SAM 2.1 tiny as a Core AI .aimodel (float16, encode + decode), with README, licence and export script

.gitattributes

@@ -33,3 +33,4 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
 *.zip filter=lfs diff=lfs merge=lfs -text
 *.zst filter=lfs diff=lfs merge=lfs -text
 *tfevents* filter=lfs diff=lfs merge=lfs -text
+cutout-sam2-tiny-float16.aimodel/main.mlirb filter=lfs diff=lfs merge=lfs -text

LICENSE

@@ -0,0 +1,201 @@
+                                 Apache License
+                           Version 2.0, January 2004
+                        http://www.apache.org/licenses/
+
+   TERMS AND CONDITIONS FOR USE, REPRODUCTION, AND DISTRIBUTION
+
+   1. Definitions.
+
+      "License" shall mean the terms and conditions for use, reproduction,
+      and distribution as defined by Sections 1 through 9 of this document.
+
+      "Licensor" shall mean the copyright owner or entity authorized by
+      the copyright owner that is granting the License.
+
+      "Legal Entity" shall mean the union of the acting entity and all
+      other entities that control, are controlled by, or are under common
+      control with that entity. For the purposes of this definition,
+      "control" means (i) the power, direct or indirect, to cause the
+      direction or management of such entity, whether by contract or
+      otherwise, or (ii) ownership of fifty percent (50%) or more of the
+      outstanding shares, or (iii) beneficial ownership of such entity.
+
+      "You" (or "Your") shall mean an individual or Legal Entity
+      exercising permissions granted by this License.
+
+      "Source" form shall mean the preferred form for making modifications,
+      including but not limited to software source code, documentation
+      source, and configuration files.
+
+      "Object" form shall mean any form resulting from mechanical
+      transformation or translation of a Source form, including but
+      not limited to compiled object code, generated documentation,
+      and conversions to other media types.
+
+      "Work" shall mean the work of authorship, whether in Source or
+      Object form, made available under the License, as indicated by a
+      copyright notice that is included in or attached to the work
+      (an example is provided in the Appendix below).
+
+      "Derivative Works" shall mean any work, whether in Source or Object
+      form, that is based on (or derived from) the Work and for which the
+      editorial revisions, annotations, elaborations, or other modifications
+      represent, as a whole, an original work of authorship. For the purposes
+      of this License, Derivative Works shall not include works that remain
+      separable from, or merely link (or bind by name) to the interfaces of,
+      the Work and Derivative Works thereof.
+
+      "Contribution" shall mean any work of authorship, including
+      the original version of the Work and any modifications or additions
+      to that Work or Derivative Works thereof, that is intentionally
+      submitted to Licensor for inclusion in the Work by the copyright owner
+      or by an individual or Legal Entity authorized to submit on behalf of
+      the copyright owner. For the purposes of this definition, "submitted"
+      means any form of electronic, verbal, or written communication sent
+      to the Licensor or its representatives, including but not limited to
+      communication on electronic mailing lists, source code control systems,
+      and issue tracking systems that are managed by, or on behalf of, the
+      Licensor for the purpose of discussing and improving the Work, but
+      excluding communication that is conspicuously marked or otherwise
+      designated in writing by the copyright owner as "Not a Contribution."
+
+      "Contributor" shall mean Licensor and any individual or Legal Entity
+      on behalf of whom a Contribution has been received by Licensor and
+      subsequently incorporated within the Work.
+
+   2. Grant of Copyright License. Subject to the terms and conditions of
+      this License, each Contributor hereby grants to You a perpetual,
+      worldwide, non-exclusive, no-charge, royalty-free, irrevocable
+      copyright license to reproduce, prepare Derivative Works of,
+      publicly display, publicly perform, sublicense, and distribute the
+      Work and such Derivative Works in Source or Object form.
+
+   3. Grant of Patent License. Subject to the terms and conditions of
+      this License, each Contributor hereby grants to You a perpetual,
+      worldwide, non-exclusive, no-charge, royalty-free, irrevocable
+      (except as stated in this section) patent license to make, have made,
+      use, offer to sell, sell, import, and otherwise transfer the Work,
+      where such license applies only to those patent claims licensable
+      by such Contributor that are necessarily infringed by their
+      Contribution(s) alone or by combination of their Contribution(s)
+      with the Work to which such Contribution(s) was submitted. If You
+      institute patent litigation against any entity (including a
+      cross-claim or counterclaim in a lawsuit) alleging that the Work
+      or a Contribution incorporated within the Work constitutes direct
+      or contributory patent infringement, then any patent licenses
+      granted to You under this License for that Work shall terminate
+      as of the date such litigation is filed.
+
+   4. Redistribution. You may reproduce and distribute copies of the
+      Work or Derivative Works thereof in any medium, with or without
+      modifications, and in Source or Object form, provided that You
+      meet the following conditions:
+
+      (a) You must give any other recipients of the Work or
+          Derivative Works a copy of this License; and
+
+      (b) You must cause any modified files to carry prominent notices
+          stating that You changed the files; and
+
