Update app.py

app.py

@@ -9,7 +9,9 @@ from fastapi import FastAPI, HTTPException
 from pydantic import BaseModel
 from sentence_transformers import SentenceTransformer
 from transformers import AutoModelForSeq2SeqLM, AutoTokenizer
-
+os.environ["HF_HOME"] = "/tmp/huggingface"
+os.environ["TRANSFORMERS_CACHE"] = "/tmp/huggingface"
+os.environ["SENTENCE_TRANSFORMERS_HOME"] = "/tmp/huggingface"
 # ✅ Initialize FastAPI
 app = FastAPI()
 
@@ -20,10 +22,9 @@ if not OPENROUTER_API_KEY:
 
 OPENROUTER_BASE_URL = "https://openrouter.ai/api/v1/chat/completions"
 
-# ✅ Load AI Models
-embedding_model = SentenceTransformer("all-MiniLM-L6-v2")
-summarization_model = AutoModelForSeq2SeqLM.from_pretrained("google/long-t5-tglobal-base")
-summarization_tokenizer = AutoTokenizer.from_pretrained("google/long-t5-tglobal-base")
+embedding_model = SentenceTransformer("sentence-transformers/paraphrase-MiniLM-L6-v2", cache_folder="/app/cache")
+summarization_model = AutoModelForSeq2SeqLM.from_pretrained("google/long-t5-tglobal-base", cache_dir="/app/cache")
+summarization_tokenizer = AutoTokenizer.from_pretrained("google/long-t5-tglobal-base", cache_dir="/app/cache")
 
 # ✅ Load Datasets
 try: