
| "row": 68, |
| "milliseconds": 0.9406250028405339 |
| }, |
| { |
| "row": 130, |
| "milliseconds": 0.9064170008059591 |
| } |
| ], |
| "compute_plan": { |
| "kind": "MLComputePlan preferred device; not a runtime utilization or power trace", |
| "assigned_operation_count": 173, |
| "device_counts": { |
| "cpu": 24, |
| "gpu": 0, |
| "ane": 149, |
| "unknown": 0 |
| }, |
| "unassigned_operation_types": { |
| "const": 239 |
| }, |
| "operations": [ |
| { |
| "name": "option_mask_1", |
| "type": "ios17.not_equal", |
| "device": "cpu" |
| }, |
| { |
| "name": "context_mask", |
| "type": "ios17.not_equal", |
| "device": "cpu" |
| }, |
| { |
| "name": "var_27", |
| "type": "ios17.slice_by_index", |
| "device": "cpu" |
| }, |
| { |
| "name": "fill_like_0_cast_fp16", |
| "type": "ios16.fill_like", |
| "device": "ane" |
| }, |
| { |
| "name": "var_43", |
| "type": "ios17.slice_by_index", |
| "device": "cpu" |
| }, |
| { |
| "name": "var_43_to_fp16", |
| "type": "ios17.cast", |
| "device": "cpu" |
| }, |
| { |
| "name": "safe_context_ids_cast_fp16", |
| "type": "ios17.concat", |
| "device": "ane" |
| }, |
| { |
| "name": "context_ids_to_int16", |
| "type": "ios17.cast", |
| "device": "cpu" |
| }, |
| { |
| "name": "cast_54", |
| "type": "ios17.cast", |
| "device": "cpu" |
| }, |
| { |
| "name": "greater_equal_0", |
| "type": "ios17.greater_equal", |
| "device": "cpu" |
| }, |
| { |
| "name": "add_3", |
| "type": "ios17.add", |
| "device": "cpu" |
| }, |
| { |
| "name": "select_0", |
| "type": "select", |
| "device": "cpu" |
| }, |
| { |
| "name": "select_0_to_int16", |
| "type": "ios17.cast", |
| "device": "cpu" |
| }, |
| { |
| "name": "var_59_cast_fp16_cast_uint16_cast_uint16", |
| "type": "ios17.gather", |
| "device": "cpu" |
| }, |
| { |
| "name": "src_1_cast_fp16", |
| "type": "ios17.add", |
| "device": "ane" |
| }, |
| { |
| "name": "mask_1_cast_fp16", |
| "type": "ios17.equal", |
| "device": "ane" |
| }, |
| { |
| "name": "key_padding_mask_1_cast_fp16", |
| "type": "select", |
| "device": "ane" |
| }, |
| { |
| "name": "query_1_cast_fp16", |
| "type": "ios17.layer_norm", |
| "device": "ane" |
| }, |
| { |
| "name": "query_3_cast_fp16", |
| "type": "ios17.transpose", |
| "device": "ane" |
| }, |
| { |
| "name": "linear_0_cast_fp16", |
| "type": "ios17.linear", |
| "device": "ane" |
| }, |
| { |
| "name": "var_124_cast_fp16", |
| "type": "ios17.reshape", |
| "device": "ane" |
| }, |
| { |
| "name": "var_125_cast_fp16", |
| "type": "ios17.expand_dims", |
| "device": "ane" |
| }, |
| { |
| "name": "var_126_cast_fp16", |
| "type": "ios17.transpose", |
| "device": "ane" |
| }, |
| { |
| "name": "var_127_cast_fp16", |
| "type": "ios17.squeeze", |
| "device": "ane" |
| }, |
| { |
| "name": "q_1_cast_fp16", |
| "type": "ios17.slice_by_index", |
| "device": "ane" |
| }, |
| { |
| "name": "k_1_cast_fp16", |
| "type": "ios17.slice_by_index", |
| "device": "ane" |
| }, |
| { |
| "name": "v_1_cast_fp16", |
| "type": "ios17.slice_by_index", |
| "device": "ane" |
| }, |
| { |
| "name": "var_136_cast_fp16", |
| "type": "ios17.reshape", |
| "device": "ane" |
| }, |
| { |
| "name": "var_143_cast_fp16", |
| "type": "ios17.reshape", |
| "device": "ane" |
| }, |
| { |
| "name": "var_150_cast_fp16", |
| "type": "ios17.reshape", |
| "device": "ane" |
| }, |
| { |
| "name": "var_154_cast_fp16", |
| "type": "ios17.reshape", |
| "device": "ane" |
| }, |
| { |
| "name": "var_156_cast_fp16", |
| "type": "tile", |
| "device": "ane" |
| }, |
| { |
| "name": "q_3_cast_fp16", |
| "type": "ios17.transpose", |
| "device": "ane" |
| }, |
| { |
| "name": "q_5_cast_fp16", |
| "type": "ios17.reshape", |
| "device": "ane" |
| }, |
| { |
| "name": "k_3_cast_fp16", |
| "type": "ios17.transpose", |
| "device": "ane" |
| }, |
| { |
| "name": "k_5_cast_fp16", |
| "type": "ios17.reshape", |
| "device": "ane" |
| }, |
| { |
| "name": "v_3_cast_fp16", |
| "type": "ios17.transpose", |
| "device": "ane" |
| }, |
| { |
| "name": "v_5_cast_fp16", |
| "type": "ios17.reshape", |
| "device": "ane" |
| }, |
| { |
| "name": "mul_1_cast_fp16", |
| "type": "ios17.mul", |
| "device": "ane" |
| }, |
| { |
| "name": "matmul_0_cast_fp16", |
| "type": "ios17.matmul", |
| "device": "ane" |
| }, |
| { |
| "name": "add_0_cast_fp16", |
| "type": "ios17.add", |
| "device": "ane" |
| }, |
| { |
| "name": "softmax_0_cast_fp16", |
| "type": "ios16.softmax", |
| "device": "ane" |
| }, |
| { |
| "name": "attn_output_1_cast_fp16", |
| "type": "ios17.matmul", |
| "device": "ane" |
