[From pretrained] Don't download .safetensors files if safetensors is… (#2057)

* [From pretrained] Don't download .safetensors files if safetensors is not available

* tests

* tests

* up

src/diffusers/pipelines/pipeline_utils.py

@@ -514,7 +514,11 @@ class DiffusionPipeline(ConfigMixin):
                 if is_safetensors_compatible(info):
                     ignore_patterns.append("*.bin")
                 else:
+                    # as a safety mechanism we also don't download safetensors if
+                    # not all safetensors files are there
                     ignore_patterns.append("*.safetensors")
+            else:
+                ignore_patterns.append("*.safetensors")
 
             # download all allow_patterns
             cached_folder = snapshot_download(

tests/test_pipelines.py

@@ -299,6 +299,16 @@ class CustomPipelineTests(unittest.TestCase):
 
 
 class PipelineFastTests(unittest.TestCase):
+    def tearDown(self):
+        # clean up the VRAM after each test
+        super().tearDown()
+        gc.collect()
+        torch.cuda.empty_cache()
+
+        import diffusers
+
+        diffusers.utils.import_utils._safetensors_available = True
+
     def dummy_image(self):
         batch_size = 1
         num_channels = 3
@@ -567,6 +577,50 @@ class PipelineFastTests(unittest.TestCase):
             assert pipeline.scheduler is not None
             assert pipeline.feature_extractor is not None
 
+    def test_no_pytorch_download_when_doing_safetensors(self):
+        # by default we don't download
+        with tempfile.TemporaryDirectory() as tmpdirname:
+            _ = StableDiffusionPipeline.from_pretrained(
+                "hf-internal-testing/diffusers-stable-diffusion-tiny-all", cache_dir=tmpdirname
+            )
+
+            path = os.path.join(
+                tmpdirname,
+                "models--hf-internal-testing--diffusers-stable-diffusion-tiny-all",
+                "snapshots",
+                "07838d72e12f9bcec1375b0482b80c1d399be843",
+                "unet",
+            )
+            # safetensors exists
+            assert os.path.exists(os.path.join(path, "diffusion_pytorch_model.safetensors"))
+            # pytorch does not
+            assert not os.path.exists(os.path.join(path, "diffusion_pytorch_model.bin"))
+
+    def test_no_safetensors_download_when_doing_pytorch(self):
+        # mock diffusers safetensors not available
+        import diffusers
+
+        diffusers.utils.import_utils._safetensors_available = False
+
+        with tempfile.TemporaryDirectory() as tmpdirname:
+            _ = StableDiffusionPipeline.from_pretrained(
+                "hf-internal-testing/diffusers-stable-diffusion-tiny-all", cache_dir=tmpdirname
+            )
+
+            path = os.path.join(
+                tmpdirname,
+                "models--hf-internal-testing--diffusers-stable-diffusion-tiny-all",
+                "snapshots",
+                "07838d72e12f9bcec1375b0482b80c1d399be843",
+                "unet",
+            )
+            # safetensors does not exists
+            assert not os.path.exists(os.path.join(path, "diffusion_pytorch_model.safetensors"))
+            # pytorch does
+            assert os.path.exists(os.path.join(path, "diffusion_pytorch_model.bin"))
+
+        diffusers.utils.import_utils._safetensors_available = True
+
     def test_optional_components(self):
         unet = self.dummy_cond_unet()
         pndm = PNDMScheduler.from_config("hf-internal-testing/tiny-stable-diffusion-torch", subfolder="scheduler")