diff --git a/tests/_protocol.py b/tests/_protocol.py
index dfc140d..a08db24 100644
--- a/tests/_protocol.py
+++ b/tests/_protocol.py
@@ -16,23 +16,27 @@ class classify_module(Protocol):
     Protocol for all `classify` modules.
     """
 
-    def token_logprobs(self, texts: str | Sequence[str], Any) -> list[list[float]]:
+    def token_logprobs(
+        self, texts: str | Sequence[str], *args, **kwargs
+    ) -> list[list[float]]:
         """
         For each text, log Pr(token[i] | tokens[:i]) for each token
         """
 
-    def cache(self, model: Model, prefix: str) -> _GeneratorContextManager[Model]:
+    def cache(
+        self, model: Model, prefix: str, *args, **kwargs
+    ) -> _GeneratorContextManager[Model]:
         """
         Optional: context manager which caches the `model` with `prefix`
         """
 
-    def cache_model(self, model: Model, prefix: str) -> Model:
+    def cache_model(self, model: Model, prefix: str, *args, **kwargs) -> Model:
         """
         Optional: returns a model cached with `prefix`
         """
 
     def log_probs_conditional(
-        self, prompts: str | Sequence[str], completions: Sequence[str], Any
+        self, prompts: str | Sequence[str], completions: Sequence[str], *args, **kwargs
     ) -> list[list[float]] | list[list[list[float]]]:
         """
         list[i][j][k] = log Pr(
@@ -41,7 +45,7 @@ def log_probs_conditional(
         """
 
     def log_probs_conditional_examples(
-        self, examples: Example | Sequence[Example], Any
+        self, examples: Example | Sequence[Example], *args, **kwargs
     ) -> list[list[float]] | list[list[list[float]]]:
         """
         list[i][j][k] = log Pr(
@@ -51,28 +55,28 @@ def log_probs_conditional_examples(
         """
 
     def predict_proba(
-        self, prompts: str | Sequence[str], completions: Sequence[str], Any
+        self, prompts: str | Sequence[str], completions: Sequence[str], *args, **kwargs
     ) -> npt.NDArray[np.floating]:
         """
         array[i, j] = Pr(completions[j] | prompts[i])
         """
 
     def predict_proba_examples(
-        self, examples: Example | Sequence[Example], Any
+        self, examples: Example | Sequence[Example], *args, **kwargs
     ) -> npt.NDArray[np.floating] | list[npt.NDArray[np.floating]]:
         """
         list[i][j] = Pr(examples[i].completions[j] | examples[i].prompt)
         """
 
     def predict(
-        self, prompts: str | Sequence[str], completions: Sequence[str], Any
+        self, prompts: str | Sequence[str], completions: Sequence[str], *args, **kwargs
     ) -> str | list[str]:
         """
         list[i] = argmax_j Pr(completions[j] | prompts[i])
         """
 
     def predict_examples(
-        self, examples: Example | Sequence[Example], Any
+        self, examples: Example | Sequence[Example], *args, **kwargs
     ) -> str | list[str]:
         """
         list[i] = argmax_j Pr(examples[i].completions[j] | examples[i].prompt)