OmniParser v2 Core ML bundle (icon_detect_v3 + Florence-2 captioner)

README.md

@@ -0,0 +1,125 @@
+---
+license: mit
+library_name: coreml
+pipeline_tag: object-detection
+tags:
+- coreml
+- apple-neural-engine
+- screen-parsing
+- gui-agent
+- ui-element-detection
+- icon-captioning
+- florence-2
+- yolo
+base_model:
+- microsoft/OmniParser-v2.0
+- microsoft/Florence-2-base
+base_model_relation: quantized
+---
+
+# OmniParser v2 for Core ML
+
+Screen parsing for macOS on the Apple Neural Engine. Given a screenshot, the detector finds interactive
+elements (buttons, icons, fields) and the captioner describes what each one does, so an agent can refer to
+and act on the screen by element instead of by pixel.
+
+This is Microsoft's OmniParser v2 converted to Core ML:
+
+- **Detector:** `icon_detect_v3` (YOLOv9-E), the MIT-licensed detector checkpoint.
+- **Captioner:** the OmniParser v2 fine-tune of Florence-2-base, split into an encoder and a decoder.
+
+Every executable op in all three graphs is planned for the Neural Engine (zero CPU or GPU fallbacks).
+
+## Requirements
+
+- macOS 15 or later on Apple Silicon
+- Core ML must be able to write its user cache; the first load compiles the packages
+
+## Download
+
+```bash
+hf download 1of2/omniparser-v2-coreml --local-dir omniparser-v2-coreml
+cd omniparser-v2-coreml && shasum -a 256 -c SHA256SUMS
+```
+
+## Contents
+
+| Path | Purpose | Size |
+| --- | --- | --- |
+| `models/detector/IconDetectV3.mlpackage` | element detector, two functions | 117 MB |
+| `models/caption/FlorenceEncoderContextEmbeds.mlpackage` | caption encoder (DaViT vision + BART encoder) | 270 MB |
+| `models/caption/FlorenceDecoderPrefixEmbeds.mlpackage` | caption decoder | 192 MB |
+| `models/caption/shared_embedding_fp16.npy` | token embedding table [51289, 768] fp16; memory-map read-only | 79 MB |
+| `models/caption/tokenizer/` | BART tokenizer and Florence-2 processor config | 2.5 MB |
+| `runtime_config.json` | every shape, threshold and generation setting below, machine-readable | |
+| `models/*/LICENSE`, `models/detector/provenance.json` | licences and detector source hash | |
+| `SHA256SUMS` | checksums for every file | |
+
+## Detector
+
+`IconDetectV3.mlpackage` is a multifunction package:
+
+| Function | Input `image` | Output `predictions` |
+| --- | --- | --- |
+| `detect_640` (default) | fp16 [1, 3, 640, 640] | [1, 5, 8400] |
+| `detect_1280` | fp16 [1, 3, 1280, 1280] | [1, 5, 33600] |
+
+Preprocessing: RGB in 0..1, letterboxed to a centred square with pad value 114, resampled with Lanczos.
+Each anchor carries one class score and a box in grid units (`[1, 5, anchors]`).
+
+Postprocessing: confidence threshold 0.05, non-maximum suppression at IoU 0.1, at most 300 detections,
+then boxes overlapping by more than IoU 0.7 are merged so each element appears once. Boxes map back to the
+original image by undoing the letterbox.
+
+Use `detect_1280` for dense or high-resolution screens; `detect_640` is faster and enough for most windows.
+
+## Captioner
+
+Each detected element is cropped, resized to 64x64, then processed like Florence-2-base input: resized to
+768x768, rescaled by 1/255 and normalised with ImageNet mean `[0.485, 0.456, 0.406]` and std
+`[0.229, 0.224, 0.225]`.
+
+| Graph | Inputs | Output |
+| --- | --- | --- |
+| Encoder | `pixel_values` fp16 [1, 3, 768, 768], `prompt_embeds` fp16 [1, 8, 768] | `encoder_hidden_states` [1, 585, 768] |
+| Decoder | `decoder_inputs_embeds` fp16 [1, 20, 768], `encoder_hidden_states` [1, 585, 768] | `logits` [1, 20, 51289] |
+
+The prompt is `<CAPTION>`, tokenised to `[0, 2264, 473, 5, 2274, 6190, 116, 2]`; look its ids up in
+`shared_embedding_fp16.npy` to form `prompt_embeds`. Run the encoder once per crop.
+
+Generation is greedy over a fixed 20-token decoder prefix: start token 2, forced BOS 0, EOS and forced EOS 2,
+pad 1, no repeated 3-grams, at most 20 new tokens. At each step embed the tokens so far into the padded
+prefix, run the decoder, and take the argmax at the current position after applying those constraints.
+Decode with special tokens skipped and whitespace trimmed.
+
+## Faithfulness
+
+The conversion targets parity with the upstream PyTorch pipeline at three points: the detector's input
+pixels, the decoded boxes, and the caption text. fp16 weights and activations can still move a borderline
+detection or change a caption token in rare cases.
+
+## Limitations
+
+- Captions are short functional descriptions ("settings", "close window"), not full sentences.
+- Text on screen is not read by these models; pair them with an OCR engine for labels and content.
+- Trained mostly on desktop and web UIs; unusual custom-drawn interfaces detect less reliably.
+
+## License and attribution
+
+Released under the MIT License.
+
+- Detector: [`microsoft/OmniParser-v2.0`](https://huggingface.co/microsoft/OmniParser-v2.0)
+  `icon_detect_v3/model.pt` at revision `f55d0750e5b94db2125ef0b45b0fa4a85ddc59b4`
+  (sha256 `11c6cbb77f22569fab22d86c76407a83ec81ab89dbfe28279854822d6e3fb00c`), MIT. See `models/detector/LICENSE`.
+- Captioner: `microsoft/OmniParser-v2.0` `icon_caption` at the same revision, fine-tuned from
+  [`microsoft/Florence-2-base`](https://huggingface.co/microsoft/Florence-2-base), MIT. See `models/caption/LICENSE`.
+
+```bibtex
+@misc{lu2024omniparserpurevisionbased,
+  title={OmniParser for Pure Vision Based GUI Agent},
+  author={Yadong Lu and Jianwei Yang and Yelong Shen and Ahmed Awadallah},
+  year={2024},
+  eprint={2408.00203},
+  archivePrefix={arXiv},
+}
+```

