change filename, parse results

Former-commit-id: 6686e2a3ddcb1ba2871be6a1786c262af1a852ba

evaluation/scores/{multi_score.py → multi_scores.py}

@@ -2,15 +2,15 @@ from comet import download_model, load_from_checkpoint
 from sacrebleu.metrics import BLEU, CHRF, TER
 import LLM_eval
 
-class multi_score:
+class multi_scores:
     def __init__(self) -> None:
         self.comet_model = load_from_checkpoint(download_model("Unbabel/wmt22-comet-da"))
         self.bleu_model = BLEU()
         self.LLM_model = LLM_eval.init_evaluator()
 
-    def get(self, src, mt, ref):
-        comet_score = self.comet_model.predict([{"src":src, "mt":mt, "ref":ref}], batch_size=8, gpus=0).scores
+    def get(self, src:str, mt:str, ref:str) -> dict:
+        comet_score = self.comet_model.predict([{"src":src, "mt":mt, "ref":ref}], batch_size=8, gpus=0).scores[0]
         bleu_score = self.bleu_model.corpus_score(mt, ref).score
-        LLM_score = LLM_eval.evaluate_prediction(src, ref, mt, self.LLM_model)
+        LLM_score = LLM_eval.evaluate_prediction(src, ref, mt, self.LLM_model).score
         return {'bleu':bleu_score, 'comet':comet_score, 'llm':LLM_score}
         

evaluation/scores/{scores.py → score.py}

@@ -2,7 +2,9 @@ from comet import download_model, load_from_checkpoint
 from sacrebleu.metrics import BLEU, CHRF, TER
 
 def COMETscore(src, mt, ref):
-    data = [{"src":src, "mt":mt, "ref":ref}]
+    data = []
+    for i in enumerate(src):
+        data.append({"src":src[i], "mt":mt[i], "ref":ref[i]})
     model_path = download_model("Unbabel/wmt22-comet-da")
     model = load_from_checkpoint(model_path)
     model_output = model.predict(data, batch_size = 8, gpus=0)