+      (c) You must retain, in the Source form of any Derivative Works
+          that You distribute, all copyright, patent, trademark, and
+          attribution notices from the Source form of the Work,
+          excluding those notices that do not pertain to any part of
+          the Derivative Works; and
+
+      (d) If the Work includes a "NOTICE" text file as part of its
+          distribution, then any Derivative Works that You distribute must
+          include a readable copy of the attribution notices contained
+          within such NOTICE file, excluding those notices that do not
+          pertain to any part of the Derivative Works, in at least one
+          of the following places: within a NOTICE text file distributed
+          as part of the Derivative Works; within the Source form or
+          documentation, if provided along with the Derivative Works; or,
+          within a display generated by the Derivative Works, if and
+          wherever such third-party notices normally appear. The contents
+          of the NOTICE file are for informational purposes only and
+          do not modify the License. You may add Your own attribution
+          notices within Derivative Works that You distribute, alongside
+          or as an addendum to the NOTICE text from the Work, provided
+          that such additional attribution notices cannot be construed
+          as modifying the License.
+
+      You may add Your own copyright statement to Your modifications and
+      may provide additional or different license terms and conditions
+      for use, reproduction, or distribution of Your modifications, or
+      for any such Derivative Works as a whole, provided Your use,
+      reproduction, and distribution of the Work otherwise complies with
+      the conditions stated in this License.
+
+   5. Submission of Contributions. Unless You explicitly state otherwise,
+      any Contribution intentionally submitted for inclusion in the Work
+      by You to the Licensor shall be under the terms and conditions of
+      this License, without any additional terms or conditions.
+      Notwithstanding the above, nothing herein shall supersede or modify
+      the terms of any separate license agreement you may have executed
+      with Licensor regarding such Contributions.
+
+   6. Trademarks. This License does not grant permission to use the trade
+      names, trademarks, service marks, or product names of the Licensor,
+      except as required for reasonable and customary use in describing the
+      origin of the Work and reproducing the content of the NOTICE file.
+
+   7. Disclaimer of Warranty. Unless required by applicable law or
+      agreed to in writing, Licensor provides the Work (and each
+      Contributor provides its Contributions) on an "AS IS" BASIS,
+      WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or
+      implied, including, without limitation, any warranties or conditions
+      of TITLE, NON-INFRINGEMENT, MERCHANTABILITY, or FITNESS FOR A
+      PARTICULAR PURPOSE. You are solely responsible for determining the
+      appropriateness of using or redistributing the Work and assume any
+      risks associated with Your exercise of permissions under this License.
+
+   8. Limitation of Liability. In no event and under no legal theory,
+      whether in tort (including negligence), contract, or otherwise,
+      unless required by applicable law (such as deliberate and grossly
+      negligent acts) or agreed to in writing, shall any Contributor be
+      liable to You for damages, including any direct, indirect, special,
+      incidental, or consequential damages of any character arising as a
+      result of this License or out of the use or inability to use the
+      Work (including but not limited to damages for loss of goodwill,
+      work stoppage, computer failure or malfunction, or any and all
+      other commercial damages or losses), even if such Contributor
+      has been advised of the possibility of such damages.
+
+   9. Accepting Warranty or Additional Liability. While redistributing
+      the Work or Derivative Works thereof, You may choose to offer,
+      and charge a fee for, acceptance of support, warranty, indemnity,
+      or other liability obligations and/or rights consistent with this
+      License. However, in accepting such obligations, You may act only
+      on Your own behalf and on Your sole responsibility, not on behalf
+      of any other Contributor, and only if You agree to indemnify,
+      defend, and hold each Contributor harmless for any liability
+      incurred by, or claims asserted against, such Contributor by reason
+      of your accepting any such warranty or additional liability.
+
+   END OF TERMS AND CONDITIONS
+
+   APPENDIX: How to apply the Apache License to your work.
+
+      To apply the Apache License to your work, attach the following
+      boilerplate notice, with the fields enclosed by brackets "[]"
+      replaced with your own identifying information. (Don't include
+      the brackets!)  The text should be enclosed in the appropriate
+      comment syntax for the file format. We also recommend that a
+      file or class name and description of purpose be included on the
+      same "printed page" as the copyright notice for easier
+      identification within third-party archives.
+
+   Copyright [yyyy] [name of copyright owner]
+
+   Licensed under the Apache License, Version 2.0 (the "License");
+   you may not use this file except in compliance with the License.
+   You may obtain a copy of the License at
+
+       http://www.apache.org/licenses/LICENSE-2.0
+
+   Unless required by applicable law or agreed to in writing, software
+   distributed under the License is distributed on an "AS IS" BASIS,
+   WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+   See the License for the specific language governing permissions and
+   limitations under the License.

