add the hack (works for Mistral/Llama, destroys performance for GPT2

lm_eval/models/huggingface.py

@@ -792,6 +792,26 @@ class HFLM(LM):
         # context_enc = self.tok_encode(context, add_special_tokens=False)
         context_enc_len = len(context_enc)
         continuation_enc = whole_enc[context_enc_len:]
+
+        # quite the hack, but what this does:
+        # circumvents the addition of an extraneous sentencepiece underline token
+        # that was produced when passing " <word>" into the Llama / Mistral tokenizer.
+        # if instead we pass "<word>" in, we don't get this extra token (29871 for Llama.)
+        # which would hurt performance if provided.
+        if (
+            len(continuation.lstrip()) + 1 == len(continuation)
+            and continuation.startswith(" ")
+        ) or (len(continuation_enc) == 0):
+            context_enc_2 = context_enc
+            continuation_enc_2 = self.tok_encode(
+                continuation[1:], add_special_tokens=False
+            )
+
+            # assert context_enc == context_enc_2
+            # assert continuation_enc == continuation_enc_2, f"{continuation_enc},{continuation_enc_2}"
+
+            return context_enc_2, continuation_enc_2
+
         return context_enc, continuation_enc
 
     def loglikelihood(self, requests: List[Instance]) -> List[Tuple[float, bool]]: