temporary skip failing whisper tests

.github/workflows/linux.yml

@@ -477,7 +477,7 @@ jobs:
           - name: 'Whisper'
             cmd: 'tests/python_tests/test_whisper_pipeline.py tests/python_tests/test_whisper_pipeline_static.py'
             run_condition: ${{ fromJSON(needs.smart_ci.outputs.affected_components).whisper.test }}
-            timeout: 45
+            timeout: 145
           - name: 'Cacheopt E2E'
             cmd: 'tests/python_tests/test_kv_cache_eviction.py'
             run_condition: ${{ fromJSON(needs.smart_ci.outputs.affected_components).continuous_batching.test }}

.github/workflows/mac.yml

@@ -425,7 +425,7 @@ jobs:
           - name: 'Whisper'
             cmd: 'tests/python_tests/test_whisper_pipeline.py'
             run_condition: ${{ fromJSON(needs.smart_ci.outputs.affected_components).whisper.test }}
-            timeout: 45
+            timeout: 145
           - name: 'Cacheopt E2E'
             cmd: 'tests/python_tests/test_kv_cache_eviction.py'
             run_condition: ${{ fromJSON(needs.smart_ci.outputs.affected_components).continuous_batching.test }}

.github/workflows/windows.yml

@@ -536,7 +536,7 @@ jobs:
           - name: 'Whisper'
             cmd: 'tests/python_tests/test_whisper_pipeline.py tests/python_tests/test_whisper_pipeline_static.py'
             run_condition: ${{ fromJSON(needs.smart_ci.outputs.affected_components).whisper.test }}
-            timeout: 45
+            timeout: 145
           - name: 'Cacheopt E2E'
             cmd: 'tests/python_tests/test_kv_cache_eviction.py'
             run_condition: ${{ fromJSON(needs.smart_ci.outputs.affected_components).continuous_batching.test }}

tests/python_tests/test_whisper_pipeline.py

@@ -180,18 +180,29 @@ def run_genai(
 @functools.lru_cache(16)
 def get_whisper_dataset(language: str, long_form: bool) -> list:
     if not long_form:
-        ds = datasets.load_dataset(
-            "mozilla-foundation/common_voice_11_0",
-            language,
-            split="test",
-            streaming=True,
-            trust_remote_code=True,
-        )
+        # TODO: temporary download librispeech dataset until https://github.com/huggingface/datasets/issues/7647 is fixed
+        if language == "en":
+            ds = datasets.load_dataset(
+                "openslr/librispeech_asr",
+                split="test.other",
+                streaming=True,
+                trust_remote_code=True,
+            )
+        else:
+            ds = datasets.load_dataset(
+                "mozilla-foundation/common_voice_11_0",
+                language,
+                split="test",
+                streaming=False,
+                trust_remote_code=True,
+            )
     else:
         ds = datasets.load_dataset(
             "distil-whisper/meanwhile",
             split="test",
-            streaming=True,
+            # TODO: temporary download the whole dataset until is fixed https://github.com/huggingface/datasets/issues/7647
+            # return streaming back once the issue is resolved.
+            # streaming=True,
             trust_remote_code=True,
         )