diff --git a/eval_llm.py b/eval_llm.py
index 9359ca8..e9f8a68 100755
--- a/eval_llm.py
+++ b/eval_llm.py
@@ -3,6 +3,10 @@ import argparse
 import random
 import warnings
 import torch
+try:
+    import torch_npu
+except:
+    torch_npu = None
 from transformers import AutoTokenizer, AutoModelForCausalLM, TextStreamer
 from model.model_minimind import MiniMindConfig, MiniMindForCausalLM
 from model.model_lora import *
@@ -45,7 +49,10 @@ def main():
     parser.add_argument('--open_thinking', default=0, type=int, help="是否开启自适应思考（0=否，1=是）")
     parser.add_argument('--historys', default=0, type=int, help="携带历史对话轮数（需为偶数，0表示不携带历史）")
     parser.add_argument('--show_speed', default=1, type=int, help="显示decode速度（tokens/s）")
-    parser.add_argument('--device', default='cuda' if torch.cuda.is_available() else 'cpu', type=str, help="运行设备")
+    parser.add_argument('--device',
+                        default='cuda' if torch.cuda.is_available() else ('npu' if torch_npu and torch_npu.npu.is_available() else 'cpu'),
+                        type=str,
+                        help="运行设备")
     args = parser.parse_args()
     
     prompts = [
diff --git a/scripts/eval_toolcall.py b/scripts/eval_toolcall.py
index da078b4..c94de4d 100644
--- a/scripts/eval_toolcall.py
+++ b/scripts/eval_toolcall.py
@@ -8,6 +8,10 @@ import random
 import argparse
 import warnings
 import torch
+try:
+    import torch_npu
+except:
+    torch_npu = None
 from datetime import datetime
 from transformers import AutoTokenizer, AutoModelForCausalLM, TextStreamer
 from openai import OpenAI
@@ -212,11 +216,14 @@ def main():
     parser.add_argument('--temperature', default=0.9, type=float, help="生成温度，控制随机性（0-1，越大越随机）")
     parser.add_argument('--top_p', default=0.9, type=float, help="nucleus采样阈值（0-1）")
     parser.add_argument('--show_speed', default=0, type=int, help="显示decode速度（tokens/s）")
-    parser.add_argument('--device', default='cuda' if torch.cuda.is_available() else 'cpu', type=str, help="运行设备")
     parser.add_argument('--api_base_url', default="http://localhost:11434/v1", type=str, help="OpenAI兼容接口的base_url")
     parser.add_argument('--api_key', default='sk-123', type=str, help="OpenAI兼容接口的api_key")
     parser.add_argument('--api_model', default='jingyaogong/minimind-3:latest', type=str, help="API请求时使用的模型名称")
     parser.add_argument('--stream', default=1, type=int, help="API模式下是否流式输出（0=否，1=是）")
+    parser.add_argument('--device',
+                        default='cuda' if torch.cuda.is_available() else ('npu' if torch_npu and torch_npu.npu.is_available() else 'cpu'),
+                        type=str,
+                        help="运行设备")
     args = parser.parse_args()
 
     model = tokenizer = client = None
diff --git a/scripts/serve_openai_api.py b/scripts/serve_openai_api.py
index 0550d4a..06612da 100644
--- a/scripts/serve_openai_api.py
+++ b/scripts/serve_openai_api.py
@@ -8,6 +8,10 @@ __package__ = "scripts"
 sys.path.append(os.path.abspath(os.path.join(os.path.dirname(__file__), '..')))
 import time
 import torch
+try:
+    import torch_npu
+except:
+    torch_npu = None
 import warnings
 import uvicorn
 
