update gemma-2 default BOS behavior (#2049)

lm_eval/models/huggingface.py

@@ -256,9 +256,8 @@ class HFLM(TemplateLM):
         # select (or create) a pad token to use
         self.tokenizer = configure_pad_token(self.tokenizer, model_config=self.config)
 
-        # TODO: override this for Gemma
         self.add_bos_token = add_bos_token
-        if getattr(self.config, "model_type", None) == "gemma":
+        if getattr(self.config, "model_type", None) in ["gemma", "gemma2"]:
             self.add_bos_token = True
             eval_logger.info(
                 f"Model type is '{self.config.model_type}', a BOS token will be used as Gemma underperforms without it."

lm_eval/models/vllm_causallms.py

@@ -125,7 +125,7 @@ class VLLM(TemplateLM):
         if "gemma" in pretrained.lower():
             self.add_bos_token = True
             eval_logger.info(
-                "Found 'gemma' in model name, a BOS token will be used as Gemma underperforms without it."
+                "Found 'gemma' in model name, a BOS token will be used as Gemma series models underperform without it."
             )
 
         self.custom_prefix_token_id = prefix_token_id