SHA256SUMS

@@ -0,0 +1,18 @@
+d6c4232f4e6be156224aa7a14e060d7163df85c050758f22b2d78b76b2a66be5  ./models/caption/FlorenceDecoderPrefixEmbeds.mlpackage/Data/com.apple.CoreML/model.mlmodel
+a6df8e1ab247880d83d5647aca2994d251b62a6723e6f7986293b5e7787bc6d2  ./models/caption/FlorenceDecoderPrefixEmbeds.mlpackage/Data/com.apple.CoreML/weights/weight.bin
+2070b35b2e4780738b53201173ad493d72d3d5a4eae4d6e21c18ecf5645f0682  ./models/caption/FlorenceDecoderPrefixEmbeds.mlpackage/Manifest.json
+ca2d0d3086bfc983932089db379dc52257cbd3bb49f11170a7f41b2932129b8f  ./models/caption/FlorenceEncoderContextEmbeds.mlpackage/Data/com.apple.CoreML/model.mlmodel
+8d9669550b24005f5e9df03440c316a2f837fa062e9e6bc576e0c47ec7b2e7c3  ./models/caption/FlorenceEncoderContextEmbeds.mlpackage/Data/com.apple.CoreML/weights/weight.bin
+6fc454f70a3cea67fa53c466525cbdfd10fe148451454cf5a173f0a7d86e08e7  ./models/caption/FlorenceEncoderContextEmbeds.mlpackage/Manifest.json
+d9a1b1e30d633d5732ea18e3cba9538d293ebc53e1a9e4e96ab739e0c5c4f1cb  ./models/caption/LICENSE
+00ded3f65a0d242bcae37c22fd1badbd0eae07a912b603a895841aea962da3fc  ./models/caption/shared_embedding_fp16.npy
+2f5921bbc53c7cc04251e1027b45b1cec726276be6db23d1bb40641bfbe2cf29  ./models/caption/tokenizer/preprocessor_config.json
+847bbeab6174d66a88898f729d52fa8d355fafe1bea101cf960dd404581df70e  ./models/caption/tokenizer/tokenizer.json
+79ffcf43af8ebda99d165f61d243180da2e2639952e41e71e11611c18770489c  ./models/caption/tokenizer/tokenizer_config.json
+394fdc63c71aabe0a9b97117f5d62fb5fcc4d59b2b3ea929a3929e6a53217b3c  ./models/caption/tokenizer/vocab.json
+82f6d36d2188433034a17c434d24257ce8b40f863f839584fbe586b71148bf98  ./models/detector/IconDetectV3.mlpackage/Data/com.apple.CoreML/model.mlmodel
+745fab6261135d6300362219e771b973fc968a61f14e476cd58ab97dd834b0d7  ./models/detector/IconDetectV3.mlpackage/Data/com.apple.CoreML/weights/weight.bin
+0d1b1d324c247745e11d8f761a7ab4d612aeae60290e36f02c11c62e7ce8487b  ./models/detector/IconDetectV3.mlpackage/Manifest.json
+646f8936b8ddcd14e13e578ff6857e368780b0d1a4f6066bee89211923a373e2  ./models/detector/LICENSE
+215bd3b5da42496ac5fdd20cf8dcde9cd0ab266236ddfb267e066b9382fc6529  ./models/detector/provenance.json
+a4f3d58e5758a39d77b7d134ba8e750aa6df4cd9321bfe85ea23e41ff8dcffba  ./runtime_config.json