| }, |
| { |
| "name": "var_172_cast_fp16", |
| "type": "ios17.transpose", |
| "device": "ane" |
| }, |
| { |
| "name": "attn_output_3_cast_fp16", |
| "type": "ios17.reshape", |
| "device": "ane" |
| }, |
| { |
| "name": "linear_1_cast_fp16", |
| "type": "ios17.linear", |
| "device": "ane" |
| }, |
| { |
| "name": "attn_output_7_cast_fp16", |
| "type": "ios17.reshape", |
| "device": "ane" |
| }, |
| { |
| "name": "input_3_cast_fp16", |
| "type": "ios17.transpose", |
| "device": "ane" |
| }, |
| { |
| "name": "input_5_cast_fp16", |
| "type": "ios17.add", |
| "device": "ane" |
| }, |
| { |
| "name": "input_7_cast_fp16", |
| "type": "ios17.layer_norm", |
| "device": "ane" |
| }, |
| { |
| "name": "linear_2_cast_fp16", |
| "type": "ios17.linear", |
| "device": "ane" |
| }, |
| { |
| "name": "input_11_cast_fp16", |
| "type": "ios16.relu", |
| "device": "ane" |
| }, |
| { |
| "name": "linear_3_cast_fp16", |
| "type": "ios17.linear", |
| "device": "ane" |
| }, |
| { |
| "name": "input_17_cast_fp16", |
| "type": "ios17.add", |
| "device": "ane" |
| }, |
| { |
| "name": "query_5_cast_fp16", |
| "type": "ios17.layer_norm", |
| "device": "ane" |
| }, |
| { |
| "name": "query_7_cast_fp16", |
| "type": "ios17.transpose", |
| "device": "ane" |
| }, |
| { |
| "name": "linear_4_cast_fp16", |
| "type": "ios17.linear", |
| "device": "ane" |
| }, |
| { |
| "name": "var_231_cast_fp16", |
| "type": "ios17.reshape", |
| "device": "ane" |
| }, |
| { |
| "name": "var_232_cast_fp16", |
| "type": "ios17.expand_dims", |
| "device": "ane" |
| }, |
| { |
| "name": "var_233_cast_fp16", |
| "type": "ios17.transpose", |
| "device": "ane" |
| }, |
| { |
| "name": "var_234_cast_fp16", |
| "type": "ios17.squeeze", |
| "device": "ane" |
| }, |
| { |
| "name": "q_7_cast_fp16", |
| "type": "ios17.slice_by_index", |
| "device": "ane" |
| }, |
| { |
| "name": "k_7_cast_fp16", |
| "type": "ios17.slice_by_index", |
| "device": "ane" |
| }, |
| { |
| "name": "v_7_cast_fp16", |
| "type": "ios17.slice_by_index", |
| "device": "ane" |
| }, |
| { |
| "name": "var_243_cast_fp16", |
| "type": "ios17.reshape", |
| "device": "ane" |
| }, |
| { |
| "name": "var_250_cast_fp16", |
| "type": "ios17.reshape", |
| "device": "ane" |
| }, |
| { |
| "name": "var_257_cast_fp16", |
| "type": "ios17.reshape", |
| "device": "ane" |
| }, |
| { |
| "name": "q_9_cast_fp16", |
| "type": "ios17.transpose", |
| "device": "ane" |
| }, |
| { |
| "name": "q_11_cast_fp16", |
| "type": "ios17.reshape", |
| "device": "ane" |
| }, |
| { |
| "name": "k_9_cast_fp16", |
| "type": "ios17.transpose", |
| "device": "ane" |
| }, |
| { |
| "name": "k_11_cast_fp16", |
| "type": "ios17.reshape", |
| "device": "ane" |
| }, |
| { |
| "name": "v_9_cast_fp16", |
| "type": "ios17.transpose", |
| "device": "ane" |
| }, |
| { |
| "name": "v_11_cast_fp16", |
| "type": "ios17.reshape", |
| "device": "ane" |
| }, |
| { |
| "name": "mul_3_cast_fp16", |
| "type": "ios17.mul", |
| "device": "ane" |
| }, |
| { |
| "name": "matmul_1_cast_fp16", |
| "type": "ios17.matmul", |
| "device": "ane" |
| }, |
| { |
| "name": "add_1_cast_fp16", |
| "type": "ios17.add", |
| "device": "ane" |
| }, |
| { |
| "name": "softmax_1_cast_fp16", |
| "type": "ios16.softmax", |
| "device": "ane" |
| }, |
| { |
| "name": "attn_output_9_cast_fp16", |
| "type": "ios17.matmul", |
| "device": "ane" |
| }, |
| { |
| "name": "var_279_cast_fp16", |
| "type": "ios17.transpose", |
| "device": "ane" |
| }, |
| { |
| "name": "attn_output_11_cast_fp16", |
| "type": "ios17.reshape", |
| "device": "ane" |
| }, |
| { |
| "name": "linear_5_cast_fp16", |
| "type": "ios17.linear", |
| "device": "ane" |
| }, |
| { |
| "name": "attn_output_15_cast_fp16", |
| "type": "ios17.reshape", |
| "device": "ane" |
| }, |
| { |
| "name": "input_19_cast_fp16", |
| "type": "ios17.transpose", |
| "device": "ane" |
| }, |
| { |
| "name": "input_21_cast_fp16", |
| "type": "ios17.add", |
| "device": "ane" |
| }, |
| { |
| "name": "input_23_cast_fp16", |
| "type": "ios17.layer_norm", |
| "device": "ane" |
| }, |
| { |
| "name": "linear_6_cast_fp16", |
| "type": "ios17.linear", |
| "device": "ane" |
| }, |
| { |
| "name": "input_27_cast_fp16", |
| "type": "ios16.relu", |
| "device": "ane" |
| }, |
| { |
| "name": "linear_7_cast_fp16", |
| "type": "ios17.linear", |
| "device": "ane" |
| }, |
| { |
| "name": "context_cast_fp16", |
| "type": "ios17.add", |
| "device": "ane" |
| }, |
| { |
| "name": "flat_ids", |
| "type": "ios17.reshape", |
| "device": "cpu" |
| }, |
