Add switchable DIFF V1 attention

attention_types.py

@@ -0,0 +1,26 @@
+"""Attention implementation identifiers and validation helpers."""
+
+
+STANDARD_ATTENTION = "standard"
+DIFF_ATTENTION = "diff"
+DEFAULT_ATTENTION_TYPE = STANDARD_ATTENTION
+SUPPORTED_ATTENTION_TYPES = (
+    STANDARD_ATTENTION,
+    DIFF_ATTENTION,
+)
+
+
+def resolve_attention_type(attention_type: str | None = None) -> str:
+    """Resolve an attention selector, defaulting old configs to standard."""
+    resolved = DEFAULT_ATTENTION_TYPE if attention_type is None else attention_type
+    if not isinstance(resolved, str):
+        raise ValueError(
+            "attention_type must be one of "
+            f"{SUPPORTED_ATTENTION_TYPES}, got {resolved!r}"
+        )
+    if resolved not in SUPPORTED_ATTENTION_TYPES:
+        raise ValueError(
+            f"Unsupported attention_type={resolved!r}; "
+            f"expected one of {SUPPORTED_ATTENTION_TYPES}."
+        )
+    return resolved