models/caption/FlorenceDecoderPrefixEmbeds.mlpackage/Data/com.apple.CoreML/model.mlmodel

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d6c4232f4e6be156224aa7a14e060d7163df85c050758f22b2d78b76b2a66be5
+size 126216

models/caption/FlorenceDecoderPrefixEmbeds.mlpackage/Data/com.apple.CoreML/weights/weight.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a6df8e1ab247880d83d5647aca2994d251b62a6723e6f7986293b5e7787bc6d2
+size 192348850

models/caption/FlorenceDecoderPrefixEmbeds.mlpackage/Manifest.json

@@ -0,0 +1,18 @@
+{
+    "fileFormatVersion": "1.0.0",
+    "itemInfoEntries": {
+        "AF61D9FD-1C1E-4F8E-BC3E-E79592088676": {
+            "author": "com.apple.CoreML",
+            "description": "CoreML Model Weights",
+            "name": "weights",
+            "path": "com.apple.CoreML/weights"
+        },
+        "BB144488-CE48-4528-9018-061781F782CD": {
+            "author": "com.apple.CoreML",
+            "description": "CoreML Model Specification",
+            "name": "model.mlmodel",
+            "path": "com.apple.CoreML/model.mlmodel"
+        }
+    },
+    "rootModelIdentifier": "BB144488-CE48-4528-9018-061781F782CD"
+}

models/caption/FlorenceEncoderContextEmbeds.mlpackage/Data/com.apple.CoreML/model.mlmodel

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ca2d0d3086bfc983932089db379dc52257cbd3bb49f11170a7f41b2932129b8f
+size 506345

models/caption/FlorenceEncoderContextEmbeds.mlpackage/Data/com.apple.CoreML/weights/weight.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8d9669550b24005f5e9df03440c316a2f837fa062e9e6bc576e0c47ec7b2e7c3
+size 269483648

models/caption/FlorenceEncoderContextEmbeds.mlpackage/Manifest.json

@@ -0,0 +1,18 @@
+{
+    "fileFormatVersion": "1.0.0",
+    "itemInfoEntries": {
+        "0665D1E6-F427-4A7A-9480-0403FF991702": {
+            "author": "com.apple.CoreML",
+            "description": "CoreML Model Specification",
+            "name": "model.mlmodel",
+            "path": "com.apple.CoreML/model.mlmodel"
+        },
+        "797EC9EA-09AE-4E04-B426-768CC64A1561": {
+            "author": "com.apple.CoreML",
+            "description": "CoreML Model Weights",
+            "name": "weights",
+            "path": "com.apple.CoreML/weights"
+        }
+    },
+    "rootModelIdentifier": "0665D1E6-F427-4A7A-9480-0403FF991702"
+}

models/caption/LICENSE

@@ -0,0 +1,21 @@
+    MIT License
+
+    Copyright (c) Microsoft Corporation.
+
+    Permission is hereby granted, free of charge, to any person obtaining a copy
+    of this software and associated documentation files (the "Software"), to deal
+    in the Software without restriction, including without limitation the rights
+    to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+    copies of the Software, and to permit persons to whom the Software is
+    furnished to do so, subject to the following conditions:
+
+    The above copyright notice and this permission notice shall be included in all
+    copies or substantial portions of the Software.
+
+    THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+    IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+    FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+    AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+    LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+    OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+    SOFTWARE

