fix: Multiple-E dataset fix go_test.go path for test execution

bigcode-project · Apr 20, 2024 · 5930726 · 5930726
1 parent 1b0147c
commit 5930726
Show file tree

Hide file tree

Showing 2 changed files with 3 additions and 2 deletions.
diff --git a/bigcode_eval/tasks/custom_metrics/multiple_metrics/containerized_eval.py b/bigcode_eval/tasks/custom_metrics/multiple_metrics/containerized_eval.py
@@ -36,6 +36,7 @@
     "jl": (eval_julia.eval_script, ".jl"),
     "ts": (eval_ts.eval_script, ".ts"),
     "go": (eval_go.eval_script, ".go"),
+    "go_test.go": (eval_go.eval_script, "_test.go"),
     "pl": (eval_pl.eval_script, ".pl"),
     "sh": (eval_sh.eval_script, ".sh"),
     "scala": (eval_scala.eval_script, ".scala"),

diff --git a/bigcode_eval/tasks/multiple.py b/bigcode_eval/tasks/multiple.py
@@ -138,7 +138,7 @@ def process_results(self, generations, references):
         """
         # get prompts and problem names
         prompts_names = [
-            {"prompt": doc["prompt"], "name": doc["name"]}
+            {"prompt": doc["prompt"], "name": doc["name"], "lang": doc["language"]}
             for i, doc in enumerate(self.get_dataset())
             if i < len(generations)
         ]
@@ -150,7 +150,7 @@ def process_results(self, generations, references):
         ):
             problem = {
                 "name": prompt_name["name"],
-                "language": self.language,
+                "language": prompt_name["lang"],
                 "prompt": prompt_name["prompt"],
                 "completions": generation,
                 "tests": reference,