diff --git a/app/main.py b/app/main.py
index c8d17d8..7c399cc 100644
--- a/app/main.py
+++ b/app/main.py
@@ -1,67 +1,258 @@
+#!/usr/bin/env python3
+# -*- coding: utf-8 -*-
+"""
+BGEM3 inference server (FastAPI) with robust GPU / CPU management.
+
+Launch examples:
+    python server.py                       # 自动选卡 / 自动降级
+    python server.py --device 1            # 固定用第 1 张 GPU
+    CUDA_VISIBLE_DEVICES=0,1 python server.py
+"""
+
+import argparse
+import logging
+import os
+import sys
+import time
+from typing import List, Union
+
+import torch
 from fastapi import FastAPI, HTTPException
 from pydantic import BaseModel
-from typing import List, Union
-import torch
 from transformers import AutoTokenizer
 from FlagEmbedding import BGEM3FlagModel
 
-# 自动检测设备
-DEVICE = "cuda" if torch.cuda.is_available() else "cpu"
-# 如果是 GPU，则启用 fp16 加速，否则使用 fp32
-USE_FP16 = DEVICE == "cuda"
+# -----------------------------------------------------------------------------#
+# Config
+# -----------------------------------------------------------------------------#
+MODEL_PATH = "model/bge-m3"          # 按需改成你的权重路径
+SAFE_MIN_FREE_MB = int(
+    os.getenv("SAFE_MIN_FREE_MB", "4096")
+)  # 启动时要求的最小空闲显存，MB
+
+# -----------------------------------------------------------------------------#
+# Logging
+# -----------------------------------------------------------------------------#
+logger = logging.getLogger("bge-m3-server")
+logging.basicConfig(
+    level=logging.INFO,
+    format="%(asctime)s [%(levelname)s] %(message)s",
+)
+
+# -----------------------------------------------------------------------------#
+# GPU memory helpers (NVML → torch fallback)
+# -----------------------------------------------------------------------------#
+try:
+    import pynvml
+
+    pynvml.nvmlInit()
+    _USE_NVML = True
+except Exception:
+    _USE_NVML = False
+    logger.warning("pynvml 不可用，将使用 torch.cuda.mem_get_info() 探测显存")
+
+
+def _gpu_mem_info(idx: int) -> tuple[int, int]:
+    """Return (free_bytes, total_bytes) for GPU `idx`."""
+    if _USE_NVML:
+        handle = pynvml.nvmlDeviceGetHandleByIndex(idx)
+        mem = pynvml.nvmlDeviceGetMemoryInfo(handle)
+        return mem.free, mem.total
+    # torch fallback
+    torch.cuda.set_device(idx)
+    free, total = torch.cuda.mem_get_info(idx)
+    return free, total
+
+
+# -----------------------------------------------------------------------------#
+# Precision & model loader
+# -----------------------------------------------------------------------------#
+def _choose_precision(device: str) -> str:
+    """Return 'fp16'|'bf16'|'fp32'."""
+    if not device.startswith("cuda"):
+        return "fp32"
+
+    major, _ = torch.cuda.get_device_capability(device)
+    if major >= 8:  # Ampere/Hopper
+        return "fp16"
+    if major >= 7:
+        return "bf16"
+    return "fp32"
+
+
+def load_model(device: str):
+    """Instantiate model on `device`; return (model, precision)."""
+    precision = _choose_precision(device)
+    use_fp16 = precision == "fp16"
+    logger.info("Loading BGEM3 on %s (%s)", device, precision)
+
+    mdl = BGEM3FlagModel(MODEL_PATH, use_fp16=use_fp16, device=device)
+
+    # Simple DataParallel for multi-GPU inference
+    if device.startswith("cuda") and torch.cuda.device_count() > 1:
+        logger.info(
+            "Wrapping model with torch.nn.DataParallel (%d GPUs)",
+            torch.cuda.device_count(),
+        )
+        mdl = torch.nn.DataParallel(mdl)
+    return mdl, precision
+
+
+# -----------------------------------------------------------------------------#
+# Auto-select device (startup)
+# -----------------------------------------------------------------------------#
+def auto_select_and_load(min_free_mb: int = 4096):
+    """
+    1. Gather GPUs with free_mem ≥ min_free_mb
+    2. Sort by free_mem desc, attempt to load
+    3. All fail → CPU
+    Return (model, device, precision)
+    """
+    if not torch.cuda.is_available():
+        logger.info("No GPU detected → CPU")
+        return (*load_model("cpu"), "cpu")
+
+    # Build candidate list
+    candidates: list[tuple[int, int]] = []  # (free_MB, idx)
+    for idx in range(torch.cuda.device_count()):
+        free, _ = _gpu_mem_info(idx)
+        free_mb = free // 2**20
+        candidates.append((free_mb, idx))
+
+    candidates = [c for c in candidates if c[0] >= min_free_mb]
+    if not candidates:
+        logger.warning("All GPUs free_mem < %d MB → CPU", min_free_mb)
+        return (*load_model("cpu"), "cpu")
+
+    candidates.sort(reverse=True)  # high free_mem first
+    for free_mb, idx in candidates:
+        dev = f"cuda:{idx}"
+        try:
+            logger.info("Trying %s (free=%d MB)", dev, free_mb)
+            mdl, prec = load_model(dev)
+            return mdl, prec, dev
+        except RuntimeError as e:
+            if "out of memory" in str(e).lower():
+                logger.warning("%s OOM → next GPU", dev)
+                torch.cuda.empty_cache()
+                continue
+            raise  # non-OOM error
+
+    logger.warning("All GPUs failed → CPU fallback")
+    return (*load_model("cpu"), "cpu")
+
+
+# -----------------------------------------------------------------------------#
+# CLI
+# -----------------------------------------------------------------------------#
+parser = argparse.ArgumentParser()
+parser.add_argument(
+    "--device", help="GPU index (e.g. 0) or 'cpu'; overrides auto-selection"
+)
+args, _ = parser.parse_known_args()
+
+if args.device is not None:
+    # Forced path
+    if args.device.lower() == "cpu":
+        DEVICE = "cpu"
+    else:
+        DEVICE = f"cuda:{int(args.device)}" if torch.cuda.is_available() else "cpu"
+    model, PRECISION = load_model(DEVICE)
+else:
+    # Auto path with VRAM check
+    model, PRECISION, DEVICE = auto_select_and_load(SAFE_MIN_FREE_MB)
 