models/caption/shared_embedding_fp16.npy

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:00ded3f65a0d242bcae37c22fd1badbd0eae07a912b603a895841aea962da3fc
+size 78780032

models/caption/tokenizer/preprocessor_config.json

@@ -0,0 +1,39 @@
+{
+  "auto_map": {
+    "AutoProcessor": "processing_florence2.Florence2Processor"
+   },
+  "_valid_processor_keys": [
+    "images",
+    "do_resize",
+    "size",
+    "resample",
+    "do_rescale",
+    "rescale_factor",
+    "do_normalize",
+    "image_mean",
+    "image_std",
+    "return_tensors",
+    "data_format",
+    "input_data_format",
+    "do_convert_rgb"
+  ],
+  "do_convert_rgb": null,
+  "do_normalize": true,
+  "do_rescale": true,
+  "do_resize": true,
+  "do_center_crop": false,
+  "image_processor_type": "CLIPImageProcessor",
+  "image_seq_length": 577,
+  "image_mean": [0.485, 0.456, 0.406],
+  "image_std":  [0.229, 0.224, 0.225],
+  "processor_class": "Florence2Processor",
+  "resample": 3,
+  "size": {
+    "height": 768,
+    "width":768 
+  },
+  "crop_size": {
+    "height": 768,
+    "width": 768
+  }
+}

models/caption/tokenizer/tokenizer_config.json

@@ -0,0 +1,4 @@
+{
+    "model_max_length": 1024
+}
+

models/detector/IconDetectV3.mlpackage/Data/com.apple.CoreML/model.mlmodel

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:82f6d36d2188433034a17c434d24257ce8b40f863f839584fbe586b71148bf98
+size 1031126

models/detector/IconDetectV3.mlpackage/Data/com.apple.CoreML/weights/weight.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:745fab6261135d6300362219e771b973fc968a61f14e476cd58ab97dd834b0d7
+size 116045184

models/detector/IconDetectV3.mlpackage/Manifest.json

@@ -0,0 +1,18 @@
+{
+    "fileFormatVersion": "1.0.0",
+    "itemInfoEntries": {
+        "60BDEF3F-0828-435D-91C0-798999E28E7F": {
+            "author": "com.apple.CoreML",
+            "description": "CoreML Model Specification",
+            "name": "model.mlmodel",
+            "path": "com.apple.CoreML/model.mlmodel"
+        },
+        "D7E03763-6377-4D2B-A56A-90648B5C5182": {
+            "author": "com.apple.CoreML",
+            "description": "CoreML Model Weights",
+            "name": "weights",
+            "path": "com.apple.CoreML/weights"
+        }
+    },
+    "rootModelIdentifier": "60BDEF3F-0828-435D-91C0-798999E28E7F"
+}

models/detector/LICENSE

@@ -0,0 +1,21 @@
+MIT License
+
+Copyright (c) Microsoft Corporation.
+
+Permission is hereby granted, free of charge, to any person obtaining a copy
+of this software and associated documentation files (the "Software"), to deal
+in the Software without restriction, including without limitation the rights
+to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+copies of the Software, and to permit persons to whom the Software is
+furnished to do so, subject to the following conditions:
+
+The above copyright notice and this permission notice shall be included in all
+copies or substantial portions of the Software.
+
+THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+SOFTWARE

models/detector/provenance.json

@@ -0,0 +1,14 @@
+{
+  "repo_id": "microsoft/OmniParser-v2.0",
+  "revision": "f55d0750e5b94db2125ef0b45b0fa4a85ddc59b4",
+  "files": {
+    "icon_detect_v3/model.pt": {
+      "size": 281182680,
+      "sha256": "11c6cbb77f22569fab22d86c76407a83ec81ab89dbfe28279854822d6e3fb00c"
+    },
+    "icon_detect_v3/LICENSE": {
+      "size": 1073,
+      "sha256": "646f8936b8ddcd14e13e578ff6857e368780b0d1a4f6066bee89211923a373e2"
+    }
+  }
+}