@@ -238,7 +242,10 @@ if __name__ == "__main__":
     parser.add_argument('--max_seq_len', default=8192, type=int, help="最大序列长度")
     parser.add_argument('--use_moe', default=0, type=int, choices=[0, 1], help="是否使用MoE架构（0=否，1=是）")
     parser.add_argument('--inference_rope_scaling', default=False, action='store_true', help="启用RoPE位置编码外推（4倍，仅解决位置编码问题）")
-    parser.add_argument('--device', default='cuda' if torch.cuda.is_available() else 'cpu', type=str, help="运行设备")
+    parser.add_argument('--device',
+                        default='cuda' if torch.cuda.is_available() else ('npu' if torch_npu and torch_npu.npu.is_available() else 'cpu'),
+                        type=str,
+                        help="运行设备")
     args = parser.parse_args()
     device = args.device
     model, tokenizer = init_model(args)
diff --git a/scripts/web_demo.py b/scripts/web_demo.py
index b29030d..d433a53 100644
--- a/scripts/web_demo.py
+++ b/scripts/web_demo.py
@@ -5,6 +5,10 @@ import os
 from threading import Thread
 
 import torch
+try:
+    import torch_npu
+except:
+    torch_npu = None
 import numpy as np
 import streamlit as st
 from transformers import AutoModelForCausalLM, AutoTokenizer, TextIteratorStreamer
@@ -67,7 +71,7 @@ st.markdown("""
     </style>
 """, unsafe_allow_html=True)
 
-device = "cuda" if torch.cuda.is_available() else "cpu"
+device = "cuda" if torch.cuda.is_available() else ("npu" if torch_npu and torch_npu.npu.is_available() else "cpu")
 
 # 多语言文本
 LANG_TEXTS = {
diff --git a/trainer/train_agent.py b/trainer/train_agent.py
index e652c3e..1d22f1c 100644
--- a/trainer/train_agent.py
+++ b/trainer/train_agent.py
@@ -13,6 +13,10 @@ import signal
 import argparse
 import warnings
 import torch
+try:
+    import torch_npu
+except:
+    torch_npu = None
 import torch.nn.functional as F
 import torch.distributed as dist
 from contextlib import nullcontext
@@ -377,7 +381,6 @@ if __name__ == "__main__":
     parser.add_argument("--epochs", type=int, default=1, help="训练轮数")
     parser.add_argument("--batch_size", type=int, default=2, help="批次大小")
     parser.add_argument("--learning_rate", type=float, default=3e-7, help="学习率")
-    parser.add_argument("--device", type=str, default="cuda:0" if torch.cuda.is_available() else "cpu", help="训练设备")
     parser.add_argument("--dtype", type=str, default="bfloat16", help="数据类型 bfloat16/float16")
     parser.add_argument("--num_workers", type=int, default=8, help="数据加载线程数")
     parser.add_argument("--accumulation_steps", type=int, default=1, help="梯度累积步数")
@@ -409,10 +412,14 @@ if __name__ == "__main__":
     parser.add_argument("--sglang_base_url", type=str, default="http://localhost:8998", help="SGLang服务器URL")
     parser.add_argument("--sglang_model_path", type=str, default="../model", help="SGLang tokenizer路径")
     parser.add_argument("--sglang_shared_path", type=str, default="./sglang_ckpt_agent", help="SGLang共享存储路径")
+    parser.add_argument("--device",
+                        type=str,
+                        default="cuda:0" if torch.cuda.is_available() else ("npu:0" if torch_npu and torch_npu.npu.is_available() else "cpu"),
+                        help="训练设备")
     args = parser.parse_args()
 
-    local_rank = init_distributed_mode()
-    if dist.is_initialized(): args.device = f"cuda:{local_rank}"
+    local_rank = init_distributed_mode("nccl" if torch.cuda.is_available() else "hccl")
+    if dist.is_initialized(): args.device = f"cuda:{local_rank}" if torch.cuda.is_available() else f"npu:{local_rank}"
     setup_seed(42 + (dist.get_rank() if dist.is_initialized() else 0))
 
     os.makedirs(args.save_dir, exist_ok=True)
@@ -420,7 +427,7 @@ if __name__ == "__main__":
                                max_seq_len=args.max_seq_len + args.max_gen_len, use_moe=bool(args.use_moe))
     ckp_data = lm_checkpoint(lm_config, weight=args.save_weight, save_dir='../checkpoints') if args.from_resume == 1 else None
 