| { |
| "name": "flat_mask", |
| "type": "ios17.not_equal", |
| "device": "cpu" |
| }, |
| { |
| "name": "var_333", |
| "type": "ios17.slice_by_index", |
| "device": "cpu" |
| }, |
| { |
| "name": "fill_like_1_cast_fp16", |
| "type": "ios16.fill_like", |
| "device": "ane" |
| }, |
| { |
| "name": "var_349", |
| "type": "ios17.slice_by_index", |
| "device": "cpu" |
| }, |
| { |
| "name": "var_349_to_fp16", |
| "type": "ios17.cast", |
| "device": "cpu" |
| }, |
| { |
| "name": "safe_ids_cast_fp16", |
| "type": "ios17.concat", |
| "device": "ane" |
| }, |
| { |
| "name": "flat_ids_to_int16", |
| "type": "ios17.cast", |
| "device": "cpu" |
| }, |
| { |
| "name": "cast_55", |
| "type": "ios17.cast", |
| "device": "cpu" |
| }, |
| { |
| "name": "greater_equal_1", |
| "type": "ios17.greater_equal", |
| "device": "cpu" |
| }, |
| { |
| "name": "add_4", |
| "type": "ios17.add", |
| "device": "cpu" |
| }, |
| { |
| "name": "select_1", |
| "type": "select", |
| "device": "cpu" |
| }, |
| { |
| "name": "select_1_to_int16", |
| "type": "ios17.cast", |
| "device": "cpu" |
| }, |
| { |
| "name": "var_365_cast_fp16_cast_uint16_cast_uint16", |
| "type": "ios17.gather", |
| "device": "cpu" |
| }, |
| { |
| "name": "src_cast_fp16", |
| "type": "ios17.add", |
| "device": "ane" |
| }, |
| { |
| "name": "mask_cast_fp16", |
| "type": "ios17.equal", |
| "device": "ane" |
| }, |
| { |
| "name": "key_padding_mask_7_cast_fp16", |
| "type": "select", |
| "device": "ane" |
| }, |
| { |
| "name": "query_9_cast_fp16", |
| "type": "ios17.layer_norm", |
| "device": "ane" |
| }, |
| { |
| "name": "query_11_cast_fp16", |
| "type": "ios17.transpose", |
| "device": "ane" |
| }, |
| { |
| "name": "linear_8_cast_fp16", |
| "type": "ios17.linear", |
| "device": "ane" |
| }, |
| { |
| "name": "var_428_cast_fp16", |
| "type": "ios17.reshape", |
| "device": "ane" |
| }, |
| { |
| "name": "var_429_cast_fp16", |
| "type": "ios17.expand_dims", |
| "device": "ane" |
| }, |
| { |
| "name": "var_430_cast_fp16", |
| "type": "ios17.transpose", |
| "device": "ane" |
| }, |
| { |
| "name": "var_431_cast_fp16", |
| "type": "ios17.squeeze", |
| "device": "ane" |
| }, |
| { |
| "name": "q_13_cast_fp16", |
| "type": "ios17.slice_by_index", |
| "device": "ane" |
| }, |
| { |
| "name": "k_13_cast_fp16", |
| "type": "ios17.slice_by_index", |
| "device": "ane" |
| }, |
| { |
| "name": "v_13_cast_fp16", |
| "type": "ios17.slice_by_index", |
| "device": "ane" |
| }, |
| { |
| "name": "var_440_cast_fp16", |
| "type": "ios17.reshape", |
| "device": "ane" |
| }, |
| { |
| "name": "var_447_cast_fp16", |
| "type": "ios17.reshape", |
| "device": "ane" |
| }, |
| { |
| "name": "var_454_cast_fp16", |
| "type": "ios17.reshape", |
| "device": "ane" |
| }, |
| { |
| "name": "var_458_cast_fp16", |
| "type": "ios17.reshape", |
| "device": "ane" |
| }, |
| { |
| "name": "var_460_cast_fp16", |
| "type": "tile", |
| "device": "ane" |
| }, |
| { |
| "name": "q_15_cast_fp16", |
| "type": "ios17.transpose", |
| "device": "ane" |
| }, |
| { |
| "name": "q_cast_fp16", |
| "type": "ios17.reshape", |
| "device": "ane" |
| }, |
| { |
| "name": "k_15_cast_fp16", |
| "type": "ios17.transpose", |
| "device": "ane" |
| }, |
| { |
| "name": "k_cast_fp16", |
| "type": "ios17.reshape", |
| "device": "ane" |
| }, |
| { |
| "name": "v_15_cast_fp16", |
| "type": "ios17.transpose", |
| "device": "ane" |
| }, |
| { |
| "name": "v_cast_fp16", |
| "type": "ios17.reshape", |
| "device": "ane" |
| }, |
| { |
| "name": "mul_5_cast_fp16", |
| "type": "ios17.mul", |
| "device": "ane" |
| }, |
| { |
| "name": "matmul_2_cast_fp16", |
| "type": "ios17.matmul", |
| "device": "ane" |
| }, |
| { |
| "name": "add_2_cast_fp16", |
| "type": "ios17.add", |
| "device": "ane" |
| }, |
| { |
| "name": "softmax_2_cast_fp16", |
| "type": "ios16.softmax", |
| "device": "ane" |
| }, |
| { |
| "name": "attn_output_17_cast_fp16", |
| "type": "ios17.matmul", |
| "device": "ane" |
| }, |
| { |
| "name": "var_476_cast_fp16", |
| "type": "ios17.transpose", |
| "device": "ane" |
| }, |
| { |
| "name": "attn_output_19_cast_fp16", |
| "type": "ios17.reshape", |
| "device": "ane" |
| }, |
| { |
| "name": "linear_9_cast_fp16", |
| "type": "ios17.linear", |
| "device": "ane" |
| }, |
| { |
| "name": "attn_output_cast_fp16", |
| "type": "ios17.reshape", |
| "device": "ane" |
| }, |
| { |
| "name": "input_35_cast_fp16", |
| "type": "ios17.transpose", |
| "device": "ane" |
| }, |
| { |
| "name": "input_37_cast_fp16", |
| "type": "ios17.add", |