NOTICE

@@ -0,0 +1,15 @@
+cutout-sam2-tiny: SAM 2.1 hiera tiny, Copyright (c) Meta Platforms, Inc. and
+affiliates. Code: facebookresearch/sam2 at commit
+2b90b9f5ceec907a1c18123530e92e794ad901a4. Weights: facebook/sam2.1-hiera-tiny
+at revision de431c4043854a71d8101e17995dfe596bf101a5 (sam2.1_hiera_tiny.pt).
+Both Apache License 2.0; see LICENSE.
+
+Changed for Core AI, in export_sam2.py:
+- The image predictor's own steps are traced as two entry points: `encode`
+  (normalisation, image encoder, no-memory embedding) and `decode` (prompt
+  encoder for one point, mask decoder, multimask output).
+- Weights and arithmetic are float16, except the prompt encoder, which stays
+  float32; every input and output is float32.
+- The 256x256 mask logits are returned unclamped (the PyTorch predictor clamps
+  them to +/-32; the host thresholds them).
+The trained weights are otherwise the authors', unmodified.

README.md

@@ -0,0 +1,96 @@
+---
+license: apache-2.0
+base_model: facebook/sam2.1-hiera-tiny
+tags:
+  - segment-anything
+  - sam2
+  - interactive-segmentation
+  - background-removal
+  - core-ai
+  - apple-silicon
+  - macos
+library_name: swift-object-cutout
+pipeline_tag: mask-generation
+---
+
+# cutout-sam2-tiny: SAM 2.1 tiny for Core AI (macOS 27)
+
+[SAM 2.1](https://github.com/facebookresearch/sam2) hiera tiny (Ravi et al., Meta, *SAM 2: Segment
+Anything in Images and Videos*, 2024; code and weights Apache-2.0), exported to an Apple Core AI
+`.aimodel`. Click a point on a picture and it returns the object under it: three candidate masks
+and the network's own estimate of how good each one is.
+
+[`cutout`](https://github.com/arraypress/swift-cutout-cli) uses it for `--at` (click to cut), through
+[swift-object-cutout](https://github.com/arraypress/swift-object-cutout). It downloads this
+repository by itself the first time it is needed, pinned to one commit, with every file checked
+against its published hash.
+
+## Files
+
+| Path | What |
+|---|---|
+| `cutout-sam2-tiny-float16.aimodel/` | Two entry points (below). Float16 weights and arithmetic, float32 inputs and outputs, 63 MB. |
+| `export_sam2.py` | The export script (`SAM2_BUILD_CUDA=0 uv run export_sam2.py --dtype float16`). |
+| `LICENSE`, `NOTICE` | The authors' Apache-2.0 licence, and what the export changed. |
+
+**`encode`**: `image` `[1,3,1024,1024]` float32, RGB in 0…1 → `image_embedding` `[1,256,64,64]`,
+`feats_s0` `[1,32,256,256]`, `feats_s1` `[1,64,128,128]`. The host stretches the picture to
+1024×1024 without keeping its aspect ratio, as SAM 2's own predictor does; the ImageNet
+normalisation and SAM 2's no-memory embedding are inside.
+
+**`decode`**: the three encoder outputs, `point` `[1,1,2]` (x then y, in pixels of the 1024² square)
+and `label` `[1,1]` (1 = the object, 0 = not it) → `masks` `[1,3,256,256]` logits and `scores`