-    device_type = "cuda" if "cuda" in args.device else "cpu"
+    device_type = "cuda" if "cuda" in args.device else ("npu" if "npu" in args.device else "cpu")
     dtype = torch.bfloat16 if args.dtype == "bfloat16" else torch.float16
     autocast_ctx = nullcontext() if device_type == "cpu" else torch.cuda.amp.autocast(dtype=dtype)
 
diff --git a/trainer/train_distillation.py b/trainer/train_distillation.py
index 63b6025..f496893 100644
--- a/trainer/train_distillation.py
+++ b/trainer/train_distillation.py
@@ -8,6 +8,10 @@ import argparse
 import time
 import warnings
 import torch
+try:
+    import torch_npu
+except:
+    torch_npu = None
 import torch.nn.functional as F
 import torch.distributed as dist
 from contextlib import nullcontext
@@ -150,7 +154,6 @@ if __name__ == "__main__":
     parser.add_argument("--epochs", type=int, default=6, help="训练轮数")
     parser.add_argument("--batch_size", type=int, default=32, help="batch size")
     parser.add_argument("--learning_rate", type=float, default=5e-6, help="初始学习率")
-    parser.add_argument("--device", type=str, default="cuda:0" if torch.cuda.is_available() else "cpu", help="训练设备")
     parser.add_argument("--dtype", type=str, default="bfloat16", help="混合精度类型")
     parser.add_argument("--num_workers", type=int, default=8, help="数据加载线程数")
     parser.add_argument("--accumulation_steps", type=int, default=1, help="梯度累积步数")
@@ -173,11 +176,15 @@ if __name__ == "__main__":
     parser.add_argument("--use_wandb", action="store_true", help="是否使用wandb")
     parser.add_argument("--wandb_project", type=str, default="MiniMind-Distillation", help="wandb项目名")
     parser.add_argument("--use_compile", default=0, type=int, choices=[0, 1], help="是否使用torch.compile加速（0=否，1=是）")
+    parser.add_argument("--device",
+                        type=str,
+                        default="cuda:0" if torch.cuda.is_available() else ("npu:0" if torch_npu and torch_npu.npu.is_available() else "cpu"),
+                        help="训练设备")
     args = parser.parse_args()
 
     # ========== 1. 初始化环境和随机种子 ==========
-    local_rank = init_distributed_mode()
-    if dist.is_initialized(): args.device = f"cuda:{local_rank}"
+    local_rank = init_distributed_mode("nccl" if torch.cuda.is_available() else "hccl")
+    if dist.is_initialized(): args.device = f"cuda:{local_rank}" if torch.cuda.is_available() else f"npu:{local_rank}"
     setup_seed(42 + (dist.get_rank() if dist.is_initialized() else 0))
     
     # ========== 2. 配置目录、模型参数、检查ckp ==========
@@ -187,7 +194,7 @@ if __name__ == "__main__":
     ckp_data = lm_checkpoint(lm_config_student, weight=args.save_weight, save_dir='../checkpoints') if args.from_resume==1 else None
     
     # ========== 3. 设置混合精度 ==========
-    device_type = "cuda" if "cuda" in args.device else "cpu"
+    device_type = "cuda" if "cuda" in args.device else ("npu" if "npu" in args.device else "cpu")
     dtype = torch.bfloat16 if args.dtype == "bfloat16" else torch.float16
     autocast_ctx = nullcontext() if device_type == "cpu" else torch.cuda.amp.autocast(dtype=dtype)
     
