fix

app.py

@@ -12,7 +12,7 @@ logger = logging.getLogger(__name__)
 app = FastAPI()
 
 # モデルロード
-model_name = "google/gemma-3-4b-it"
+model_name = "google/gemma-3-4b-it"  # 軽量な2Bモデルに変更
 try:
     logger.info(f"Loading model: {model_name}")
     processor = AutoProcessor.from_pretrained(model_name, token=os.getenv("HF_TOKEN"))
@@ -21,7 +21,8 @@ try:
         torch_dtype=torch.bfloat16,
         device_map="auto",
         token=os.getenv("HF_TOKEN"),
-        low_cpu_mem_usage=True
+        low_cpu_mem_usage=True,
+        load_in_4bit=True  # 量子化でメモリ節約
     )
     logger.info("Model loaded successfully")
 except Exception as e:

requirements.txt

@@ -1,3 +1,7 @@
 huggingface_hub==0.25.2
-torch
-transformers
+torch==2.1.0+cpu
+transformers==4.44.2
+bitsandbytes==0.42.0
+accelerate==0.26.1
+fastapi==0.115.0
+uvicorn==0.30.6