+`[1,3]`, the predicted IoU of each. Take the best-scoring mask, threshold it at 0 and stretch it back
+over the picture. One encoding serves any number of clicks.
+
+Nothing in the network is rewritten: the two entry points are the official
+`SAM2ImagePredictor.set_image` and `_predict` steps, traced as they are. In float32 on the CPU the
+traced steps match the predictor to 8 × 10⁻⁵ in the logits.
+
+## Fidelity
+
+60 clicks (two per picture) on 30 generated pictures: 20 AI-generated room and house scenes at
+1024², and 10 synthetic shape scenes at sizes from 640×480 to 1600×900. The same 1024² pixels went
+to every runtime. The reference is the PyTorch original in float32 on the CPU; "chosen" is the
+best-scoring mask, which is the one a click returns.
+
+| Runtime | Chosen-mask IoU, mean | worst | Clicks ≥ 0.99 | Same mask chosen | Score error, worst |
+|---|---|---|---|---|---|
+| this export, float16 (this repo) | 0.9996 | 0.9959 | 60 of 60 | 60 of 60 | 0.006 |
+| this export, float32 (not published) | 1.0000 | 1.0000 | 60 of 60 | 60 of 60 | < 10⁻⁴ |
+| Apple's Core ML conversion (`apple/coreml-sam2.1-tiny`, FLOAT16) | 0.968 | 0.500 | 34 of 60 | 56 of 60 | 0.22 |
+
+Apple's Core ML conversion is what `cutout` ran before. Its masks drift from the original on
+textured scenes, where it sometimes picks a different mask (IoU 0.50 on one click). This export
+agrees with the original on every click. On the synthetic scenes the two runtimes agree to
+0.99 or better, end to end; on the generated rooms they differ by as much as Apple's conversion differs
+from the original.
+
+## Speed and memory
+
+Apple M3 Max, macOS 27, GPU, measured through `cutout`:
+
+- **First load:** compiles the model for the Mac in well under a second and keeps a 60 MB compiled
+  copy in `~/Library/Caches/coreai-cache`.
+- **Per click:** `encode` takes 0.15 s and `decode` 5 ms. One `cutout --at` run takes 0.6–1.0 s from
+  start to finish, against 4.5 s for the Core ML conversion, most of which went on loading.
+- **Batches:** a folder of 30 pictures took 0.51 s per picture (decode, model, edge, PNG), against
+  0.42 s for the Core ML conversion, which ran on the Neural Engine.
+- **Memory:** the folder run peaked at 1.2 GB, against 1.4 GB for the Core ML conversion.
+
+## Use
+
+```sh
+cutout photo.jpg --at 0.32,0.6          # fetches this model on first use
+cutout ./shots --at 0.5,0.45 --out cut  # a folder, the model loaded once
+cutout model remove click               # the model and its compiled copy
+```
+
+Or by hand: `hf download arraypress/cutout-sam2-tiny --local-dir models`, then point
+`$CUTOUT_CLICK_MODEL` at the `.aimodel`. Requires macOS 27 (Core AI) on Apple silicon. Converted
+with coreai-torch 0.4.3, coreai-core 1.0.0b3, torch 2.9.0, hydra-core 1.3.2 and sam2 at 2b90b9f.
+
+## Licence
+
+Apache-2.0, as the original: see `LICENSE` (the authors' licence) and `NOTICE` (the copyright, and
+what the export changed).

