Script 'mail_helper' called by obssrc
Hello community,
here is the log from the commit of package python-compressed-tensors for
openSUSE:Factory checked in at 2026-08-12 16:12:26
++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
Comparing /work/SRC/openSUSE:Factory/python-compressed-tensors (Old)
and /work/SRC/openSUSE:Factory/.python-compressed-tensors.new.17972 (New)
++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
Package is "python-compressed-tensors"
Wed Aug 12 16:12:26 2026 rev:3 rq:1370784 version:0.18.0
Changes:
--------
---
/work/SRC/openSUSE:Factory/python-compressed-tensors/python-compressed-tensors.changes
2026-07-18 22:26:30.757159978 +0200
+++
/work/SRC/openSUSE:Factory/.python-compressed-tensors.new.17972/python-compressed-tensors.changes
2026-08-12 16:13:38.177332721 +0200
@@ -1,0 +2,20 @@
+Wed Aug 12 05:22:58 UTC 2026 - Martin Pluskal <[email protected]>
+
+- Update to 0.18.0:
+ * Add QuantizationConfig.merge and TransformConfig.merge, allowing
+ compression of models that already carry a quantization config
+ * Only apply quantization schemes to Linear, Embedding and Attention
+ layers, and ignore non-linear MoE gate layers
+ * Add a Triton implementation of cast_to_fp4 and make FP4 packing
+ faster and less memory hungry
+ * Fix NVFP4 packing for float32 inputs
+ * Fix KV cache quantization for composite multimodal models
+ * Several offloading fixes: force offloading of buffers, fix disk
+ distributed module compression, fix tied transform weight
+ detection for offloaded models
+ * Fix MTP tensor copy from indexed checkpoints
+ * Validate block structure dimensions and length
+ * Deprecate actorder group
+- Add Requires on python-psutil, newly required by upstream
+
+-------------------------------------------------------------------
Old:
----
compressed_tensors-0.17.1.tar.gz
New:
----
compressed_tensors-0.18.0.tar.gz
++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
Other differences:
------------------
++++++ python-compressed-tensors.spec ++++++
--- /var/tmp/diff_new_pack.wHpo7G/_old 2026-08-12 16:13:38.809359417 +0200
+++ /var/tmp/diff_new_pack.wHpo7G/_new 2026-08-12 16:13:38.813359586 +0200
@@ -18,7 +18,7 @@
%{?sle15_python_module_pythons}
Name: python-compressed-tensors
-Version: 0.17.1
+Version: 0.18.0
Release: 0
Summary: Library for utilization of compressed safetensors of neural
network models
License: Apache-2.0
@@ -31,6 +31,7 @@
BuildRequires: fdupes
BuildRequires: python-rpm-macros
Requires: python-loguru
+Requires: python-psutil
Requires: python-pydantic >= 2.0
Requires: python-torch >= 2.10.0
Requires: python-transformers >= 4.45.0
@@ -56,10 +57,10 @@
%pyproject_install
%python_expand %fdupes %{buildroot}%{$python_sitelib}
-# Tests are not run: the test suite requires the transformers Python
-# module (not available in openSUSE:Factory) and most tests additionally
-# require a GPU (upstream states "most tests do require a GPU"), so even
-# an import smoke test fails without transformers in the build root.
+# Tests are not run: importing the module pulls in torch, which is built
+# for x86_64 only in openSUSE:Factory, so even an import smoke test would
+# make this noarch package fail to build on every other architecture.
+# Most of the test suite additionally requires a GPU.
%files %{python_files}
%license LICENSE
++++++ compressed_tensors-0.17.1.tar.gz -> compressed_tensors-0.18.0.tar.gz
++++++
++++ 7433 lines of diff (skipped)
