INF_MULT

app.py

@@ -13,6 +13,7 @@ QUANTIZE = {
     "Q5_K_M": 5.7,
     "Q6_K": 6.6,
 }
+INF_MULT = 1.25
 
 def get_model_api_info(model_name: str) -> dict:
     """
@@ -107,7 +108,7 @@ def estimate_gpu_memory(model_name: str) -> str:
     # GPUメモリサイズの計算
     total_params = sum(parameters.values())
     for quant, multiplier in QUANTIZE.items():
-        gpu_memory = total_params * multiplier / 8 / (1024 ** 3) * 1.3
+        gpu_memory = total_params * multiplier / 8 / (1024 ** 3) * INF_MULT
         result_lines.append(f"【{quant}】約 {gpu_memory:.2f} GB")
 
     result_lines.append("となります。")