dev-python/compressed-tensors
Library for compressed safetensors of neural network models
ChangeLog
commit 1c9de71084f5bcc49ebe2866f2f423cb905f3d02
Author: Ivan S. Titov <iohann.s.titov@gmail.com>
Date: Fri Aug 14 12:58:26 2026 +0200
dev-python/compressed-tensors: complete isolated build checks
The exact 0.18.0 build metadata requires wheel. Three selected tests load remote
model configurations even in offline mode, so exclude only those parameterized cases
while retaining the rest of the quantization configuration coverage.
commit 1144a656f3d9de6b3085d7ca9669395fe56a00a1
Author: Ivan S. Titov <iohann.s.titov@gmail.com>
Date: Sat Aug 8 09:13:15 2026 +0200
dev-python/compressed-tensors: drop 0.17.1-r1
commit 0f896386869c365fed95080ff021d4c3a12bf98d
Author: Ivan S. Titov <iohann.s.titov@gmail.com>
Date: Sat Aug 8 09:13:09 2026 +0200
dev-python/compressed-tensors: add 0.18.0
commit d69c676c2b50b6e344ff9a71b9c80857687130a4
Author: Ivan S. Titov <iohann.s.titov@gmail.com>
Date: Tue Aug 4 19:33:18 2026 +0200
dev-python/compressed-tensors: fix dependencies and tests in 0.17.1
commit 2f14c0c7bbbc6308130a816d94714a5e67e58a85
Author: Ivan S. Titov <iohann.s.titov@gmail.com>
Date: Tue Aug 4 19:33:10 2026 +0200
dev-python/compressed-tensors: fix dependencies and tests in 0.17.0
commit b9b9bd1d8127b57d9bce35f185092fb0f1a34430
Author: Ivan S. Titov <iohann.s.titov@gmail.com>
Date: Tue Aug 4 19:33:03 2026 +0200
dev-python/compressed-tensors: fix dependencies and tests in 0.15.0.1
commit 77eacb348475287ca975346d1f5d80e9f71e1899
Author: Ivan S. Titov <iohann.s.titov@gmail.com>
Date: Wed Jul 8 10:47:44 2026 +0200
dev-python/compressed-tensors: add 0.17.1
commit 303edf149bbd1276d664c2ca5bf94a93f264af20
Author: Ivan S. Titov <iohann.s.titov@gmail.com>
Date: Sat Jun 13 13:50:42 2026 +0200
dev-python/compressed-tensors: add 0.17.0
Required by vllm 0.23.0, whose requirements/common.txt pins
compressed-tensors==0.17.0 (this is the long-held bump that only moves
alongside a vllm re-pin). Upstream raised the torch floor to
torch>=2.10.0 (was >=1.7.0); transformers/pydantic/loguru unchanged.
0.15.0.1 is kept for the vllm 0.21.0/0.22.1 ~0.15.0.1 pin.
commit d8b19db7a3b4b3037a76f5aa7fc6f837120d4ab4
Author: Ivan S. Titov <iohann.s.titov@gmail.com>
Date: Wed May 13 14:34:33 2026 +0200
dev-python/compressed-tensors: disable py3.11
commit b030d0e613b59844434cb1154ee81a9b7659db9c
Author: Ivan S. Titov <iohann.s.titov@gmail.com>
Date: Sun May 10 18:38:21 2026 +0200
dev-python/compressed-tensors: switch to DISTUTILS_SINGLE_IMPL
sci-ml/{pytorch,transformers} are SINGLE_IMPL; multi-impl consumer with
bare $ on them produces python_targets_python3_*(-)?
that the single-impl children can't expose. Single-impl deps split out
to bare $; remaining multi-impl deps wrapped in
python_gen_cond_dep.
commit 1f6959a49776699a3b67eda3b6c4656a64ac12c5
Author: Ivan S. Titov <iohann.s.titov@gmail.com>
Date: Thu May 7 00:06:52 2026 +0200
dev-python/compressed-tensors: new package
Pure-Python wrapper over hf safetensors that adds compressed-tensor
serialization. Required by dev-python/vllm.
