Merge branch 'master' into feat/ModelCheckpointException

vsey · web-flow · commit f2cd64fdda9d · 2025-06-25T19:17:08.000+02:00
diff --git a/.azure/gpu-tests-fabric.yml b/.azure/gpu-tests-fabric.yml
@@ -100,6 +100,12 @@ jobs:
           pip list
         displayName: "Image info & NVIDIA"
 
+      - bash: |
+          python .actions/assistant.py replace_oldest_ver
+          pip install "cython<3.0" wheel  # for compatibility
+        condition: contains(variables['Agent.JobName'], 'oldest')
+        displayName: "setting oldest dependencies"
+
       - bash: |
           PYTORCH_VERSION=$(python -c "import torch; print(torch.__version__.split('+')[0])")
           pip install -q wget packaging
diff --git a/.azure/gpu-tests-pytorch.yml b/.azure/gpu-tests-pytorch.yml
@@ -104,6 +104,12 @@ jobs:
           pip list
         displayName: "Image info & NVIDIA"
 
+      - bash: |
+          python .actions/assistant.py replace_oldest_ver
+          pip install "cython<3.0" wheel  # for compatibility
+        condition: contains(variables['Agent.JobName'], 'oldest')
+        displayName: "setting oldest dependencies"
+
       - bash: |
           PYTORCH_VERSION=$(python -c "import torch; print(torch.__version__.split('+')[0])")
           pip install -q wget packaging
diff --git a/requirements/fabric/base.txt b/requirements/fabric/base.txt
@@ -4,5 +4,5 @@
 torch >=2.1.0, <2.8.0
 fsspec[http] >=2022.5.0, <2025.6.0
 packaging >=20.0, <=25.0
-typing-extensions >=4.4.0, <4.15.0
+typing-extensions >=4.5.0, <4.15.0
 lightning-utilities >=0.10.0, <0.15.0
diff --git a/requirements/fabric/examples.txt b/requirements/fabric/examples.txt
@@ -3,4 +3,3 @@
 
 torchvision >=0.16.0, <0.23.0
 torchmetrics >=0.10.0, <1.8.0
-lightning-utilities >=0.8.0, <0.15.0
diff --git a/requirements/fabric/strategies.txt b/requirements/fabric/strategies.txt
@@ -5,6 +5,6 @@
 
 # note: is a bug around 0.10 with `MPS_Accelerator must implement all abstract methods`
 #  shall be resolved by https://github.com/microsoft/DeepSpeed/issues/4372
-deepspeed >=0.8.2, <=0.9.3; platform_system != "Windows" and platform_system != "Darwin"  # strict
+deepspeed >=0.9.3, <=0.9.3; platform_system != "Windows" and platform_system != "Darwin"  # strict
 # skip bitsandbytes==0.46, due to ValueError: too many values to unpack (expected 2)
 bitsandbytes >=0.45.2,!=0.46,<0.47.0; platform_system != "Darwin"
diff --git a/requirements/pytorch/base.txt b/requirements/pytorch/base.txt
@@ -3,9 +3,9 @@
 
 torch >=2.1.0, <2.8.0
 tqdm >=4.57.0, <4.68.0
-PyYAML >=5.4, <6.1.0
+PyYAML >5.4, <6.1.0
 fsspec[http] >=2022.5.0, <2025.6.0
-torchmetrics >=0.7.0, <1.8.0
+torchmetrics >0.7.0, <1.8.0
 packaging >=20.0, <=25.0
-typing-extensions >=4.4.0, <4.15.0
+typing-extensions >=4.5.0, <4.15.0
 lightning-utilities >=0.10.0, <0.15.0
diff --git a/requirements/pytorch/examples.txt b/requirements/pytorch/examples.txt
@@ -5,4 +5,3 @@ requests <2.33.0
 torchvision >=0.16.0, <0.23.0
 ipython[all] <8.19.0
 torchmetrics >=0.10.0, <1.8.0
-lightning-utilities >=0.8.0, <0.15.0
diff --git a/requirements/pytorch/strategies.txt b/requirements/pytorch/strategies.txt
@@ -3,4 +3,4 @@
 
 # note: is a bug around 0.10 with `MPS_Accelerator must implement all abstract methods`
 #  shall be resolved by https://github.com/microsoft/DeepSpeed/issues/4372
-deepspeed >=0.8.2, <=0.9.3; platform_system != "Windows" and platform_system != "Darwin"  # strict
+deepspeed >=0.9.3, <=0.9.3; platform_system != "Windows" and platform_system != "Darwin"  # strict
diff --git a/src/version.info b/src/version.info
@@ -1 +1 @@
-2.5.1rc2
+2.6.0dev
diff --git a/tests/tests_fabric/utilities/test_distributed.py b/tests/tests_fabric/utilities/test_distributed.py
@@ -105,8 +105,8 @@ def _test_all_reduce(strategy):
         assert result is tensor  # inplace
 
 
-# flaky with "process 0 terminated with signal SIGABRT" (GLOO)
-@pytest.mark.flaky(reruns=3, only_rerun="torch.multiprocessing.spawn.ProcessExitedException")
+# flaky with "torch.multiprocessing.spawn.ProcessExitedException: process 0 terminated with signal SIGABRT" (GLOO)
+@pytest.mark.flaky(reruns=3)
 @RunIf(skip_windows=True)
 @pytest.mark.parametrize(
     "process",
diff --git a/tests/tests_pytorch/callbacks/test_stochastic_weight_avg.py b/tests/tests_pytorch/callbacks/test_stochastic_weight_avg.py
@@ -354,8 +354,8 @@ def test_swa_resume_training_from_checkpoint_custom_scheduler(tmp_path, crash_on
 
 
 @RunIf(skip_windows=True)
-# flaky with "process 0 terminated with signal SIGABRT" (GLOO)
-@pytest.mark.flaky(reruns=3, only_rerun="torch.multiprocessing.spawn.ProcessExitedException")
+# flaky with "torch.multiprocessing.spawn.ProcessExitedException: process 0 terminated with signal SIGABRT" (GLOO)
+@pytest.mark.flaky(reruns=3)
 def test_swa_resume_training_from_checkpoint_ddp(tmp_path):
     model = SwaTestModel(crash_on_epoch=3)
     resume_model = SwaTestModel()
diff --git a/tests/tests_pytorch/core/test_results.py b/tests/tests_pytorch/core/test_results.py
@@ -49,8 +49,8 @@ def result_reduce_ddp_fn(strategy):
     assert actual.item() == dist.get_world_size()
 
 
-# flaky with "process 0 terminated with signal SIGABRT"
-@pytest.mark.flaky(reruns=3, only_rerun="torch.multiprocessing.spawn.ProcessExitedException")
+# flaky with "torch.multiprocessing.spawn.ProcessExitedException: process 0 terminated with signal SIGABRT"
+@pytest.mark.flaky(reruns=3)
 @RunIf(skip_windows=True)
 def test_result_reduce_ddp():
     spawn_launch(result_reduce_ddp_fn, [torch.device("cpu")] * 2)

Original file line number	Diff line number	Diff line change
`@@ -3,4 +3,3 @@`
`3`	`3`
`4`	`4`	`torchvision >=0.16.0, <0.23.0`
`5`	`5`	`torchmetrics >=0.10.0, <1.8.0`
`6`		`-lightning-utilities >=0.8.0, <0.15.0`