fix params

lm_eval/api/model.py

@@ -336,7 +336,7 @@ class TemplateLM(LM):
         return self.eot_token_id
 
     @abc.abstractmethod
-    def tok_encode(self, string: str, **kwargs) -> list[int]:
+    def tok_encode(self, string: str, add_special_tokens=False, **kwargs) -> list[int]:
         """
         Tokenize a string using the model's tokenizer and return a list of token IDs.
         """
@@ -377,6 +377,7 @@ class TemplateLM(LM):
             This method does NOT handle empty context. The caller should
             handle empty context (see loglikelihood method).
         """
+        assert context, "Context cannot be empty!"
         import transformers
 
         n_spaces = len(context) - len(context.rstrip())
@@ -429,7 +430,9 @@ class TemplateLM(LM):
         new_reqs = []
         for context, continuation in [req.args for req in requests]:
             if context == "":
-                continuation_enc = self.tok_encode(continuation)
+                continuation_enc = self.tok_encode(
+                    continuation, add_special_tokens=False
+                )
                 # BOS or EOS as context
                 context_enc, continuation_enc = (
                     ([self.prefix_token_id], continuation_enc)

lm_eval/models/utils.py

@@ -150,7 +150,7 @@ class Grouper:
 
 def pad_and_concat(
     max_length: int,
-    tensors: List[torch.Tensor],
+    tensors: list[torch.Tensor],
     padding_side: Literal["right", "left"] = "right",
 ):
     """