REFACTOR rename SeedDatasetProvider.fetch_dataset to fetch_dataset_async

doc/code/datasets/4_dataset_coding.ipynb

@@ -67,7 +67,7 @@
     "    def dataset_name(self) -> str:\n",
     "        return \"dark_bench\"\n",
     "\n",
-    "    async def fetch_dataset(self, *, cache: bool = True) -> SeedDataset:\n",
+    "    async def fetch_dataset_async(self, *, cache: bool = True) -> SeedDataset:\n",
     "        # Fetch from HuggingFace\n",
     "        data = await self._fetch_from_huggingface(\n",
     "            dataset_name=\"apart/darkbench\",\n",

doc/code/datasets/4_dataset_coding.py

@@ -64,7 +64,7 @@ class SimpleDarkBench(_RemoteDatasetLoader):
     def dataset_name(self) -> str:
         return "dark_bench"
 
-    async def fetch_dataset(self, *, cache: bool = True) -> SeedDataset:
+    async def fetch_dataset_async(self, *, cache: bool = True) -> SeedDataset:
         # Fetch from HuggingFace
         data = await self._fetch_from_huggingface(
             dataset_name="apart/darkbench",

pyrit/datasets/seed_datasets/local/local_dataset_loader.py

@@ -53,7 +53,7 @@ def dataset_name(self) -> str:
         """Return the dataset name."""
         return self._dataset_name
 
-    async def fetch_dataset(self, *, cache: bool = True) -> SeedDataset:
+    async def fetch_dataset_async(self, *, cache: bool = True) -> SeedDataset:
         """
         Load the dataset from the local YAML file.
 

pyrit/datasets/seed_datasets/remote/aegis_ai_content_safety_dataset.py

@@ -120,7 +120,7 @@ def dataset_name(self) -> str:
         """Return the dataset name."""
         return "aegis_content_safety"
 
-    async def fetch_dataset(self, *, cache: bool = True) -> SeedDataset:
+    async def fetch_dataset_async(self, *, cache: bool = True) -> SeedDataset:
         """
         Fetch NVIDIA Aegis AI Content Safety dataset with optional filtering and return as SeedDataset.
 

pyrit/datasets/seed_datasets/remote/aya_redteaming_dataset.py

@@ -84,7 +84,7 @@ def dataset_name(self) -> str:
         """Return the dataset name."""
         return "aya_redteaming"
 
-    async def fetch_dataset(self, *, cache: bool = True) -> SeedDataset:
+    async def fetch_dataset_async(self, *, cache: bool = True) -> SeedDataset:
         """
         Fetch Aya Red-teaming dataset with optional filtering and return as SeedDataset.
 

pyrit/datasets/seed_datasets/remote/babelscape_alert_dataset.py

@@ -51,7 +51,7 @@ def dataset_name(self) -> str:
         """Return the dataset name."""
         return "babelscape_alert"
 
-    async def fetch_dataset(self, *, cache: bool = True) -> SeedDataset:
+    async def fetch_dataset_async(self, *, cache: bool = True) -> SeedDataset:
         """
         Fetch Babelscape ALERT dataset and return as SeedDataset.
 

pyrit/datasets/seed_datasets/remote/beaver_tails_dataset.py

@@ -52,7 +52,7 @@ def dataset_name(self) -> str:
         """Return the dataset name."""
         return "beaver_tails"
 
-    async def fetch_dataset(self, *, cache: bool = True) -> SeedDataset:
+    async def fetch_dataset_async(self, *, cache: bool = True) -> SeedDataset:
         """
         Fetch BeaverTails dataset from HuggingFace and return as SeedDataset.
 

pyrit/datasets/seed_datasets/remote/cbt_bench_dataset.py

@@ -52,7 +52,7 @@ def dataset_name(self) -> str:
         """Return the dataset name."""
         return "cbt_bench"
 
-    async def fetch_dataset(self, *, cache: bool = True) -> SeedDataset:
+    async def fetch_dataset_async(self, *, cache: bool = True) -> SeedDataset:
         """
         Fetch CBT-Bench dataset from HuggingFace and return as SeedDataset.
 

pyrit/datasets/seed_datasets/remote/ccp_sensitive_prompts_dataset.py

@@ -39,7 +39,7 @@ def dataset_name(self) -> str:
         """Return the dataset name."""
         return "ccp_sensitive_prompts"
 
-    async def fetch_dataset(self, *, cache: bool = True) -> SeedDataset:
+    async def fetch_dataset_async(self, *, cache: bool = True) -> SeedDataset:
         """
         Fetch CCP-sensitive prompts dataset and return as SeedDataset.
 