cutout-sam2-tiny-float16.aimodel/main.hash

@@ -0,0 +1 @@
+�B��o��ɸp��%��
+�i|�<1�k��Y�;�@

cutout-sam2-tiny-float16.aimodel/main.mlirb

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:df42c3de6f9ee9c9b870b2a225b4fc0d2b91697c843c318c6bf08c59b13bb940
+size 62768685

cutout-sam2-tiny-float16.aimodel/metadata.json

@@ -0,0 +1,8 @@
+{
+  "assetVersion" : "2.0",
+  "description" : "SAM 2.1 hiera tiny (facebook\/sam2.1-hiera-tiny@de431c4, code facebookresearch\/sam2@2b90b9f), float16. 'encode': image [1,3,1024,1024] float32 RGB 0…1 → image_embedding, feats_s0, feats_s1. 'decode': those + point [1,1,2] (pixels of the 1024² square) + label [1,1] → masks [1,3,256,256] logits, scores [1,3].",
+  "author" : "Ravi, Gabeur, Hu, Hu, Ryali, Ma, Khedr, Rädle, Rolland, Gustafson, et al. (Meta) — SAM 2.1 tiny; Core AI export by ObjectCutout",
+  "creationDate" : "20261017T062235Z",
+  "license" : "Apache-2.0",
+  "producer" : "coreai-core 1.0.0b3"
+}