-# 加载模型和分词器
-MODEL_PATH = "model/bge-m3"
-model = BGEM3FlagModel(MODEL_PATH, use_fp16=USE_FP16, device=DEVICE)
 tokenizer = AutoTokenizer.from_pretrained(MODEL_PATH)
 
+# -----------------------------------------------------------------------------#
+# FastAPI
+# -----------------------------------------------------------------------------#
 app = FastAPI()
 
+
 class EmbeddingRequest(BaseModel):
     input: Union[str, List[str]]
     model: str = "text-embedding-bge-m3"
 
-@app.post("/v1/embeddings")
-def create_embedding(request: EmbeddingRequest):
-    # 统一成列表
-    texts = [request.input] if isinstance(request.input, str) else request.input
+
+fallback_done = False  # prevent endless downgrade loop
+
+
+def _encode(texts: List[str]):
+    """Encode with single downgrade to CPU on OOM / CUDA failure."""
+    global model, DEVICE, PRECISION, fallback_done
 
     try:
-        # 先用 tokenizer 统计 token 数
-        encoding = tokenizer(
-            texts,
-            padding=True,
-            truncation=True,
-            max_length=8192,
-            return_tensors="pt"
-        )
-        # attention_mask 中 1 的数量即为实际 tokens（不含 padding）
-        mask = encoding["attention_mask"]
-        # prompt_tokens = 所有输入 tokens 之和
-        prompt_tokens = int(mask.sum().item())
-        total_tokens = prompt_tokens  # embedding 不产生额外 tokens
+        return model.encode(texts, return_dense=True)
 
-        # 生成 dense 向量
-        output = model.encode(texts, return_dense=True)
+    except RuntimeError as err:
+        is_oom = "out of memory" in str(err).lower()
+        is_cuda_fail = "cuda error" in str(err).lower() or "device-side assert" in str(
+            err
+        ).lower()
+
+        if (is_oom or is_cuda_fail) and not fallback_done:
+            logger.error("GPU failure (%s). Falling back to CPU…", err)
+            fallback_done = True
+            torch.cuda.empty_cache()
+            DEVICE = "cpu"
+            model, PRECISION = load_model(DEVICE)
+            return model.encode(texts, return_dense=True)
+
+        raise  # second failure → propagate
+
+
+@app.post("/v1/embeddings")
+def create_embedding(request: EmbeddingRequest):
+    texts = [request.input] if isinstance(request.input, str) else request.input
+
+    # Token stats
+    enc = tokenizer(
+        texts,
+        padding=True,
+        truncation=True,
+        max_length=8192,
+        return_tensors="pt",
+    )
+    prompt_tokens = int(enc["attention_mask"].sum().item())
+
+    try:
+        output = _encode(texts)
         embeddings = output["dense_vecs"]
-
-        return {
-            "object": "list",
-            "data": [
-                {
-                    "object": "embedding",
-                    "index": idx,
-                    "embedding": emb.tolist() if hasattr(emb, "tolist") else emb
-                }
-                for idx, emb in enumerate(embeddings)
-            ],
-            "model": request.model,
-            "usage": {
-                "prompt_tokens": prompt_tokens,
-                "total_tokens": total_tokens
-            }
-        }
-
     except Exception as e:
+        logger.exception("Embedding failed")
         raise HTTPException(status_code=500, detail=str(e))
 
