Update app.py

app.py

@@ -3,10 +3,14 @@ from pydantic import BaseModel
 from transformers import AutoTokenizer, AutoModel
 import torch
 import torch.nn.functional as F
+import os
+
+# 💡 Correction ici
+os.environ['HF_HOME'] = '/data'
 
 app = FastAPI()
 
-# Charger le modèle depuis HF sans passer par SentenceTransformer
+# Charger le modèle
 MODEL_NAME = "thenlper/gte-small"
 tokenizer = AutoTokenizer.from_pretrained(MODEL_NAME)
 model = AutoModel.from_pretrained(MODEL_NAME)
@@ -21,4 +25,4 @@ async def embed_text(payload: EmbedInput):
         outputs = model(**inputs)
         embeddings = outputs.last_hidden_state[:, 0]  # CLS token
         normalized = F.normalize(embeddings, p=2, dim=1)
-    return {"embedding": normalized[0].tolist()}
+    return {'embedding': normalized[0].tolist()}