Merge pull request #32 from shing100/main

feat: update score, add tqdm

generator_openai.py

@@ -3,6 +3,7 @@
 from openai import OpenAI
 import time
 import os
+from tqdm import tqdm
 
 MAX_MODEL_LEN = 1600
 # MODEL = "solar-1-mini-chat"
@@ -19,7 +20,7 @@ def format_single_turn_question(question):
 single_turn_questions = df_questions['questions'].map(format_single_turn_question)
 single_turn_outputs = []
 
-for question in single_turn_questions:
+for question in tqdm(single_turn_questions, desc="Processing Single Turn Questions"):
     response = client.chat.completions.create(
         model=MODEL,
         messages=[
@@ -39,7 +40,7 @@ def format_double_turn_question(question, single_turn_output):
 
 multi_turn_questions = df_questions[['questions', 'id']].apply(lambda x: format_double_turn_question(x['questions'], single_turn_outputs[x['id']-1]), axis=1)
 multi_turn_outputs = []
-for question in multi_turn_questions:
+for question in tqdm(multi_turn_questions, desc="Processing Multi Turn Questions"):
     response = client.chat.completions.create(
         model=MODEL,
         messages=[
@@ -56,4 +57,4 @@ def format_double_turn_question(question, single_turn_output):
     multi_turn_outputs.append(response.choices[0].message.content.strip())
 
 df_output = pd.DataFrame({'id': df_questions['id'], 'category': df_questions['category'], 'questions': df_questions['questions'], 'outputs': list(zip(single_turn_outputs, multi_turn_outputs)), "references": df_questions['references']})
-df_output.to_json(f'{str(MODEL).replace("/", "_")}.jsonl', orient='records', lines=True, force_ascii=False)
+df_output.to_json(f'{str(MODEL).replace("/", "_")}.jsonl', orient='records', lines=True, force_ascii=False)

score.py

@@ -37,16 +37,31 @@
             total_single_scores.append(single_score)
             total_multi_scores.append(multi_score)
 
-# 카테고리별 평균 점수 계산
+# 표의 헤더 생성
+table_header = "| Category | Single turn | Multi turn |\n|---|---|---|"
+
+# 표의 내용 생성
+table_rows = []
 for category, scores in category_scores.items():
     avg_single = sum(scores['single_scores']) / len(scores['single_scores'])
     avg_multi = sum(scores['multi_scores']) / len(scores['multi_scores'])
-    print(f"카테고리: {category}, 싱글 점수 평균: {avg_single:.2f}, 멀티 점수 평균: {avg_multi:.2f}")
+    table_rows.append(f"| {category} | {avg_single:.2f} | {avg_multi:.2f} |")
+
+    total_single_scores.extend(scores['single_scores'])
+    total_multi_scores.extend(scores['multi_scores'])
+
+# 카테고리별 점수 평균 출력
+print(table_header)
+for row in table_rows:
+    print(row)
 
 # 전체 점수의 평균 계산 및 출력
 avg_total_single = sum(total_single_scores) / len(total_single_scores)
 avg_total_multi = sum(total_multi_scores) / len(total_multi_scores)
 avg_total = (avg_total_single + avg_total_multi) / 2
-print(f"전체 싱글 점수 평균: {avg_total_single:.2f}")
-print(f"전체 멀티 점수 평균: {avg_total_multi:.2f}")
-print(f"전체 점수: {avg_total:.2f}")
+
+# 전체 점수 평균 출력
+print(f"\n| Category | Score |\n|---|---|")
+print(f"| Single turn | {avg_total_single:.2f} |")
+print(f"| Multi turn | {avg_total_multi:.2f} |")
+print(f"| Overall | {avg_total:.2f} |")