| "device": "ane" |
| }, |
| { |
| "name": "input_39_cast_fp16", |
| "type": "ios17.layer_norm", |
| "device": "ane" |
| }, |
| { |
| "name": "linear_10_cast_fp16", |
| "type": "ios17.linear", |
| "device": "ane" |
| }, |
| { |
| "name": "input_43_cast_fp16", |
| "type": "ios16.relu", |
| "device": "ane" |
| }, |
| { |
| "name": "linear_11_cast_fp16", |
| "type": "ios17.linear", |
| "device": "ane" |
| }, |
| { |
| "name": "hidden_cast_fp16", |
| "type": "ios17.add", |
| "device": "ane" |
| }, |
| { |
| "name": "var_504", |
| "type": "ios17.expand_dims", |
| "device": "ane" |
| }, |
| { |
| "name": "var_504_to_fp16", |
| "type": "ios17.cast", |
| "device": "ane" |
| }, |
| { |
| "name": "var_510_cast_fp16", |
| "type": "ios17.mul", |
| "device": "ane" |
| }, |
| { |
| "name": "var_515_cast_fp16", |
| "type": "ios16.reduce_sum", |
| "device": "ane" |
| }, |
| { |
| "name": "var_520_cast_fp16", |
| "type": "ios16.reduce_sum", |
| "device": "ane" |
| }, |
| { |
| "name": "var_522_cast_fp16", |
| "type": "ios17.maximum", |
| "device": "ane" |
| }, |
| { |
| "name": "pooled_cast_fp16", |
| "type": "ios17.real_div", |
| "device": "ane" |
| }, |
| { |
| "name": "options_cast_fp16", |
| "type": "ios17.reshape", |
| "device": "ane" |
| }, |
| { |
| "name": "input_cast_fp16", |
| "type": "ios17.layer_norm", |
| "device": "ane" |
| }, |
| { |
| "name": "input_53_cast_fp16", |
| "type": "ios17.layer_norm", |
| "device": "ane" |
| }, |
| { |
| "name": "linear_12_cast_fp16", |
| "type": "ios17.linear", |
| "device": "ane" |
| }, |
| { |
| "name": "linear_13_cast_fp16", |
| "type": "ios17.linear", |
| "device": "ane" |
| }, |
| { |
| "name": "linear_14_cast_fp16", |
| "type": "ios17.linear", |
| "device": "ane" |
| }, |
| { |
| "name": "matmul_3_cast_fp16", |
| "type": "ios17.matmul", |
| "device": "ane" |
| }, |
| { |
| "name": "_inversed_scores_1_cast_fp16", |
| "type": "ios17.mul", |
| "device": "ane" |
| }, |
| { |
| "name": "var_567", |
| "type": "ios17.expand_dims", |
| "device": "ane" |
| }, |
| { |
| "name": "var_569", |
| "type": "ios17.logical_not", |
| "device": "ane" |
| }, |
| { |
| "name": "scores_cast_fp16", |
| "type": "select", |
| "device": "ane" |
| }, |
| { |
| "name": "var_571_cast_fp16", |
| "type": "ios16.softmax", |
| "device": "ane" |
| }, |
| { |
| "name": "matmul_4_cast_fp16", |
| "type": "ios17.matmul", |
| "device": "ane" |
| }, |
| { |
| "name": "var_574_cast_fp16", |
| "type": "ios17.mul", |
| "device": "ane" |
| }, |
| { |
| "name": "var_576_cast_fp16", |
| "type": "ios16.reduce_sum", |
| "device": "ane" |
| }, |
| { |
| "name": "_inversed_logits_1_cast_fp16", |
| "type": "ios17.mul", |
| "device": "ane" |
| }, |
| { |
| "name": "var_579", |
| "type": "ios17.logical_not", |
| "device": "ane" |
| }, |
| { |
| "name": "logits_3_cast_fp16", |
| "type": "select", |
| "device": "ane" |
| }, |
| { |
| "name": "var_587_cast_fp16", |
| "type": "ios16.fill_like", |
| "device": "ane" |
| }, |
| { |
| "name": "logits_cast_fp16", |
| "type": "select", |
| "device": "ane" |
| }, |
| { |
| "name": "var_591_cast_fp16", |
| "type": "ios16.softmax", |
| "device": "ane" |
| }, |
| { |
| "name": "probabilities", |
| "type": "ios17.cast", |
| "device": "ane" |
| }, |
| { |
| "name": "logits", |
| "type": "ios17.cast", |
| "device": "ane" |
| } |
| ] |
| } |
| }, |
| "ALL": { |
| "load_ms": 115.82287499913946, |
| "first_prediction_ms": 17.531749996123835, |
| "warm_prediction": { |
| "count": 30, |
| "median_ms": 0.9119170135818422, |
| "p95_ms": 1.2285901117138565, |
| "min_ms": 0.8736249874345958, |
| "max_ms": 1.7777080065570772 |
| }, |
| "timed_predictions_correct": 30, |
| "calls": [ |
| { |
| "row": 0, |
| "milliseconds": 1.7777080065570772 |
| }, |
| { |
| "row": 68, |
| "milliseconds": 1.2005419994238764 |
| }, |
| { |
| "row": 130, |
| "milliseconds": 0.9608750115148723 |
| }, |
| { |
| "row": 0, |
| "milliseconds": 1.0891249985434115 |
| }, |
| { |
| "row": 68, |
| "milliseconds": 1.2131670082453638 |
| }, |
| { |
| "row": 130, |
| "milliseconds": 1.241209014551714 |
| }, |
| { |
| "row": 0, |
| "milliseconds": 1.1116660025436431 |
| }, |
| { |
| "row": 68, |
| "milliseconds": 1.179958024295047 |
| }, |
| { |
| "row": 130, |
| "milliseconds": 0.8888749871402979 |
| }, |
| { |
| "row": 0, |
| "milliseconds": 0.8894159982446581 |
| }, |
| { |
| "row": 68, |
| "milliseconds": 0.9866250038612634 |
| }, |
| { |
| "row": 130, |
| "milliseconds": 0.9067920036613941 |
| }, |
| { |
| "row": 0, |