pyrit/datasets/seed_datasets/remote/comic_jailbreak_dataset.py

@@ -137,7 +137,7 @@ def dataset_name(self) -> str:
         """Return the dataset name."""
         return "comic_jailbreak"
 
-    async def fetch_dataset(self, *, cache: bool = True) -> SeedDataset:
+    async def fetch_dataset_async(self, *, cache: bool = True) -> SeedDataset:
         """
         Fetch ComicJailbreak dataset and return as SeedDataset of image+text pairs.
 

pyrit/datasets/seed_datasets/remote/darkbench_dataset.py

@@ -47,7 +47,7 @@ def dataset_name(self) -> str:
         """Return the dataset name."""
         return "dark_bench"
 
-    async def fetch_dataset(self, *, cache: bool = True) -> SeedDataset:
+    async def fetch_dataset_async(self, *, cache: bool = True) -> SeedDataset:
         """
         Fetch DarkBench dataset from HuggingFace and return as SeedDataset.
 

pyrit/datasets/seed_datasets/remote/equitymedqa_dataset.py

@@ -112,7 +112,7 @@ def dataset_name(self) -> str:
         """Return the dataset name."""
         return "equitymedqa"
 
-    async def fetch_dataset(self, *, cache: bool = True) -> SeedDataset:
+    async def fetch_dataset_async(self, *, cache: bool = True) -> SeedDataset:
         """
         Fetch EquityMedQA dataset and return as SeedDataset.
 

pyrit/datasets/seed_datasets/remote/forbidden_questions_dataset.py

@@ -46,7 +46,7 @@ def dataset_name(self) -> str:
         """Return the dataset name."""
         return "forbidden_questions"
 
-    async def fetch_dataset(self, *, cache: bool = True) -> SeedDataset:
+    async def fetch_dataset_async(self, *, cache: bool = True) -> SeedDataset:
         """
         Fetch Forbidden Questions dataset and return as SeedDataset.
 

pyrit/datasets/seed_datasets/remote/harmbench_dataset.py

@@ -49,7 +49,7 @@ def dataset_name(self) -> str:
         """Return the dataset name."""
         return "harmbench"
 
-    async def fetch_dataset(self, *, cache: bool = True) -> SeedDataset:
+    async def fetch_dataset_async(self, *, cache: bool = True) -> SeedDataset:
         """
         Fetch HarmBench dataset and return as SeedDataset.
 

pyrit/datasets/seed_datasets/remote/harmbench_multimodal_dataset.py

@@ -75,7 +75,7 @@ def dataset_name(self) -> str:
         """Return the dataset name."""
         return "harmbench_multimodal"
 
-    async def fetch_dataset(self, *, cache: bool = True) -> SeedDataset:
+    async def fetch_dataset_async(self, *, cache: bool = True) -> SeedDataset:
         """
         Fetch HarmBench multimodal examples and return as SeedDataset.
 

pyrit/datasets/seed_datasets/remote/harmful_qa_dataset.py

@@ -46,7 +46,7 @@ def dataset_name(self) -> str:
         """Return the dataset name."""
         return "harmful_qa"
 
-    async def fetch_dataset(self, *, cache: bool = True) -> SeedDataset:
+    async def fetch_dataset_async(self, *, cache: bool = True) -> SeedDataset:
         """
         Fetch HarmfulQA dataset from HuggingFace and return as SeedDataset.
 

pyrit/datasets/seed_datasets/remote/jbb_behaviors_dataset.py

@@ -45,7 +45,7 @@ def dataset_name(self) -> str:
         """Return the dataset name."""
         return "jbb_behaviors"
 
-    async def fetch_dataset(self, *, cache: bool = True) -> SeedDataset:
+    async def fetch_dataset_async(self, *, cache: bool = True) -> SeedDataset:
         """
         Fetch JBB-Behaviors dataset and return as SeedDataset.
 

pyrit/datasets/seed_datasets/remote/librai_do_not_answer_dataset.py

@@ -40,7 +40,7 @@ def dataset_name(self) -> str:
         """Return the dataset name."""
         return "librai_do_not_answer"
 
-    async def fetch_dataset(self, *, cache: bool = True) -> SeedDataset:
+    async def fetch_dataset_async(self, *, cache: bool = True) -> SeedDataset:
         """
         Fetch LibrAI Do Not Answer dataset and return as SeedDataset.
 

pyrit/datasets/seed_datasets/remote/llm_latent_adversarial_training_dataset.py

@@ -39,7 +39,7 @@ def dataset_name(self) -> str:
         """Return the dataset name."""
         return "llm_lat_harmful"
 
-    async def fetch_dataset(self, *, cache: bool = True) -> SeedDataset:
+    async def fetch_dataset_async(self, *, cache: bool = True) -> SeedDataset:
         """
         Fetch LLM-LAT harmful dataset and return as SeedDataset.
 

