chat tempalte evaluations

src/leaderboard/read_evals.py

@@ -46,6 +46,7 @@ class EvalResult:
         config = data.get("config")
         n_shot = data.get("n-shot")
         start_date = data.get("date", 0)
+        chat_template = data.get("chat_template", None)
 
         # Precision
         precision = Precision.from_str(config.get("model_dtype"))
@@ -87,6 +88,9 @@ class EvalResult:
             result_key = f"{org}_{model}" # _{precision.value.name}
         full_model = "/".join(org_and_model)
 
+        if chat_template:
+            result_key = f"{result_key}_chat"
+
         still_on_hub, err, model_config = is_model_on_hub(
             full_model.split(',')[0], config.get("model_sha", "main"), trust_remote_code=True, test_tokenizer=False
         )