| "milliseconds": 0.8964589796960354 |
| }, |
| { |
| "row": 68, |
| "milliseconds": 0.901333027286455 |
| }, |
| { |
| "row": 130, |
| "milliseconds": 0.9048339852597564 |
| }, |
| { |
| "row": 0, |
| "milliseconds": 0.9170420235022902 |
| }, |
| { |
| "row": 68, |
| "milliseconds": 0.8820410002954304 |
| }, |
| { |
| "row": 130, |
| "milliseconds": 0.8736249874345958 |
| }, |
| { |
| "row": 0, |
| "milliseconds": 0.8920000109355897 |
| }, |
| { |
| "row": 68, |
| "milliseconds": 0.8972920186351985 |
| }, |
| { |
| "row": 130, |
| "milliseconds": 0.888916983967647 |
| }, |
| { |
| "row": 0, |
| "milliseconds": 0.8755000017117709 |
| }, |
| { |
| "row": 68, |
| "milliseconds": 0.9061669989023358 |
| }, |
| { |
| "row": 130, |
| "milliseconds": 1.0763329919427633 |
| }, |
| { |
| "row": 0, |
| "milliseconds": 0.9489590011071414 |
| }, |
| { |
| "row": 68, |
| "milliseconds": 0.9267920104321092 |
| }, |
| { |
| "row": 130, |
| "milliseconds": 0.9253750031348318 |
| }, |
| { |
| "row": 0, |
| "milliseconds": 0.9227920090779662 |
| }, |
| { |
| "row": 68, |
| "milliseconds": 0.8931249903980643 |
| }, |
| { |
| "row": 130, |
| "milliseconds": 0.8973750227596611 |
| } |
| ], |
| "compute_plan": { |
| "kind": "MLComputePlan preferred device; not a runtime utilization or power trace", |
| "assigned_operation_count": 173, |
| "device_counts": { |
| "cpu": 0, |
| "gpu": 173, |
| "ane": 0, |
| "unknown": 0 |
| }, |
| "unassigned_operation_types": { |
| "const": 239 |
| }, |
| "operations": [ |
| { |
| "name": "option_mask_1", |
| "type": "ios17.not_equal", |
| "device": "gpu" |
| }, |
| { |
| "name": "context_mask", |
| "type": "ios17.not_equal", |
| "device": "gpu" |
| }, |
| { |
| "name": "var_27", |
| "type": "ios17.slice_by_index", |
| "device": "gpu" |
| }, |
| { |
| "name": "fill_like_0_cast_fp16", |
| "type": "ios16.fill_like", |
| "device": "gpu" |
| }, |
| { |
| "name": "var_43", |
| "type": "ios17.slice_by_index", |
| "device": "gpu" |
| }, |
| { |
| "name": "var_43_to_fp16", |
| "type": "ios17.cast", |
| "device": "gpu" |
| }, |
| { |
| "name": "safe_context_ids_cast_fp16", |
| "type": "ios17.concat", |
| "device": "gpu" |
| }, |
| { |
| "name": "context_ids_to_int16", |
| "type": "ios17.cast", |
| "device": "gpu" |
| }, |
| { |
| "name": "cast_54", |
| "type": "ios17.cast", |
| "device": "gpu" |
| }, |
| { |
| "name": "greater_equal_0", |
| "type": "ios17.greater_equal", |
| "device": "gpu" |
| }, |
| { |
| "name": "add_3", |
| "type": "ios17.add", |
| "device": "gpu" |
| }, |
| { |
| "name": "select_0", |
| "type": "select", |
| "device": "gpu" |
| }, |
| { |
| "name": "select_0_to_int16", |
| "type": "ios17.cast", |
| "device": "gpu" |
| }, |
| { |
| "name": "var_59_cast_fp16_cast_uint16_cast_uint16", |
| "type": "ios17.gather", |
| "device": "gpu" |
| }, |
| { |
| "name": "src_1_cast_fp16", |
| "type": "ios17.add", |
| "device": "gpu" |
| }, |
| { |
| "name": "mask_1_cast_fp16", |
| "type": "ios17.equal", |
| "device": "gpu" |
| }, |
| { |
| "name": "key_padding_mask_1_cast_fp16", |
| "type": "select", |
| "device": "gpu" |
| }, |
| { |
| "name": "query_1_cast_fp16", |
| "type": "ios17.layer_norm", |
| "device": "gpu" |
| }, |
| { |
| "name": "query_3_cast_fp16", |
| "type": "ios17.transpose", |
| "device": "gpu" |
| }, |
| { |
| "name": "linear_0_cast_fp16", |
| "type": "ios17.linear", |
| "device": "gpu" |
| }, |
| { |
| "name": "var_124_cast_fp16", |
| "type": "ios17.reshape", |
| "device": "gpu" |
| }, |
| { |
| "name": "var_125_cast_fp16", |
| "type": "ios17.expand_dims", |
| "device": "gpu" |
| }, |
| { |
| "name": "var_126_cast_fp16", |
| "type": "ios17.transpose", |
| "device": "gpu" |
| }, |
| { |
| "name": "var_127_cast_fp16", |
| "type": "ios17.squeeze", |
| "device": "gpu" |
| }, |
| { |
| "name": "q_1_cast_fp16", |
| "type": "ios17.slice_by_index", |
| "device": "gpu" |
| }, |
| { |
| "name": "k_1_cast_fp16", |
| "type": "ios17.slice_by_index", |
| "device": "gpu" |
| }, |
| { |
| "name": "v_1_cast_fp16", |
| "type": "ios17.slice_by_index", |
| "device": "gpu" |
| }, |
| { |
| "name": "var_136_cast_fp16", |
| "type": "ios17.reshape", |
| "device": "gpu" |
| }, |
| { |
| "name": "var_143_cast_fp16", |
| "type": "ios17.reshape", |
| "device": "gpu" |
| }, |
| { |
| "name": "var_150_cast_fp16", |
| "type": "ios17.reshape", |
| "device": "gpu" |
| }, |
| { |
| "name": "var_154_cast_fp16", |
| "type": "ios17.reshape", |
| "device": "gpu" |
| }, |
| { |
| "name": "var_156_cast_fp16", |
