refactor: 모델 이름 일괄 변경

- gpt-3.5-turbo-16k -> gpt-4o-mini #41
boostcampaitech6 · Aug 8, 2024 · 007d67c · 007d67c
1 parent e854b6c
commit 007d67c
Show file tree

Hide file tree

Showing 6 changed files with 7 additions and 18 deletions.
diff --git a/pages/3_gene_question.py b/pages/3_gene_question.py
@@ -26,7 +26,7 @@
 from src.rule_based import list_extend_questions_based_on_keywords
 from src.util import local_css, read_prompt_from_txt
 from src.semantic_search import faiss_inference, reranker
-from config import OPENAI_API_KEY, DATA_DIR, IMG_PATH, CSS_PATH, PORT
+from config import OPENAI_API_KEY, DATA_DIR, IMG_PATH, CSS_PATH, PORT, MODEL_NAME
 
 st.session_state["FAV_IMAGE_PATH"] = os.path.join(IMG_PATH, "favicon.png")
 st.set_page_config(
@@ -80,8 +80,6 @@
             </style>
             """,unsafe_allow_html=True)
 
-## set variables
-MODEL_NAME = "gpt-3.5-turbo-16k"
 
 ## set save dir
 USER_RESUME_SAVE_DIR = os.path.join(st.session_state["save_dir"], "2_generate_question_user_resume.pdf")

diff --git a/pages/3_gene_question_no_resume.py b/pages/3_gene_question_no_resume.py
@@ -27,7 +27,7 @@
 from src.rule_based import list_extend_questions_based_on_keywords
 from src.util import local_css, read_prompt_from_txt
 from src.semantic_search import faiss_inference, reranker
-from config import OPENAI_API_KEY, DATA_DIR, IMG_PATH, CSS_PATH, PORT
+from config import OPENAI_API_KEY, DATA_DIR, IMG_PATH, CSS_PATH, MODEL_NAME
 
 st.session_state["FAV_IMAGE_PATH"] = os.path.join(IMG_PATH, "favicon.png")
 st.set_page_config(
@@ -81,8 +81,6 @@
             </style>
             """,unsafe_allow_html=True)
 
-## set variables
-MODEL_NAME = "gpt-3.5-turbo-16k"
 
 ## set save dir
 USER_RESUME_SAVE_DIR = os.path.join(st.session_state["save_dir"], "2_generate_question_user_resume.pdf")

diff --git a/pages/4_show_questions_hint.py b/pages/4_show_questions_hint.py
@@ -12,7 +12,7 @@
 from src.generate_question import (create_prompt_feedback,  # 추가
                                    create_prompt_hint)
 from src.util import read_prompt_from_txt
-from config import DATA_DIR, IMG_PATH, OPENAI_API_KEY
+from config import DATA_DIR, IMG_PATH, OPENAI_API_KEY, MODEL_NAME
 
 st.session_state["FAV_IMAGE_PATH"] = os.path.join(IMG_PATH, "favicon.png")
 st.set_page_config(
@@ -23,8 +23,6 @@
     layout="wide",
     initial_sidebar_state="collapsed",
 )
-#MODEL_NAME = "gpt-4-0125-preview"
-MODEL_NAME = "gpt-3.5-turbo-16k"
 NEXT_PAGE = "introduction"
 
 st.session_state.logger.info("start show_questions page")

diff --git a/src/gene_question_2chain_ver.py b/src/gene_question_2chain_ver.py
@@ -20,7 +20,7 @@
 from streamlit_extras.switch_page_button import switch_page
 from util import local_css, read_prompt_from_txt
 
-from config import OPENAI_API_KEY  # OPENAI_API_KEY 불러오기
+from config import OPENAI_API_KEY, MODEL_NAME
 
 DATA_DIR = os.path.join(os.path.dirname(os.path.dirname(__file__)), "data")
 st.session_state["FAV_IMAGE_PATH"] = os.path.join(DATA_DIR, "images/favicon.png")
@@ -109,9 +109,6 @@
 #             """,
             unsafe_allow_html=True,)
 
-## set variables
-MODEL_NAME = "gpt-3.5-turbo-16k"
-
 ## set save dir
 USER_RESUME_SAVE_DIR = os.path.join(st.session_state["save_dir"], "2_generate_question_user_resume.pdf")
 ### 추가
@@ -191,9 +188,7 @@
             st.session_state.logger.info("create prompt JD object")
 
             ### 모델 세팅 그대로
-            llm = ChatOpenAI(
-                temperature=st.session_state.temperature, model_name=MODEL_NAME, openai_api_key=OPENAI_API_KEY
-            )
+            llm = ChatOpenAI(temperature=st.session_state.temperature, model_name=MODEL_NAME)
 
             st.session_state.logger.info("create llm object")
 

diff --git a/src/make_faiss_index.py b/src/make_faiss_index.py
@@ -12,7 +12,7 @@
 data = pd.read_csv('hellojobits_tag.csv')
 data = data.iloc[:,0].tolist()
 
-model_name = "sentence-transformers/distiluse-base-multilingual-cased-v2"
+model_name = "jhgan/ko-sroberta-multitask"
 model_kwargs = {'device': 'cuda'}
 encode_kwargs = {'normalize_embeddings': True} # set True to compute cosine similarity
 model_norm = HuggingFaceEmbeddings(

diff --git a/src/semantic_search.py b/src/semantic_search.py
@@ -15,7 +15,7 @@ def faiss_inference(query):
     output으로는 (4 * 키워드 줄 개수 - 중복된 문장)개의 질문이 반환됩니다
     '''
     embeddings = HuggingFaceEmbeddings(
-            model_name = "sentence-transformers/distiluse-base-multilingual-cased-v2",
+            model_name = "BM-K/KoSimCSE-roberta-multitask",
             model_kwargs = {'device': 'cuda'}
         )
     store_name="./FAISS_INDEX_TAG"