NonsolvableDepsInStable is policy noise — ~amd64-only because pytorch
and transformers are ~amd64-only too.
Author: Ivan S. Titov <iohann.s.titov@gmail.com>
Date: Fri Aug 14 12:58:26 2026 +0200
dev-python/compressed-tensors: complete isolated build checks
The exact 0.18.0 build metadata requires wheel. Three selected tests load remote
model configurations even in offline mode, so exclude only those parameterized cases
while retaining the rest of the quantization configuration coverage.
commit 1144a656f3d9de6b3085d7ca9669395fe56a00a1
Author: Ivan S. Titov <iohann.s.titov@gmail.com>
Date: Sat Aug 8 09:13:15 2026 +0200
dev-python/compressed-tensors: drop 0.17.1-r1
commit 0f896386869c365fed95080ff021d4c3a12bf98d
Author: Ivan S. Titov <iohann.s.titov@gmail.com>
Date: Sat Aug 8 09:13:09 2026 +0200
dev-python/compressed-tensors: add 0.18.0
commit d69c676c2b50b6e344ff9a71b9c80857687130a4
Author: Ivan S. Titov <iohann.s.titov@gmail.com>
Date: Tue Aug 4 19:33:18 2026 +0200
dev-python/compressed-tensors: fix dependencies and tests in 0.17.1
commit 2f14c0c7bbbc6308130a816d94714a5e67e58a85
Author: Ivan S. Titov <iohann.s.titov@gmail.com>
Date: Tue Aug 4 19:33:10 2026 +0200
dev-python/compressed-tensors: fix dependencies and tests in 0.17.0
commit b9b9bd1d8127b57d9bce35f185092fb0f1a34430
Author: Ivan S. Titov <iohann.s.titov@gmail.com>
Date: Tue Aug 4 19:33:03 2026 +0200
dev-python/compressed-tensors: fix dependencies and tests in 0.15.0.1
commit 77eacb348475287ca975346d1f5d80e9f71e1899
Author: Ivan S. Titov <iohann.s.titov@gmail.com>
Date: Wed Jul 8 10:47:44 2026 +0200
dev-python/compressed-tensors: add 0.17.1
commit 303edf149bbd1276d664c2ca5bf94a93f264af20
Author: Ivan S. Titov <iohann.s.titov@gmail.com>
Date: Sat Jun 13 13:50:42 2026 +0200
dev-python/compressed-tensors: add 0.17.0
Required by vllm 0.23.0, whose requirements/common.txt pins
compressed-tensors==0.17.0 (this is the long-held bump that only moves
alongside a vllm re-pin). Upstream raised the torch floor to
torch>=2.10.0 (was >=1.7.0); transformers/pydantic/loguru unchanged.
0.15.0.1 is kept for the vllm 0.21.0/0.22.1 ~0.15.0.1 pin.
commit d8b19db7a3b4b3037a76f5aa7fc6f837120d4ab4
Author: Ivan S. Titov <iohann.s.titov@gmail.com>
Date: Wed May 13 14:34:33 2026 +0200
dev-python/compressed-tensors: disable py3.11
commit b030d0e613b59844434cb1154ee81a9b7659db9c
Author: Ivan S. Titov <iohann.s.titov@gmail.com>
Date: Sun May 10 18:38:21 2026 +0200
dev-python/compressed-tensors: switch to DISTUTILS_SINGLE_IMPL
sci-ml/{pytorch,transformers} are SINGLE_IMPL; multi-impl consumer with
bare $ on them produces python_targets_python3_*(-)?
that the single-impl children can't expose. Single-impl deps split out
to bare $; remaining multi-impl deps wrapped in
python_gen_cond_dep.
commit 1f6959a49776699a3b67eda3b6c4656a64ac12c5
Author: Ivan S. Titov <iohann.s.titov@gmail.com>
Date: Thu May 7 00:06:52 2026 +0200
dev-python/compressed-tensors: new package
Pure-Python wrapper over hf safetensors that adds compressed-tensor
serialization. Required by dev-python/vllm.
NonsolvableDepsInStable is policy noise — ~amd64-only because pytorch
and transformers are ~amd64-only too.


View
Download
Browse