runtime_config.json

@@ -0,0 +1,314 @@
+{
+  "app_runtime_metrics": {},
+  "bundle_id": "local_unverified_IconDetectV3",
+  "caption_boundary_tradeoff": {
+    "available": false
+  },
+  "caption_runtime": {
+    "crop_size_before_processor": [
+      64,
+      64
+    ],
+    "generation": {
+      "decode": {
+        "skip_special_tokens": true,
+        "trim_whitespace": true
+      },
+      "decoder_prefix_length": 20,
+      "decoder_start_token_id": 2,
+      "eos_token_id": 2,
+      "forced_bos_token_id": 0,
+      "forced_eos_token_id": 2,
+      "max_new_tokens": 20,
+      "no_repeat_ngram_size": 3,
+      "pad_prefix": true,
+      "pad_token_id": 1,
+      "selection": "greedy_argmax_after_generation_constraints"
+    },
+    "image_preprocessing": {
+      "do_convert_rgb": null,
+      "do_normalize": true,
+      "do_rescale": true,
+      "do_resize": true,
+      "image_mean": [
+        0.485,
+        0.456,
+        0.406
+      ],
+      "image_std": [
+        0.229,
+        0.224,
+        0.225
+      ],
+      "rescale_factor": 0.00392156862745098,
+      "size": {
+        "height": 768,
+        "width": 768
+      }
+    },
+    "preprocessor_config": "models/caption/tokenizer/preprocessor_config.json",
+    "processor_source_model_id": "microsoft/Florence-2-base",
+    "prompt": {
+      "attention_mask": [
+        1,
+        1,
+        1,
+        1,
+        1,
+        1,
+        1,
+        1
+      ],
+      "input_ids": [
+        0,
+        2264,
+        473,
+        5,
+        2274,
+        6190,
+        116,
+        2
+      ],
+      "pixel_values_dtype": "float32",
+      "pixel_values_shape": [
+        1,
+        3,
+        768,
+        768
+      ],
+      "prompt": "<CAPTION>"
+    },
+    "tokenizer_config": "models/caption/tokenizer/tokenizer_config.json",
+    "tokenizer_json": "models/caption/tokenizer/tokenizer.json",
+    "vocab_json": "models/caption/tokenizer/vocab.json"
+  },
+  "compute_units": "cpu-and-ne",
+  "detector_postprocessing": {
+    "confidence_threshold": 0.05,
+    "max_detections": 300,
+    "nms_iou_threshold": 0.1,
+    "remove_overlap_iou_threshold": 0.7
+  },
+  "energy": {},
+  "facade_runtime_metrics": {},
+  "hard_ane_blocker": {
+    "ane_dominant_bundle": true,
+    "available": true,
+    "evidence_scope": "current_copied_package_hashes",
+    "expected_blocker": {
+      "allowed_non_neural_engine_preferred_count": 0,
+      "component": "caption_encoder"
+    },
+    "hard_ane_shippable": true,
+    "observed_blocker": {
+      "non_neural_engine_preferred_count": 0,
+      "operation": null
+    },
+    "passed": true,
+    "source": "manifest.placement_evidence",
+    "source_sha256": "8b29c6f27756f1d91d426ab286b9dc88c9ea71919840c90c7eb9895dc2ae2691",
+    "state": "hard_ane_resolved"
+  },
+  "minimum_macos": "15.0",
+  "models": {
+    "caption_decoder": {
+      "inputs": {
+        "decoder_inputs_embeds": {
+          "dtype": "float16",
+          "shape": [
+            1,
+            20,
+            768
+          ]
+        },
+        "encoder_hidden_states": {
+          "dtype": "float16",
+          "shape": [
+            1,
+            585,
+            768
+          ]
+        }
+      },
+      "output": {
+        "name": "logits",
+        "shape": [
+          1,
+          20,
+          51289
+        ]
+      },
+      "path": "models/caption/FlorenceDecoderPrefixEmbeds.mlpackage",
+      "placement": {
+        "first_non_neural_engine_preferred": null,
+        "hard_ane": true,
+        "non_neural_engine_preferred_count": 0,
+        "planned_operation_count": 315
+      }
+    },
+    "caption_encoder": {
+      "inputs": {
+        "pixel_values": {
+          "dtype": "float16",
+          "layout": "NCHW",
+          "shape": [
+            1,
+            3,
+            768,
+            768
+          ]
+        },
+        "prompt_embeds": {
+          "dtype": "float16",
+          "shape": [
+            1,
+            8,
+            768
+          ]
+        }
+      },
