Spaces:
Sleeping
Sleeping
Commit
·
d72603f
1
Parent(s):
c15db97
update cache
Browse files- app.py +2 -2
- code/rag_core.py +3 -2
app.py
CHANGED
|
@@ -6,11 +6,11 @@ from code.rag_core import RAGConfig, SongIndex, Responder, classify_title
|
|
| 6 |
|
| 7 |
CSV_PATH = os.environ.get("CSV_PATH", "data.songs.csv")
|
| 8 |
GENRE_NAME = os.environ.get("GENRE_NAME", "Rock & Roll")
|
| 9 |
-
|
| 10 |
app = FastAPI(title="RAG Rock & Roll Checker API", version="1.0.0")
|
| 11 |
|
| 12 |
# Cargar índice al arrancar
|
| 13 |
-
cfg = RAGConfig(songs_csv=CSV_PATH, genre_name=GENRE_NAME)
|
| 14 |
song_index = SongIndex(cfg)
|
| 15 |
responder = Responder(cfg.genre_name)
|
| 16 |
|
|
|
|
| 6 |
|
| 7 |
CSV_PATH = os.environ.get("CSV_PATH", "data.songs.csv")
|
| 8 |
GENRE_NAME = os.environ.get("GENRE_NAME", "Rock & Roll")
|
| 9 |
+
TRANSFORMERS_CACHE = os.environ.get("TRANSFORMERS_CACHE", "")
|
| 10 |
app = FastAPI(title="RAG Rock & Roll Checker API", version="1.0.0")
|
| 11 |
|
| 12 |
# Cargar índice al arrancar
|
| 13 |
+
cfg = RAGConfig(songs_csv=CSV_PATH, genre_name=GENRE_NAME,cache = TRANSFORMERS_CACHE)
|
| 14 |
song_index = SongIndex(cfg)
|
| 15 |
responder = Responder(cfg.genre_name)
|
| 16 |
|
code/rag_core.py
CHANGED
|
@@ -29,12 +29,13 @@ def normalize_title(t: str) -> str:
|
|
| 29 |
class RAGConfig:
|
| 30 |
songs_csv: str
|
| 31 |
genre_name: str = "Rock & Roll"
|
| 32 |
-
|
|
|
|
| 33 |
class SongIndex:
|
| 34 |
def __init__(self, cfg: RAGConfig):
|
| 35 |
self.cfg = cfg
|
| 36 |
self.df = self._load_dataset(cfg.songs_csv)
|
| 37 |
-
self.model = SentenceTransformer(EMBEDDING_MODEL_NAME)
|
| 38 |
self.index, self.embeddings = self._build_faiss(self.df["title"].tolist())
|
| 39 |
self.norm_to_idx = {normalize_title(t): i for i, t in enumerate(self.df["title"].tolist())}
|
| 40 |
|
|
|
|
| 29 |
class RAGConfig:
|
| 30 |
songs_csv: str
|
| 31 |
genre_name: str = "Rock & Roll"
|
| 32 |
+
cache : str
|
| 33 |
+
|
| 34 |
class SongIndex:
|
| 35 |
def __init__(self, cfg: RAGConfig):
|
| 36 |
self.cfg = cfg
|
| 37 |
self.df = self._load_dataset(cfg.songs_csv)
|
| 38 |
+
self.model = SentenceTransformer(EMBEDDING_MODEL_NAME,cache_folder=self.cfg.cache)
|
| 39 |
self.index, self.embeddings = self._build_faiss(self.df["title"].tolist())
|
| 40 |
self.norm_to_idx = {normalize_title(t): i for i, t in enumerate(self.df["title"].tolist())}
|
| 41 |
|