export_sam2.py

@@ -0,0 +1,152 @@
+# /// script
+# requires-python = ">=3.11,<3.13"
+# dependencies = ["torch==2.9.0", "torchvision==0.24.0", "numpy", "pillow", "huggingface_hub", "hydra-core==1.3.2",
+#                 "sam-2 @ git+https://github.com/facebookresearch/sam2.git@2b90b9f5ceec907a1c18123530e92e794ad901a4",
+#                 "coreai-torch==0.4.3", "coreai-core==1.0.0b3"]
+# ///
+"""SAM 2.1 tiny (Meta, Apache-2.0) → cutout-sam2-tiny-<dtype>.aimodel, for the click.
+
+    SAM2_BUILD_CUDA=0 uv run Tools/export_sam2.py [--dtype float16|float32] [--out Tools/exports] [--check image.jpg …]
+
+The network is the authors' own: facebookresearch/sam2 at a pinned commit, with the official SAM 2.1 tiny checkpoint
+(facebook/sam2.1-hiera-tiny at a pinned revision). Nothing in it is rewritten; the two entry points are the image
+predictor's own steps (SAM2ImagePredictor.set_image and ._predict), traced as they are.
+
+Two entry points, so one encoded picture serves any number of clicks:
+
+  encode  `image` [1,3,1024,1024] float32 RGB in 0…1 (stretched to the square by the host)
+          → `image_embedding` [1,256,64,64], `feats_s0` [1,32,256,256], `feats_s1` [1,64,128,128]
+          The ImageNet normalisation and SAM 2's no-memory embedding are inside.
+  decode  those three, `point` [1,1,2] float32 (pixels of the 1024² square, x then y) and `label` [1,1] float32
+          (1 = the object, 0 = not it) → `masks` [1,3,256,256] logits and `scores` [1,3] (predicted IoU).
+          The prompt encoder and the mask decoder, multimask output, SAM's padding point added inside.
+
+In a float16 export the weights and arithmetic are half; every input and output stays float32.
+
+--check runs the official SAM2ImagePredictor and the two traced steps on each image (a click at the centre) in
+float32 on the CPU and prints the difference, before anything is exported."""
+import argparse, sys, time, shutil
+from pathlib import Path
+import numpy as np, torch, torch.nn as nn
+
+HERE = Path(__file__).resolve().parent
+REPOSITORY = "facebook/sam2.1-hiera-tiny"
+REVISION = "de431c4043854a71d8101e17995dfe596bf101a5"
+CHECKPOINT = "sam2.1_hiera_tiny.pt"
+CODE = "facebookresearch/sam2@2b90b9f"
+SIZE = 1024
+
+ap = argparse.ArgumentParser()
+ap.add_argument("--dtype", default="float16", choices=["float16", "float32"])
+ap.add_argument("--out", default=str(HERE / "exports"))
+ap.add_argument("--weights", default=None, help="a folder to keep the checkpoint in (default: the HF cache)")
+ap.add_argument("--check", nargs="*", default=[], help="images to compare the predictor and the traced steps on")
+ap.add_argument("--no-export", action="store_true")
+args = ap.parse_args()
+
+from huggingface_hub import hf_hub_download
+from sam2.build_sam import build_sam2
+from sam2.sam2_image_predictor import SAM2ImagePredictor
+
+torch.set_num_threads(4)
+ckpt = hf_hub_download(REPOSITORY, CHECKPOINT, revision=REVISION, local_dir=args.weights)
+sam = build_sam2("configs/sam2.1/sam2.1_hiera_t.yaml", ckpt, device="cpu", mode="eval")
+print(f"SAM 2.1 tiny @ {REVISION[:7]}: {sum(p.numel() for p in sam.parameters())/1e6:.1f} M parameters", flush=True)
+FEAT_SIZES = [(256, 256), (128, 128), (64, 64)]       # SAM2ImagePredictor._bb_feat_sizes at 1024²
+MEAN = torch.tensor([0.485, 0.456, 0.406]).reshape(1, 3, 1, 1)
+STD = torch.tensor([0.229, 0.224, 0.225]).reshape(1, 3, 1, 1)
+
+
+class Encode(nn.Module):
+    """SAM2ImagePredictor.set_image after its resize: normalise, backbone, no-memory embedding, reshape."""
+    def __init__(self, m, dtype):
+        super().__init__(); self.m = m; self.dtype = dtype
+        self.register_buffer("mean", MEAN.to(dtype)); self.register_buffer("std", STD.to(dtype))
+    def forward(self, image):
+        x = (image.to(self.dtype) - self.mean) / self.std
+        backbone_out = self.m.forward_image(x)
+        _, vision_feats, _, _ = self.m._prepare_backbone_features(backbone_out)
+        if self.m.directly_add_no_mem_embed:
+            vision_feats[-1] = vision_feats[-1] + self.m.no_mem_embed
+        feats = [f.permute(1, 2, 0).reshape(1, -1, *s) for f, s in zip(vision_feats[::-1], FEAT_SIZES[::-1])][::-1]
+        return feats[2].to(torch.float32), feats[0].to(torch.float32), feats[1].to(torch.float32)
+
+
+class Decode(nn.Module):
+    """SAM2ImagePredictor._predict for one point, multimask: prompt encoder, mask decoder."""
+    def __init__(self, m, dtype):
+        super().__init__(); self.m = m; self.dtype = dtype