pyrit/datasets/seed_datasets/remote/medsafetybench_dataset.py

@@ -73,7 +73,7 @@ def dataset_name(self) -> str:
         """Return the dataset name."""
         return "medsafetybench"
 
-    async def fetch_dataset(self, *, cache: bool = True) -> SeedDataset:
+    async def fetch_dataset_async(self, *, cache: bool = True) -> SeedDataset:
         """
         Fetch MedSafetyBench dataset and return as SeedDataset.
 

pyrit/datasets/seed_datasets/remote/mlcommons_ailuminate_dataset.py

@@ -65,7 +65,7 @@ def dataset_name(self) -> str:
         """Return the dataset name."""
         return "mlcommons_ailuminate"
 
-    async def fetch_dataset(self, *, cache: bool = True) -> SeedDataset:
+    async def fetch_dataset_async(self, *, cache: bool = True) -> SeedDataset:
         """
         Fetch AILuminate dataset and return as SeedDataset.
 

pyrit/datasets/seed_datasets/remote/multilingual_vulnerability_dataset.py

@@ -40,7 +40,7 @@ def dataset_name(self) -> str:
         """Return the dataset name."""
         return "multilingual_vulnerability"
 
-    async def fetch_dataset(self, *, cache: bool = True) -> SeedDataset:
+    async def fetch_dataset_async(self, *, cache: bool = True) -> SeedDataset:
         """
         Fetch Multilingual Vulnerability dataset and return as SeedDataset.
 

pyrit/datasets/seed_datasets/remote/or_bench_dataset.py

@@ -39,7 +39,7 @@ def __init__(self, *, split: str = "train") -> None:
         """
         self.split = split
 
-    async def fetch_dataset(self, *, cache: bool = True) -> SeedDataset:
+    async def fetch_dataset_async(self, *, cache: bool = True) -> SeedDataset:
         """
         Fetch OR-Bench dataset from HuggingFace and return as SeedDataset.
 

pyrit/datasets/seed_datasets/remote/pku_safe_rlhf_dataset.py

@@ -72,7 +72,7 @@ def dataset_name(self) -> str:
         """Return the dataset name."""
         return "pku_safe_rlhf"
 
-    async def fetch_dataset(self, *, cache: bool = True) -> SeedDataset:
+    async def fetch_dataset_async(self, *, cache: bool = True) -> SeedDataset:
         """
         Fetch PKU-SafeRLHF dataset and return as SeedDataset.
 

pyrit/datasets/seed_datasets/remote/promptintel_dataset.py

@@ -305,7 +305,7 @@ def _convert_record_to_seed_prompt(self, record: dict[str, Any]) -> Optional[See
             metadata=metadata,
         )
 
-    async def fetch_dataset(self, *, cache: bool = True) -> SeedDataset:
+    async def fetch_dataset_async(self, *, cache: bool = True) -> SeedDataset:
         """
         Fetch prompts from the PromptIntel API and return as a SeedDataset.
 

pyrit/datasets/seed_datasets/remote/red_team_social_bias_dataset.py

@@ -42,7 +42,7 @@ def dataset_name(self) -> str:
         """Return the dataset name."""
         return "red_team_social_bias"
 
-    async def fetch_dataset(self, *, cache: bool = True) -> SeedDataset:
+    async def fetch_dataset_async(self, *, cache: bool = True) -> SeedDataset:
         """
         Fetch Red Team Social Bias dataset and return as SeedDataset.
 

pyrit/datasets/seed_datasets/remote/remote_dataset_loader.py

@@ -49,7 +49,7 @@ class _RemoteDatasetLoader(SeedDatasetProvider, ABC):
     - HuggingFace Hub
 
     Subclasses must implement:
-    - fetch_dataset(): Fetch and return the dataset as a SeedDataset
+    - fetch_dataset_async(): Fetch and return the dataset as a SeedDataset
     - dataset_name property: Human-readable name for the dataset
     """
 

pyrit/datasets/seed_datasets/remote/salad_bench_dataset.py

@@ -67,7 +67,7 @@ def _parse_category(category: str) -> str:
         """
         return re.sub(r"^O\d+:\s*", "", category)
 
-    async def fetch_dataset(self, *, cache: bool = True) -> SeedDataset:
+    async def fetch_dataset_async(self, *, cache: bool = True) -> SeedDataset:
         """
         Fetch SALAD-Bench dataset from HuggingFace and return as SeedDataset.
 

pyrit/datasets/seed_datasets/remote/simple_safety_tests_dataset.py

@@ -46,7 +46,7 @@ def dataset_name(self) -> str:
         """Return the dataset name."""
         return "simple_safety_tests"
 