+    return {
+        "object": "list",
+        "data": [
+            {
+                "object": "embedding",
+                "index": i,
+                "embedding": emb.tolist() if hasattr(emb, "tolist") else emb,
+            }
+            for i, emb in enumerate(embeddings)
+        ],
+        "model": request.model,
+        "usage": {
+            "prompt_tokens": prompt_tokens,
+            "total_tokens": prompt_tokens,
+        },
+        "device": DEVICE,
+        "precision": PRECISION,
+    }
+
+
+# -----------------------------------------------------------------------------#
+# Entry-point for `python server.py`
+# -----------------------------------------------------------------------------#
+if __name__ == "__main__":
+    import uvicorn
+
+    uvicorn.run(
+        "server:app",
+        host="0.0.0.0",
+        port=int(os.getenv("PORT", 8000)),
+        log_level="info",
+        workers=1,  # multi-process → use gunicorn/uvicorn externally
+    )
diff --git a/app/main.py.old b/app/main.py.old
new file mode 100644
index 0000000..c8d17d8
--- /dev/null
+++ b/app/main.py.old
@@ -0,0 +1,67 @@
+from fastapi import FastAPI, HTTPException
+from pydantic import BaseModel
+from typing import List, Union
+import torch
+from transformers import AutoTokenizer
+from FlagEmbedding import BGEM3FlagModel
+
+# 自动检测设备
+DEVICE = "cuda" if torch.cuda.is_available() else "cpu"
+# 如果是 GPU，则启用 fp16 加速，否则使用 fp32
+USE_FP16 = DEVICE == "cuda"
+
+# 加载模型和分词器
+MODEL_PATH = "model/bge-m3"
+model = BGEM3FlagModel(MODEL_PATH, use_fp16=USE_FP16, device=DEVICE)
+tokenizer = AutoTokenizer.from_pretrained(MODEL_PATH)
+
+app = FastAPI()
+
+class EmbeddingRequest(BaseModel):
+    input: Union[str, List[str]]
+    model: str = "text-embedding-bge-m3"
+
+@app.post("/v1/embeddings")
+def create_embedding(request: EmbeddingRequest):
+    # 统一成列表
+    texts = [request.input] if isinstance(request.input, str) else request.input
+
+    try:
+        # 先用 tokenizer 统计 token 数
+        encoding = tokenizer(
+            texts,
+            padding=True,
+            truncation=True,
+            max_length=8192,
+            return_tensors="pt"
+        )
+        # attention_mask 中 1 的数量即为实际 tokens（不含 padding）
+        mask = encoding["attention_mask"]
+        # prompt_tokens = 所有输入 tokens 之和
+        prompt_tokens = int(mask.sum().item())
+        total_tokens = prompt_tokens  # embedding 不产生额外 tokens
+
+        # 生成 dense 向量
+        output = model.encode(texts, return_dense=True)
+        embeddings = output["dense_vecs"]
+
+        return {
+            "object": "list",
+            "data": [
+                {
+                    "object": "embedding",
+                    "index": idx,
+                    "embedding": emb.tolist() if hasattr(emb, "tolist") else emb
+                }
+                for idx, emb in enumerate(embeddings)
+            ],
+            "model": request.model,
+            "usage": {
+                "prompt_tokens": prompt_tokens,
+                "total_tokens": total_tokens
+            }
+        }
+
+    except Exception as e:
+        raise HTTPException(status_code=500, detail=str(e))
+
diff --git a/requirements.txt b/requirements.txt
index ad250fc..73d86f1 100644
--- a/requirements.txt
+++ b/requirements.txt
@@ -4,3 +4,4 @@ torch
 transformers
 datasets
 peft
+pynvml
\ No newline at end of file