+    def forward(self, image_embedding, feats_s0, feats_s1, point, label):
+        d = self.dtype
+        # The prompt encoder stays float32 (it is tiny, and its padding point is made float32 inside).
+        sparse, dense = self.m.sam_prompt_encoder(points=(point, label.to(torch.int32)), boxes=None, masks=None)
+        masks, scores, _, _ = self.m.sam_mask_decoder(
+            image_embeddings=image_embedding.to(d), image_pe=self.m.sam_prompt_encoder.get_dense_pe().to(d),
+            sparse_prompt_embeddings=sparse.to(d), dense_prompt_embeddings=dense.to(d), multimask_output=True,
+            repeat_image=False, high_res_features=[feats_s0.to(d), feats_s1.to(d)])
+        return masks.to(torch.float32), scores.to(torch.float32)
+
+
+def square(path):
+    from PIL import Image
+    im = Image.open(path).convert("RGB")
+    return im, torch.from_numpy(np.asarray(im.resize((SIZE, SIZE), Image.BILINEAR), np.float32) / 255).permute(2, 0, 1)[None]
+
+
+if args.check:
+    predictor = SAM2ImagePredictor(sam)
+    enc, dec = Encode(sam, torch.float32).eval(), Decode(sam, torch.float32).eval()
+    with torch.no_grad():
+        for path in args.check:
+            im, x = square(path)
+            predictor.set_image(np.asarray(im))
+            w, h = im.size
+            _, s_ref, low_ref = predictor.predict(point_coords=np.array([[w / 2, h / 2]]), point_labels=np.array([1]),
+                                                  multimask_output=True, return_logits=True)
+            # The predictor resizes with its own transform; feed the traced steps its tensor so only the steps differ.
+            x = predictor._transforms(np.asarray(im))[None]
+            x = x * STD + MEAN
+            e, s0, s1 = enc(x)
+            m, s = dec(e, s0, s1, torch.tensor([[[SIZE / 2, SIZE / 2]]]), torch.ones(1, 1))
+            # The predictor clamps its low-resolution logits to ±32; the export does not (the host thresholds).
+            dm = np.abs(m[0].clamp(-32, 32).numpy() - low_ref).max()
+            print(f"{Path(path).name}: |predictor − traced| masks {dm:.2e} scores {np.abs(s[0].numpy() - s_ref).max():.2e}",
+                  flush=True)
+
+if args.no_export:
+    sys.exit(0)
+
+import coreai_torch
+from coreai.runtime import AIModelAssetMetadata
+dtype = torch.float16 if args.dtype == "float16" else torch.float32
+sam = sam.to(dtype)
+sam.sam_prompt_encoder.float()
+enc, dec = Encode(sam, dtype).eval(), Decode(sam, dtype).eval()
+t0 = time.time()
+converter = coreai_torch.TorchConverter(mode=coreai_torch.TorchConverter.Mode.RELEASE)
+with torch.no_grad():
+    x = torch.rand(1, 3, SIZE, SIZE)
+    ex = torch.export.export(enc, (x,)).run_decompositions(coreai_torch.get_decomp_table())
+    converter.add_exported_program(ex, input_names=["image"],
+                                   output_names=["image_embedding", "feats_s0", "feats_s1"], entrypoint_name="encode")
+    e, s0, s1 = enc(x)
+    ex = torch.export.export(dec, (e, s0, s1, torch.tensor([[[300.0, 400.0]]]), torch.ones(1, 1))
+                             ).run_decompositions(coreai_torch.get_decomp_table())
+    converter.add_exported_program(ex, input_names=["image_embedding", "feats_s0", "feats_s1", "point", "label"],
+                                   output_names=["masks", "scores"], entrypoint_name="decode")
+    print(f"traced in {time.time()-t0:.0f} s", flush=True)
+program = converter.to_coreai()
+print(f"converted in {time.time()-t0:.0f} s", flush=True)
+meta = AIModelAssetMetadata()
+meta.author = "Ravi, Gabeur, Hu, Hu, Ryali, Ma, Khedr, Rädle, Rolland, Gustafson, et al. (Meta) — SAM 2.1 tiny; Core AI export by ObjectCutout"
+meta.license = "Apache-2.0"
+meta.model_description = (f"SAM 2.1 hiera tiny ({REPOSITORY}@{REVISION[:7]}, code {CODE}), {args.dtype}. "
+                          f"'encode': image [1,3,{SIZE},{SIZE}] float32 RGB 0…1 → image_embedding, feats_s0, feats_s1. "
+                          f"'decode': those + point [1,1,2] (pixels of the {SIZE}² square) + label [1,1] "
+                          f"→ masks [1,3,256,256] logits, scores [1,3].")
+meta.creation_date = int(time.time())
+out = Path(args.out); out.mkdir(parents=True, exist_ok=True)
+asset = out / f"cutout-sam2-tiny-{args.dtype}.aimodel"
+if asset.exists(): shutil.rmtree(asset)
+program.save_asset(asset, meta)
+print(f"saved {asset} ({sum(f.stat().st_size for f in asset.rglob('*'))/1e6:.0f} MB) in {time.time()-t0:.0f} s", flush=True)