| "type": "tile", |
| "device": "gpu" |
| }, |
| { |
| "name": "q_3_cast_fp16", |
| "type": "ios17.transpose", |
| "device": "gpu" |
| }, |
| { |
| "name": "q_5_cast_fp16", |
| "type": "ios17.reshape", |
| "device": "gpu" |
| }, |
| { |
| "name": "k_3_cast_fp16", |
| "type": "ios17.transpose", |
| "device": "gpu" |
| }, |
| { |
| "name": "k_5_cast_fp16", |
| "type": "ios17.reshape", |
| "device": "gpu" |
| }, |
| { |
| "name": "v_3_cast_fp16", |
| "type": "ios17.transpose", |
| "device": "gpu" |
| }, |
| { |
| "name": "v_5_cast_fp16", |
| "type": "ios17.reshape", |
| "device": "gpu" |
| }, |
| { |
| "name": "mul_1_cast_fp16", |
| "type": "ios17.mul", |
| "device": "gpu" |
| }, |
| { |
| "name": "matmul_0_cast_fp16", |
| "type": "ios17.matmul", |
| "device": "gpu" |
| }, |
| { |
| "name": "add_0_cast_fp16", |
| "type": "ios17.add", |
| "device": "gpu" |
| }, |
| { |
| "name": "softmax_0_cast_fp16", |
| "type": "ios16.softmax", |
| "device": "gpu" |
| }, |
| { |
| "name": "attn_output_1_cast_fp16", |
| "type": "ios17.matmul", |
| "device": "gpu" |
| }, |
| { |
| "name": "var_172_cast_fp16", |
| "type": "ios17.transpose", |
| "device": "gpu" |
| }, |
| { |
| "name": "attn_output_3_cast_fp16", |
| "type": "ios17.reshape", |
| "device": "gpu" |
| }, |
| { |
| "name": "linear_1_cast_fp16", |
| "type": "ios17.linear", |
| "device": "gpu" |
| }, |
| { |
| "name": "attn_output_7_cast_fp16", |
| "type": "ios17.reshape", |
| "device": "gpu" |
| }, |
| { |
| "name": "input_3_cast_fp16", |
| "type": "ios17.transpose", |
| "device": "gpu" |
| }, |
| { |
| "name": "input_5_cast_fp16", |
| "type": "ios17.add", |
| "device": "gpu" |
| }, |
| { |
| "name": "input_7_cast_fp16", |
| "type": "ios17.layer_norm", |
| "device": "gpu" |
| }, |
| { |
| "name": "linear_2_cast_fp16", |
| "type": "ios17.linear", |
| "device": "gpu" |
| }, |
| { |
| "name": "input_11_cast_fp16", |
| "type": "ios16.relu", |
| "device": "gpu" |
| }, |
| { |
| "name": "linear_3_cast_fp16", |
| "type": "ios17.linear", |
| "device": "gpu" |
| }, |
| { |
| "name": "input_17_cast_fp16", |
| "type": "ios17.add", |
| "device": "gpu" |
| }, |
| { |
| "name": "query_5_cast_fp16", |
| "type": "ios17.layer_norm", |
| "device": "gpu" |
| }, |
| { |
| "name": "query_7_cast_fp16", |
| "type": "ios17.transpose", |
| "device": "gpu" |
| }, |
| { |
| "name": "linear_4_cast_fp16", |
| "type": "ios17.linear", |
| "device": "gpu" |
| }, |
| { |
| "name": "var_231_cast_fp16", |
| "type": "ios17.reshape", |
| "device": "gpu" |
| }, |
| { |
| "name": "var_232_cast_fp16", |
| "type": "ios17.expand_dims", |
| "device": "gpu" |
| }, |
| { |
| "name": "var_233_cast_fp16", |
| "type": "ios17.transpose", |
| "device": "gpu" |
| }, |
| { |
| "name": "var_234_cast_fp16", |
| "type": "ios17.squeeze", |
| "device": "gpu" |
| }, |
| { |
| "name": "q_7_cast_fp16", |
| "type": "ios17.slice_by_index", |
| "device": "gpu" |
| }, |
| { |
| "name": "k_7_cast_fp16", |
| "type": "ios17.slice_by_index", |
| "device": "gpu" |
| }, |
| { |
| "name": "v_7_cast_fp16", |
| "type": "ios17.slice_by_index", |
| "device": "gpu" |
| }, |
| { |
| "name": "var_243_cast_fp16", |
| "type": "ios17.reshape", |
| "device": "gpu" |
| }, |
| { |
| "name": "var_250_cast_fp16", |
| "type": "ios17.reshape", |
| "device": "gpu" |
| }, |
| { |
| "name": "var_257_cast_fp16", |
| "type": "ios17.reshape", |
| "device": "gpu" |
| }, |
| { |
| "name": "q_9_cast_fp16", |
| "type": "ios17.transpose", |
| "device": "gpu" |
| }, |
| { |
| "name": "q_11_cast_fp16", |
| "type": "ios17.reshape", |
| "device": "gpu" |
| }, |
| { |
| "name": "k_9_cast_fp16", |
| "type": "ios17.transpose", |
| "device": "gpu" |
| }, |
| { |
| "name": "k_11_cast_fp16", |
| "type": "ios17.reshape", |
| "device": "gpu" |
| }, |
| { |
| "name": "v_9_cast_fp16", |
| "type": "ios17.transpose", |
| "device": "gpu" |
| }, |
| { |
| "name": "v_11_cast_fp16", |
| "type": "ios17.reshape", |
| "device": "gpu" |
| }, |
| { |
| "name": "mul_3_cast_fp16", |
| "type": "ios17.mul", |
| "device": "gpu" |
| }, |
| { |
| "name": "matmul_1_cast_fp16", |
| "type": "ios17.matmul", |
| "device": "gpu" |
| }, |
| { |
| "name": "add_1_cast_fp16", |
| "type": "ios17.add", |
| "device": "gpu" |
| }, |
| { |
| "name": "softmax_1_cast_fp16", |
| "type": "ios16.softmax", |
| "device": "gpu" |
| }, |
| { |
| "name": "attn_output_9_cast_fp16", |
| "type": "ios17.matmul", |
| "device": "gpu" |
| }, |
| { |
| "name": "var_279_cast_fp16", |
| "type": "ios17.transpose", |
| "device": "gpu" |