-    async def fetch_dataset(self, *, cache: bool = True) -> SeedDataset:
+    async def fetch_dataset_async(self, *, cache: bool = True) -> SeedDataset:
         """
         Fetch SimpleSafetyTests dataset from HuggingFace and return as SeedDataset.
 

pyrit/datasets/seed_datasets/remote/sorry_bench_dataset.py

@@ -140,7 +140,7 @@ def dataset_name(self) -> str:
         """Return the dataset name."""
         return "sorry_bench"
 
-    async def fetch_dataset(self, *, cache: bool = True) -> SeedDataset:
+    async def fetch_dataset_async(self, *, cache: bool = True) -> SeedDataset:
         """
         Fetch Sorry-Bench dataset and return as SeedDataset.
 

pyrit/datasets/seed_datasets/remote/sosbench_dataset.py

@@ -40,7 +40,7 @@ def dataset_name(self) -> str:
         """Return the dataset name."""
         return "sosbench"
 
-    async def fetch_dataset(self, *, cache: bool = True) -> SeedDataset:
+    async def fetch_dataset_async(self, *, cache: bool = True) -> SeedDataset:
         """
         Fetch SOSBench dataset and return as SeedDataset.
 

pyrit/datasets/seed_datasets/remote/tdc23_redteaming_dataset.py

@@ -40,7 +40,7 @@ def dataset_name(self) -> str:
         """Return the dataset name."""
         return "tdc23_redteaming"
 
-    async def fetch_dataset(self, *, cache: bool = True) -> SeedDataset:
+    async def fetch_dataset_async(self, *, cache: bool = True) -> SeedDataset:
         """
         Fetch TDC23-RedTeaming dataset and return as SeedDataset.
 

pyrit/datasets/seed_datasets/remote/toxic_chat_dataset.py

@@ -83,7 +83,7 @@ def _extract_harm_categories(self, item: dict[str, Any]) -> list[str]:
 
         return categories
 
-    async def fetch_dataset(self, *, cache: bool = True) -> SeedDataset:
+    async def fetch_dataset_async(self, *, cache: bool = True) -> SeedDataset:
         """
         Fetch ToxicChat dataset from HuggingFace and return as SeedDataset.
 

pyrit/datasets/seed_datasets/remote/transphobia_awareness_dataset.py

@@ -52,7 +52,7 @@ def dataset_name(self) -> str:
         """Return the dataset name."""
         return "transphobia_awareness"
 
-    async def fetch_dataset(self, *, cache: bool = True) -> SeedDataset:
+    async def fetch_dataset_async(self, *, cache: bool = True) -> SeedDataset:
         """
         Fetch Transphobia-Awareness dataset and return as SeedDataset.
 

pyrit/datasets/seed_datasets/remote/visual_leak_bench_dataset.py

@@ -119,7 +119,7 @@ def dataset_name(self) -> str:
         """Return the dataset name."""
         return "visual_leak_bench"
 
-    async def fetch_dataset(self, *, cache: bool = True) -> SeedDataset:
+    async def fetch_dataset_async(self, *, cache: bool = True) -> SeedDataset:
         """
         Fetch VisualLeakBench examples and return as SeedDataset.
 

pyrit/datasets/seed_datasets/remote/vlguard_dataset.py

@@ -135,7 +135,7 @@ def dataset_name(self) -> str:
         """Return the dataset name."""
         return "vlguard"
 
-    async def fetch_dataset(self, *, cache: bool = True) -> SeedDataset:
+    async def fetch_dataset_async(self, *, cache: bool = True) -> SeedDataset:
         """
         Fetch VLGuard multimodal examples and return as SeedDataset.
 

pyrit/datasets/seed_datasets/remote/vlsu_multimodal_dataset.py

@@ -93,7 +93,7 @@ def dataset_name(self) -> str:
         """Return the dataset name."""
         return "ml_vlsu"
 
-    async def fetch_dataset(self, *, cache: bool = True) -> SeedDataset:
+    async def fetch_dataset_async(self, *, cache: bool = True) -> SeedDataset:
         """
         Fetch ML-VLSU multimodal examples and return as SeedDataset.
 

pyrit/datasets/seed_datasets/remote/xstest_dataset.py

@@ -42,7 +42,7 @@ def dataset_name(self) -> str:
         """Return the dataset name."""
         return "xstest"
 
-    async def fetch_dataset(self, *, cache: bool = True) -> SeedDataset:
+    async def fetch_dataset_async(self, *, cache: bool = True) -> SeedDataset:
         """
         Fetch XSTest dataset and return as SeedDataset.