+      "output": {
+        "dtype": "float16",
+        "name": "encoder_hidden_states",
+        "shape": [
+          1,
+          585,
+          768
+        ]
+      },
+      "path": "models/caption/FlorenceEncoderContextEmbeds.mlpackage",
+      "placement": {
+        "first_non_neural_engine_preferred": null,
+        "hard_ane": true,
+        "non_neural_engine_preferred_count": 0,
+        "planned_operation_count": 1042
+      }
+    },
+    "caption_shared_embedding": {
+      "dtype": "float16",
+      "path": "models/caption/shared_embedding_fp16.npy",
+      "runtime_policy": "memory_map_read_only",
+      "shape": [
+        51289,
+        768
+      ]
+    },
+    "detector": {
+      "default_function": "detect_640",
+      "family": "icon_detect_v3",
+      "functions": [
+        {
+          "image_size": 640,
+          "input": {
+            "dtype": "float16",
+            "layout": "NCHW",
+            "name": "image",
+            "shape": [
+              1,
+              3,
+              640,
+              640
+            ]
+          },
+          "name": "detect_640",
+          "output": {
+            "name": "predictions",
+            "shape": [
+              1,
+              5,
+              8400
+            ]
+          }
+        },
+        {
+          "image_size": 1280,
+          "input": {
+            "dtype": "float16",
+            "layout": "NCHW",
+            "name": "image",
+            "shape": [
+              1,
+              3,
+              1280,
+              1280
+            ]
+          },
+          "name": "detect_1280",
+          "output": {
+            "name": "predictions",
+            "shape": [
+              1,
+              5,
+              33600
+            ]
+          }
+        }
+      ],
+      "license": "MIT",
+      "output_layout": "[1, 5, anchors]: class logit, then l, t, r, b in grid units",
+      "path": "models/detector/IconDetectV3.mlpackage",
+      "placement": {
+        "first_non_neural_engine_preferred": null,
+        "hard_ane": true,
+        "non_neural_engine_preferred_count": 0,
+        "planned_operation_count": 1594
+      },
+      "preprocessing": {
+        "color_space": "RGB",
+        "letterbox": "square_center_floor",
+        "pad_value": 114,
+        "resample": "pillow_lanczos",
+        "value_range": "0..1"
+      },
+      "source_revision": "f55d0750e5b94db2125ef0b45b0fa4a85ddc59b4",
+      "source_sha256": "11c6cbb77f22569fab22d86c76407a83ec81ab89dbfe28279854822d6e3fb00c",
+      "strides": [
+        8,
+        16,
+        32
+      ]
+    }
+  },
+  "policy": {
+    "ane_as_much_as_possible_without_behavior_change": "unverified",
+    "ane_dominant_faithful_candidate": false,
+    "blocker": {
+      "component": null,
+      "interpretation": "fresh placement evidence reports zero non-ANE executable ops",
+      "operation": null
+    },
+    "caption_encoder_allowed_cpu_fallbacks": 0,
+    "hard_ane_shippable": true,
+    "literal_hard_ane": "satisfied"
+  },
+  "runtime_cache": {
+    "available": false,
+    "failure_reason": "missing report",
+    "source": "artifacts/release/coreml_runtime_cache_report.json"
+  },
+  "runtime_metrics": {},
+  "schema_version": 3,
+  "storage_footprint": {
+    "available": true,
+    "caption_decoder_bytes": 192475683,
+    "caption_encoder_bytes": 269990610,
+    "caption_model_bytes": 541246325,
+    "content_bytes_excluding_runtime_config": 660781322,
+    "detector_bytes": 117076927,
+    "model_bytes": 658323252,
+    "policy": {
+      "compiled_mlmodelc_in_bundle_allowed": false,
+      "max_caption_decoder_bytes": 251658240,
+      "max_caption_encoder_bytes": 335544320,
+      "max_detector_bytes": 83886080,
+      "max_model_bytes": 681574400,
+      "max_shared_embedding_bytes": 100663296,
+      "max_tokenizer_bytes": 4194304,
+      "ship_mlpackage_only": true
+    },
+    "runtime_config_self_size_excluded": true,
+    "shared_embedding_bytes": 78780032,
+    "tokenizer_bytes": 2456587
+  },
+  "target": "macos"
+}