Upload folder using huggingface_hub

onnxruntime-gpu-cuda13/onnxruntime_gpu-1.24.0-cp311-cp311-linux_x86_64.whl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:554aff20928cee1fff07f914937cf49046ba661536edeb8a5f149544b6f1877e
+size 396952334

onnxruntime-gpu-cuda13/onnxruntime_gpu-1.24.0-cp311-cp311-win_amd64.whl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d53c9b73e4075cb5b79cd2ad8a947d03d62dea47ae4e4447728a152ded555a7d
+size 386281727

onnxruntime-gpu-cuda13/onnxruntime_gpu-1.24.0-cp312-cp312-linux_x86_64.whl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b411ad0a75c6812e76631b7e090688596f51f32227a81e45cc911b28c2d4e9e2
+size 396956282

onnxruntime-gpu-cuda13/onnxruntime_gpu-1.24.0-cp312-cp312-win_amd64.whl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:519ec75ff5a2b78ceea35d18c61c5ba944e81d7453855d48e6979ffecae97bdf
+size 386243087

onnxsim/onnxsim-0.4.36-cp312-cp312-linux_x86_64.whl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:82452637624f98233fd9a504a60eca8a2f21f2a771546bbdc518ccffd7059822
+size 2199905

onnxsim/onnxsim-0.4.36-cp312-cp312-win_amd64.whl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:05f541c93a7a77ceb2c8a6b8e10d03cfcf3a4ef1fb693652baf3a552a1e6d4c6
+size 1316462

wheels-for-A1111-on-py312/README.md

@@ -0,0 +1,33 @@
+`scikit-image==0.21.0`:
+Directly required by A1111;
+
+Alternatively, bump it up to 0.24.0 seems to work well.
+
+`scikit-image>=0.25` requires `pillow>=10`, which breaks xyz plot.
+
+Compiling it on Windows need `clang-cl.exe`, then set `CC=clang-cl` and `CXX=clang-cl`.
+
+---
+
+`tokenizers==0.13.3`:
+Required by `transformers==4.30.2`.
+
+`transformers>=4.35.2` requires tokenizers with cp312 wheel.
+
+However it will break model loading: `Failed to create model quickly; will retry using slow method.`Slow method works well.
+
+Compiling it needs to set `RUSTFLAGS="-W invalid_reference_casting"`.
+
+Tips: Default version of cargo/rustc is 1.75.0 in Ubuntu 24.04, which is too low to compile. Although `rustc-1.82` is available, it can't be used out-of-box.
+So the most convenient way is just install `rustup` and then it will download correct version.
+
+---
+
+`Pillow==9.5.0`:
+There's only Windows version for cp312.
+
+If compiling from source, unlike official wheel, no libs would be bundled in the wheel. So it will link system libs.
+
+libtiff5 version is built on colab(Ubuntu 22.04); libtiff6 version is built on wsl Ubuntu 24.04, and also work well on Arch.
+
+Test by `ldd /path-to-site-packages/PIL/_imaging.cpython-312-x86_64-linux-gnu.so`

wheels-for-A1111-on-py312/pillow-9.5.0-libtiff5(colab)/pillow-9.5.0-cp312-cp312-linux_x86_64.whl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9a8cbb0b74970b66d1c387d5e1ec5245a7138603520aad7df4e0477df6d73b0a
+size 1210155

wheels-for-A1111-on-py312/pillow-9.5.0-libtiff6/pillow-9.5.0-cp312-cp312-linux_x86_64.whl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d60841d6fecdbe041b3644fd256e1c78815f7edc255348b4b891937f6a7ad0c6
+size 1259183

wheels-for-A1111-on-py312/scikit_image-0.21.0-cp312-cp312-linux_x86_64.whl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4e8e62ba31ca92f914011f9f957614af176d1dd7f210403101b47689fb7c4d01
+size 14235553

wheels-for-A1111-on-py312/scikit_image-0.21.0-cp312-cp312-win_amd64.whl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f3f685d68ad2e01c3895bac904930e3ba59b52d5ad2eeb1b02e8198023eef307
+size 12569642

wheels-for-A1111-on-py312/tokenizers-0.13.3-cp312-cp312-linux_x86_64.whl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7c64ec832ca4553b517e3645e33cbd4e1a84649d9d47d266ed1197c5eecce0a9
+size 4494204

wheels-for-A1111-on-py312/tokenizers-0.13.3-cp312-cp312-win_amd64.whl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:51fe938720674a722f190d2a7a7be11759964fc383b58ff326de426d02842068
+size 3583717