| }, |
| { |
| "name": "attn_output_11_cast_fp16", |
| "type": "ios17.reshape", |
| "device": "gpu" |
| }, |
| { |
| "name": "linear_5_cast_fp16", |
| "type": "ios17.linear", |
| "device": "gpu" |
| }, |
| { |
| "name": "attn_output_15_cast_fp16", |
| "type": "ios17.reshape", |
| "device": "gpu" |
| }, |
| { |
| "name": "input_19_cast_fp16", |
| "type": "ios17.transpose", |
| "device": "gpu" |
| }, |
| { |
| "name": "input_21_cast_fp16", |
| "type": "ios17.add", |
| "device": "gpu" |
| }, |
| { |
| "name": "input_23_cast_fp16", |
| "type": "ios17.layer_norm", |
| "device": "gpu" |
| }, |
| { |
| "name": "linear_6_cast_fp16", |
| "type": "ios17.linear", |
| "device": "gpu" |
| }, |
| { |
| "name": "input_27_cast_fp16", |
| "type": "ios16.relu", |
| "device": "gpu" |
| }, |
| { |
| "name": "linear_7_cast_fp16", |
| "type": "ios17.linear", |
| "device": "gpu" |
| }, |
| { |
| "name": "context_cast_fp16", |
| "type": "ios17.add", |
| "device": "gpu" |
| }, |
| { |
| "name": "flat_ids", |
| "type": "ios17.reshape", |
| "device": "gpu" |
| }, |
| { |
| "name": "flat_mask", |
| "type": "ios17.not_equal", |
| "device": "gpu" |
| }, |
| { |
| "name": "var_333", |
| "type": "ios17.slice_by_index", |
| "device": "gpu" |
| }, |
| { |
| "name": "fill_like_1_cast_fp16", |
| "type": "ios16.fill_like", |
| "device": "gpu" |
| }, |
| { |
| "name": "var_349", |
| "type": "ios17.slice_by_index", |
| "device": "gpu" |
| }, |
| { |
| "name": "var_349_to_fp16", |
| "type": "ios17.cast", |
| "device": "gpu" |
| }, |
| { |
| "name": "safe_ids_cast_fp16", |
| "type": "ios17.concat", |
| "device": "gpu" |
| }, |
| { |
| "name": "flat_ids_to_int16", |
| "type": "ios17.cast", |
| "device": "gpu" |
| }, |
| { |
| "name": "cast_55", |
| "type": "ios17.cast", |
| "device": "gpu" |
| }, |
| { |
| "name": "greater_equal_1", |
| "type": "ios17.greater_equal", |
| "device": "gpu" |
| }, |
| { |
| "name": "add_4", |
| "type": "ios17.add", |
| "device": "gpu" |
| }, |
| { |
| "name": "select_1", |
| "type": "select", |
| "device": "gpu" |
| }, |
| { |
| "name": "select_1_to_int16", |
| "type": "ios17.cast", |
| "device": "gpu" |
| }, |
| { |
| "name": "var_365_cast_fp16_cast_uint16_cast_uint16", |
| "type": "ios17.gather", |
| "device": "gpu" |
| }, |
| { |
| "name": "src_cast_fp16", |
| "type": "ios17.add", |
| "device": "gpu" |
| }, |
| { |
| "name": "mask_cast_fp16", |
| "type": "ios17.equal", |
| "device": "gpu" |
| }, |
| { |
| "name": "key_padding_mask_7_cast_fp16", |
| "type": "select", |
| "device": "gpu" |
| }, |
| { |
| "name": "query_9_cast_fp16", |
| "type": "ios17.layer_norm", |
| "device": "gpu" |
| }, |
| { |
| "name": "query_11_cast_fp16", |
| "type": "ios17.transpose", |
| "device": "gpu" |
| }, |
| { |
| "name": "linear_8_cast_fp16", |
| "type": "ios17.linear", |
| "device": "gpu" |
| }, |
| { |
| "name": "var_428_cast_fp16", |
| "type": "ios17.reshape", |
| "device": "gpu" |
| }, |
| { |
| "name": "var_429_cast_fp16", |
| "type": "ios17.expand_dims", |
| "device": "gpu" |
| }, |
| { |
| "name": "var_430_cast_fp16", |
| "type": "ios17.transpose", |
| "device": "gpu" |
| }, |
| { |
| "name": "var_431_cast_fp16", |
| "type": "ios17.squeeze", |
| "device": "gpu" |
| }, |
| { |
| "name": "q_13_cast_fp16", |
| "type": "ios17.slice_by_index", |
| "device": "gpu" |
| }, |
| { |
| "name": "k_13_cast_fp16", |
| "type": "ios17.slice_by_index", |
| "device": "gpu" |
| }, |
| { |
| "name": "v_13_cast_fp16", |
| "type": "ios17.slice_by_index", |
| "device": "gpu" |
| }, |
| { |
| "name": "var_440_cast_fp16", |
| "type": "ios17.reshape", |
| "device": "gpu" |
| }, |
| { |
| "name": "var_447_cast_fp16", |
| "type": "ios17.reshape", |
| "device": "gpu" |
| }, |
| { |
| "name": "var_454_cast_fp16", |
| "type": "ios17.reshape", |
| "device": "gpu" |
| }, |
| { |
| "name": "var_458_cast_fp16", |
| "type": "ios17.reshape", |
| "device": "gpu" |
| }, |
| { |
| "name": "var_460_cast_fp16", |
| "type": "tile", |
| "device": "gpu" |
| }, |
| { |
| "name": "q_15_cast_fp16", |
| "type": "ios17.transpose", |
| "device": "gpu" |
| }, |
| { |
| "name": "q_cast_fp16", |
| "type": "ios17.reshape", |
| "device": "gpu" |
| }, |
| { |
| "name": "k_15_cast_fp16", |
| "type": "ios17.transpose", |
| "device": "gpu" |
| }, |
| { |
| "name": "k_cast_fp16", |
| "type": "ios17.reshape", |
| "device": "gpu" |
| }, |
| { |
| "name": "v_15_cast_fp16", |
| "type": "ios17.transpose", |
| "device": "gpu" |
| }, |
| { |
| "name": "v_cast_fp16", |
| "type": "ios17.reshape", |
