mlboydaisuke commited on
Commit
14a992c
·
verified ·
1 Parent(s): ab9eaa7

Upload README.md with huggingface_hub

Browse files
Files changed (1) hide show
  1. README.md +12 -4
README.md CHANGED
@@ -7,7 +7,7 @@ tags:
7
  - on-device
8
  - image-segmentation
9
  ---
10
- # dis_isnet — ExecuTorch XNNPACK
11
 
12
  - **Source**: xuebinqin/DIS + NimaBoscarino/IS-Net_DIS-general-use weights
13
  - **License**: Apache-2.0
@@ -18,14 +18,22 @@ tags:
18
 
19
  All variants take and return fp32 tensors — swap the `.pte` file, keep your app code.
20
 
21
- | precision | file | size (MB) | parity vs fp32 eager (worst corr) | Mac median (ms)* |
22
  |-----------|------|-----------|------------------------------------|------------------|
23
  | fp32 | `dis_isnet_xnnpack_fp32.pte` | 176.1 | 1.000000 | 123.4 |
 
 
 
 
 
 
 
 
24
 
25
  \*Mac arm64, single process, median of 10 — a reference point for relative cost
26
  only, not a device number (torch eager fp32 on the same machine: 364.6 ms).
27
 
28
- ### Precisions that did not earn a slot
29
 
30
  - **fp16 is not shipped**: worst-output corr 0.986 against fp32 eager, below the 0.995 bar for this precision. The file converts and runs; the numbers do not hold up, so it is left out rather than shipped with a warning.
31
  - **int8 is not shipped**: measured in the units that matter for this model — mask IoU at 0.5: median 0.9106 over 10 real images, worst 0.4647.
@@ -43,5 +51,5 @@ XNNPACK delegate coverage (fp32): 100.0% (467/467 ops)
43
 
44
  ## Conversion
45
 
46
- torch.export -> to_edge_transform_and_lower(XnnpackPartitioner) -> .pte
47
  (conversion scripts: [executorch-models](https://github.com/john-rocky/executorch-models))
 
7
  - on-device
8
  - image-segmentation
9
  ---
10
+ # dis_isnet — ExecuTorch
11
 
12
  - **Source**: xuebinqin/DIS + NimaBoscarino/IS-Net_DIS-general-use weights
13
  - **License**: Apache-2.0
 
18
 
19
  All variants take and return fp32 tensors — swap the `.pte` file, keep your app code.
20
 
21
+ | build | file | size (MB) | parity vs fp32 eager (worst corr) | Mac median (ms)* |
22
  |-----------|------|-----------|------------------------------------|------------------|
23
  | fp32 | `dis_isnet_xnnpack_fp32.pte` | 176.1 | 1.000000 | 123.4 |
24
+ | Core ML (fp16, iOS) | `dis_isnet_coreml_all.pte` | 89.0 | 0.999984 | 29.2 |
25
+
26
+
27
+ The Core ML build is the same graph lowered to Apple's Neural Engine instead of
28
+ XNNPACK, which is CPU-only. On an iPhone 17 Pro, Depth-Anything-V2-Small runs
29
+ 500.8 ms through XNNPACK and 42.7 ms through Core ML, at half the file size. It
30
+ computes in fp16 and is iOS-only; the XNNPACK files stay the portable option and
31
+ are what runs on Android.
32
 
33
  \*Mac arm64, single process, median of 10 — a reference point for relative cost
34
  only, not a device number (torch eager fp32 on the same machine: 364.6 ms).
35
 
36
+ ### Builds that did not earn a slot
37
 
38
  - **fp16 is not shipped**: worst-output corr 0.986 against fp32 eager, below the 0.995 bar for this precision. The file converts and runs; the numbers do not hold up, so it is left out rather than shipped with a warning.
39
  - **int8 is not shipped**: measured in the units that matter for this model — mask IoU at 0.5: median 0.9106 over 10 real images, worst 0.4647.
 
51
 
52
  ## Conversion
53
 
54
+ torch.export -> to_edge_transform_and_lower(partitioner) -> .pte
55
  (conversion scripts: [executorch-models](https://github.com/john-rocky/executorch-models))