Script 'mail_helper' called by obssrc
Hello community,
here is the log from the commit of package python-transformers for
openSUSE:Factory checked in at 2026-10-01 22:21:03
++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
Comparing /work/SRC/openSUSE:Factory/python-transformers (Old)
and /work/SRC/openSUSE:Factory/.python-transformers.new.1253 (New)
++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
Package is "python-transformers"
Thu Oct 1 22:21:03 2026 rev:7 rq:1381945 version:5.18.0
Changes:
--------
--- /work/SRC/openSUSE:Factory/python-transformers/python-transformers.changes
2026-09-10 17:40:58.888673169 +0200
+++
/work/SRC/openSUSE:Factory/.python-transformers.new.1253/python-transformers.changes
2026-10-01 22:21:37.368896624 +0200
@@ -1,0 +2,23 @@
+Thu Oct 1 16:29:23 UTC 2026 - Martin Pluskal <[email protected]>
+
+- Update to 5.18.0:
+ * Add Nemotron 3 Diarization, NemotronH Omni, HyperCLOVAX
+ Vision V2 and the GTE bidirectional encoder
+ * huggingface-hub moves to >= 1.31.0, < 3.0 and the spec
+ follows; that also lifts the 2.0 exclusion which left 5.17.0
+ unimportable next to Factory's hub 2.0.0
+ * Breaking: the cache was reworked for mixed attention types
+ and the indexers layer_type mapping remapped
+ * Restore the legacy tensor-parallel initialization 5.16.0
+ removed, and unify device_mesh init for pipeline plus
+ tensor parallelism
+ * Fix from_pretrained dropping modules_to_save weights, RTDetr
+ and SEWD loading with random weights, MusicgenMelody losing
+ its conditioning at step 0 and AutoImageProcessor demanding
+ torchvision when only Pillow is installed
+ * Deprecate the use_mamba_kernels config flag and the min-max
+ pixels preprocessor argument
+ * Many more fixes and improvements; see upstream's release notes
+ for the full list
+
+-------------------------------------------------------------------
Old:
----
transformers-5.17.0.tar.gz
New:
----
transformers-5.18.0.tar.gz
++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
Other differences:
------------------
++++++ python-transformers.spec ++++++
--- /var/tmp/diff_new_pack.33l3wi/_old 2026-10-01 22:21:39.562988362 +0200
+++ /var/tmp/diff_new_pack.33l3wi/_new 2026-10-01 22:21:39.565988487 +0200
@@ -18,7 +18,7 @@
%bcond_without libalternatives
Name: python-transformers
-Version: 5.17.0
+Version: 5.18.0
Release: 0
Summary: State-of-the-art Machine Learning for JAX, PyTorch and
TensorFlow
License: Apache-2.0
@@ -33,13 +33,15 @@
BuildRequires: python-rpm-macros
Requires: alts
Requires: python-PyYAML >= 5.1
-Requires: python-huggingface-hub >= 1.5.0
+# dependency_versions_check.py enforces these upper bounds at import time, so
+# an undeclared tokenizers 0.24 or huggingface-hub 3.0 would install fine and
+# then fail on "import transformers". 5.17.0 already excluded Factory's 2.0.0.
+Requires: python-huggingface-hub < 3.0
+Requires: python-huggingface-hub >= 1.31.0
Requires: python-numpy >= 1.17
Requires: python-packaging >= 20.0
Requires: python-regex >= 2025.10.22
Requires: python-safetensors >= 0.8.0
-# dependency_versions_check.py enforces the upper bound at import time, so an
-# undeclared 0.24 would install fine and then fail on "import transformers".
Requires: python-tokenizers < 0.24.0
Requires: python-tokenizers >= 0.23.1
Requires: python-tqdm >= 4.60
@@ -47,7 +49,7 @@
BuildArch: noarch
# SECTION runtime dependencies (also needed for the %%check import test)
BuildRequires: %{python_module PyYAML >= 5.1}
-BuildRequires: %{python_module huggingface-hub >= 1.5.0}
+BuildRequires: %{python_module huggingface-hub >= 1.31.0}
BuildRequires: %{python_module numpy >= 1.17}
BuildRequires: %{python_module packaging >= 20.0}
BuildRequires: %{python_module regex >= 2025.10.22}
++++++ transformers-5.17.0.tar.gz -> transformers-5.18.0.tar.gz ++++++
/work/SRC/openSUSE:Factory/python-transformers/transformers-5.17.0.tar.gz
/work/SRC/openSUSE:Factory/.python-transformers.new.1253/transformers-5.18.0.tar.gz
differ: char 5, line 1