| "device": "gpu" |
| }, |
| { |
| "name": "mul_5_cast_fp16", |
| "type": "ios17.mul", |
| "device": "gpu" |
| }, |
| { |
| "name": "matmul_2_cast_fp16", |
| "type": "ios17.matmul", |
| "device": "gpu" |
| }, |
| { |
| "name": "add_2_cast_fp16", |
| "type": "ios17.add", |
| "device": "gpu" |
| }, |
| { |
| "name": "softmax_2_cast_fp16", |
| "type": "ios16.softmax", |
| "device": "gpu" |
| }, |
| { |
| "name": "attn_output_17_cast_fp16", |
| "type": "ios17.matmul", |
| "device": "gpu" |
| }, |
| { |
| "name": "var_476_cast_fp16", |
| "type": "ios17.transpose", |
| "device": "gpu" |
| }, |
| { |
| "name": "attn_output_19_cast_fp16", |
| "type": "ios17.reshape", |
| "device": "gpu" |
| }, |
| { |
| "name": "linear_9_cast_fp16", |
| "type": "ios17.linear", |
| "device": "gpu" |
| }, |
| { |
| "name": "attn_output_cast_fp16", |
| "type": "ios17.reshape", |
| "device": "gpu" |
| }, |
| { |
| "name": "input_35_cast_fp16", |
| "type": "ios17.transpose", |
| "device": "gpu" |
| }, |
| { |
| "name": "input_37_cast_fp16", |
| "type": "ios17.add", |
| "device": "gpu" |
| }, |
| { |
| "name": "input_39_cast_fp16", |
| "type": "ios17.layer_norm", |
| "device": "gpu" |
| }, |
| { |
| "name": "linear_10_cast_fp16", |
| "type": "ios17.linear", |
| "device": "gpu" |
| }, |
| { |
| "name": "input_43_cast_fp16", |
| "type": "ios16.relu", |
| "device": "gpu" |
| }, |
| { |
| "name": "linear_11_cast_fp16", |
| "type": "ios17.linear", |
| "device": "gpu" |
| }, |
| { |
| "name": "hidden_cast_fp16", |
| "type": "ios17.add", |
| "device": "gpu" |
| }, |
| { |
| "name": "var_504", |
| "type": "ios17.expand_dims", |
| "device": "gpu" |
| }, |
| { |
| "name": "var_504_to_fp16", |
| "type": "ios17.cast", |
| "device": "gpu" |
| }, |
| { |
| "name": "var_510_cast_fp16", |
| "type": "ios17.mul", |
| "device": "gpu" |
| }, |
| { |
| "name": "var_515_cast_fp16", |
| "type": "ios16.reduce_sum", |
| "device": "gpu" |
| }, |
| { |
| "name": "var_520_cast_fp16", |
| "type": "ios16.reduce_sum", |
| "device": "gpu" |
| }, |
| { |
| "name": "var_522_cast_fp16", |
| "type": "ios17.maximum", |
| "device": "gpu" |
| }, |
| { |
| "name": "pooled_cast_fp16", |
| "type": "ios17.real_div", |
| "device": "gpu" |
| }, |
| { |
| "name": "options_cast_fp16", |
| "type": "ios17.reshape", |
| "device": "gpu" |
| }, |
| { |
| "name": "input_cast_fp16", |
| "type": "ios17.layer_norm", |
| "device": "gpu" |
| }, |
| { |
| "name": "input_53_cast_fp16", |
| "type": "ios17.layer_norm", |
| "device": "gpu" |
| }, |
| { |
| "name": "linear_12_cast_fp16", |
| "type": "ios17.linear", |
| "device": "gpu" |
| }, |
| { |
| "name": "linear_13_cast_fp16", |
| "type": "ios17.linear", |
| "device": "gpu" |
| }, |
| { |
| "name": "linear_14_cast_fp16", |
| "type": "ios17.linear", |
| "device": "gpu" |
| }, |
| { |
| "name": "matmul_3_cast_fp16", |
| "type": "ios17.matmul", |
| "device": "gpu" |
| }, |
| { |
| "name": "_inversed_scores_1_cast_fp16", |
| "type": "ios17.mul", |
| "device": "gpu" |
| }, |
| { |
| "name": "var_567", |
| "type": "ios17.expand_dims", |
| "device": "gpu" |
| }, |
| { |
| "name": "var_569", |
| "type": "ios17.logical_not", |
| "device": "gpu" |
| }, |
| { |
| "name": "scores_cast_fp16", |
| "type": "select", |
| "device": "gpu" |
| }, |
| { |
| "name": "var_571_cast_fp16", |
| "type": "ios16.softmax", |
| "device": "gpu" |
| }, |
| { |
| "name": "matmul_4_cast_fp16", |
| "type": "ios17.matmul", |
| "device": "gpu" |
| }, |
| { |
| "name": "var_574_cast_fp16", |
| "type": "ios17.mul", |
| "device": "gpu" |
| }, |
| { |
| "name": "var_576_cast_fp16", |
| "type": "ios16.reduce_sum", |
| "device": "gpu" |
| }, |
| { |
| "name": "_inversed_logits_1_cast_fp16", |
| "type": "ios17.mul", |
| "device": "gpu" |
| }, |
| { |
| "name": "var_579", |
| "type": "ios17.logical_not", |
| "device": "gpu" |
| }, |
| { |
| "name": "logits_3_cast_fp16", |
| "type": "select", |
| "device": "gpu" |
| }, |
| { |
| "name": "var_587_cast_fp16", |
| "type": "ios16.fill_like", |
| "device": "gpu" |
| }, |
| { |
| "name": "logits_cast_fp16", |
| "type": "select", |
| "device": "gpu" |
| }, |
| { |
| "name": "var_591_cast_fp16", |
| "type": "ios16.softmax", |
| "device": "gpu" |
| }, |
| { |
| "name": "probabilities", |
| "type": "ios17.cast", |
| "device": "gpu" |
| }, |
| { |
| "name": "logits", |
| "type": "ios17.cast", |
| "device": "gpu" |
| } |
| ] |
| } |
| } |
| }, |
| "script_sha256": "d5e1386f54fc6c482647120e593e7ffbbe4bbb277c2cc6d4949154660fe239fd" |
| } |
|
|