Upload src/mine_memorized.py with huggingface_hub

src/mine_memorized.py

@@ -49,7 +49,7 @@ def sample_sequences(
     min_text_tokens: int = None,
     seed: int = 42,
     streaming: bool = True,
-    dataset_config: Optional[str] = None,
+    dataset_config: Optional[str] = "en",  # Default to English for C4
     text_column: str = "text",
 ) -> list[MemorizedSequence]:
     """
@@ -253,14 +253,14 @@ def filter_memorized(
 def mine_memorized_sequences(
     model: nn.Module,
     tokenizer,
-    dataset_name: str = "allenai/olmo-mix-1124",
+    dataset_name: str = "allenai/c4",
     target_count: int = 1000,
     max_candidates: int = 50000,
     prefix_len: int = 64,
     suffix_len: int = 48,
     batch_size: int = 8,
     seed: int = 42,
-    dataset_config: Optional[str] = None,
+    dataset_config: Optional[str] = "en",  # Default to English for C4
     strict: bool = True,
 ) -> list[MemorizedSequence]:
     """