[CI][RFC] Replace black-jupyter with ruff-format (#37210)

Co-authored-by: Alison Shao <a.shao@wustl.edu>

test/manual/models/test_clip_models.py

@@ -29,7 +29,6 @@ TORCH_DTYPES = [torch.float16]
 
 
 class TestClipModels(unittest.TestCase):
-
     @classmethod
     def setUpClass(cls):
         mp.set_start_method("spawn", force=True)
@@ -61,12 +60,12 @@ class TestClipModels(unittest.TestCase):
         )
         print("text similarity diff", abs(text_similarity - 1))
         print("image similarity diff", abs(image_similarity - 1))
-        assert torch.all(
-            abs(text_similarity - 1) < prefill_tolerance
-        ), "embeddings are not all close"
-        assert torch.all(
-            abs(image_similarity - 1) < prefill_tolerance
-        ), "embeddings are not all close"
+        assert torch.all(abs(text_similarity - 1) < prefill_tolerance), (
+            "embeddings are not all close"
+        )
+        assert torch.all(abs(image_similarity - 1) < prefill_tolerance), (
+            "embeddings are not all close"
+        )
 
     def test_accuracy(self):
         for model, prefill_tolerance in MODELS:

test/manual/models/test_gme_qwen_models.py

@@ -64,16 +64,16 @@ class TestQmeQwenModels(CustomTestCase):
             hf_text_embeddings.embed_logits[0], srt_text_embeddings.embed_logits[0]
         )
         print("texts similarity diff", abs(similarity - 1))
-        assert torch.all(
-            abs(similarity - 1) < prefill_tolerance
-        ), "embeddings are not all close"
+        assert torch.all(abs(similarity - 1) < prefill_tolerance), (
+            "embeddings are not all close"
+        )
         similarity = get_similarities(
             hf_image_embeddings.embed_logits[0], srt_image_embeddings.embed_logits[0]
         )
         print("images similarity diff", abs(similarity - 1))
-        assert torch.all(
-            abs(similarity - 1) < prefill_tolerance
-        ), "embeddings are not all close"
+        assert torch.all(abs(similarity - 1) < prefill_tolerance), (
+            "embeddings are not all close"
+        )
 
     def test_accuracy(self):
         for model, prefill_tolerance in MODELS:

test/manual/models/test_kimi_k2_models.py

@@ -61,7 +61,7 @@ class TestKimiK2Thinking(CustomTestCase):
 
         if is_in_ci():
             write_github_step_summary(
-                f"### test_gsm8k (Kimi-K2-Thinking)\n" f'{metrics["score"]=:.3f}\n'
+                f'### test_gsm8k (Kimi-K2-Thinking)\n{metrics["score"]=:.3f}\n'
             )
             self.assertGreater(metrics["score"], 0.95)
 

test/manual/models/test_mistral_large3_basic.py

@@ -67,7 +67,7 @@ class TestMistralLarge3Basic(CustomTestCase):
 
         if is_in_ci():
             write_github_step_summary(
-                f"### test_gsm8k (mistral-large-3)\n" f'{metrics["score"]=:.3f}\n'
+                f'### test_gsm8k (mistral-large-3)\n{metrics["score"]=:.3f}\n'
             )
             self.assertGreater(metrics["score"], 0.90)
 
@@ -79,7 +79,7 @@ class TestMistralLarge3Basic(CustomTestCase):
 
         if is_in_ci():
             write_github_step_summary(
-                f"### test_bs_1_speed (mistral-large-3)\n" f"{speed=:.2f} token/s\n"
+                f"### test_bs_1_speed (mistral-large-3)\n{speed=:.2f} token/s\n"
             )
             self.assertGreater(speed, 50)
 

test/manual/models/test_qwen3_asr.py

@@ -156,9 +156,9 @@ async def _stream_websocket_async(
 
     async with websockets.connect(websocket_url) as websocket:
         created = json.loads(await websocket.recv())
-        assert (
-            created.get("type") == "session.created"
-        ), f"expected session.created, got {created!r}"
+        assert created.get("type") == "session.created", (
+            f"expected session.created, got {created!r}"
+        )
         session_id = created["session"]["id"]
 
         transcription_cfg = {"model": "qwen3-asr"}
@@ -200,9 +200,9 @@ async def _stream_websocket_async(
                 if t == "conversation.item.input_audio_transcription.delta":
                     deltas.append(resp["delta"])
                 elif t == "conversation.item.input_audio_transcription.completed":
-                    assert (
-                        "usage" in resp
-                    ), f"transcription.completed missing required usage field: {resp!r}"
+                    assert "usage" in resp, (
+                        f"transcription.completed missing required usage field: {resp!r}"
+                    )
                     assert resp["usage"].get("type") == "duration", resp["usage"]
                     completed_msg.update(resp)
                     return
@@ -317,7 +317,7 @@ class TestQwen3ASRTranscription(CustomTestCase):
             self.assertEqual(
                 results[0],
                 results[i],
-                f"Request {i+1} differs from first request",
+                f"Request {i + 1} differs from first request",
             )
         print(f"[Consistency] All 3 requests match: {results[0][:80]}...")