diff --git a/trainer/train_dpo.py b/trainer/train_dpo.py
index 558d108..eeea5d9 100644
--- a/trainer/train_dpo.py
+++ b/trainer/train_dpo.py
@@ -8,6 +8,10 @@ import argparse
 import time
 import warnings
 import torch
+try:
+    import torch_npu
+except:
+    torch_npu = None
 import torch.nn.functional as F
 import torch.distributed as dist
 from contextlib import nullcontext
@@ -134,7 +138,6 @@ if __name__ == "__main__":
     parser.add_argument("--epochs", type=int, default=1, help="训练轮数")
     parser.add_argument("--batch_size", type=int, default=4, help="batch size")
     parser.add_argument("--learning_rate", type=float, default=4e-8, help="初始学习率（建议<=5e-8避免遗忘）")
-    parser.add_argument("--device", type=str, default="cuda:0" if torch.cuda.is_available() else "cpu", help="训练设备")
     parser.add_argument("--dtype", type=str, default="bfloat16", help="混合精度类型")
     parser.add_argument("--num_workers", type=int, default=8, help="数据加载线程数")
     parser.add_argument("--accumulation_steps", type=int, default=1, help="梯度累积步数")
@@ -152,11 +155,15 @@ if __name__ == "__main__":
     parser.add_argument("--use_wandb", action="store_true", help="是否使用wandb")
     parser.add_argument("--wandb_project", type=str, default="MiniMind-DPO", help="wandb项目名")
     parser.add_argument("--use_compile", default=0, type=int, choices=[0, 1], help="是否使用torch.compile加速（0=否，1=是）")
+    parser.add_argument("--device",
+                        type=str,
+                        default="cuda:0" if torch.cuda.is_available() else ("npu:0" if torch_npu and torch_npu.npu.is_available() else "cpu"),
+                        help="训练设备")
     args = parser.parse_args()
 
     # ========== 1. 初始化环境和随机种子 ==========
-    local_rank = init_distributed_mode()
-    if dist.is_initialized(): args.device = f"cuda:{local_rank}"
+    local_rank = init_distributed_mode("nccl" if torch.cuda.is_available() else "hccl")
+    if dist.is_initialized(): args.device = f"cuda:{local_rank}" if torch.cuda.is_available() else f"npu:{local_rank}"
     setup_seed(42 + (dist.get_rank() if dist.is_initialized() else 0))
     
     # ========== 2. 配置目录、模型参数、检查ckp ==========
@@ -165,7 +172,7 @@ if __name__ == "__main__":
     ckp_data = lm_checkpoint(lm_config, weight=args.save_weight, save_dir='../checkpoints') if args.from_resume==1 else None
     
     # ========== 3. 设置混合精度 ==========
-    device_type = "cuda" if "cuda" in args.device else "cpu"
+    device_type = "cuda" if "cuda" in args.device else ("npu" if "npu" in args.device else "cpu")
     dtype = torch.bfloat16 if args.dtype == "bfloat16" else torch.float16
     autocast_ctx = nullcontext() if device_type == "cpu" else torch.cuda.amp.autocast(dtype=dtype)
     
diff --git a/trainer/train_full_sft.py b/trainer/train_full_sft.py
index 51c755b..90969b7 100644
--- a/trainer/train_full_sft.py
+++ b/trainer/train_full_sft.py
@@ -8,6 +8,10 @@ import argparse
 import time
 import warnings
 import torch
+try:
+    import torch_npu
+except:
+    torch_npu = None
 import torch.distributed as dist
 from contextlib import nullcontext
 from torch import optim, nn
@@ -87,7 +91,6 @@ if __name__ == "__main__":
     parser.add_argument("--epochs", type=int, default=2, help="训练轮数")
     parser.add_argument("--batch_size", type=int, default=16, help="batch size")
     parser.add_argument("--learning_rate", type=float, default=1e-5, help="初始学习率")
-    parser.add_argument("--device", type=str, default="cuda:0" if torch.cuda.is_available() else "cpu", help="训练设备")
     parser.add_argument("--dtype", type=str, default="bfloat16", help="混合精度类型")
     parser.add_argument("--num_workers", type=int, default=8, help="数据加载线程数")
     parser.add_argument("--accumulation_steps", type=int, default=1, help="梯度累积步数")
@@ -104,11 +107,15 @@ if __name__ == "__main__":
     parser.add_argument("--use_wandb", action="store_true", help="是否使用wandb")
     parser.add_argument("--wandb_project", type=str, default="MiniMind-Full-SFT", help="wandb项目名")
     parser.add_argument("--use_compile", default=0, type=int, choices=[0, 1], help="是否使用torch.compile加速（0=否，1=是）")
+    parser.add_argument("--device",
+                        type=str,
+                        default="cuda:0" if torch.cuda.is_available() else ("npu:0" if torch_npu and torch_npu.npu.is_available() else "cpu"),
+                        help="训练设备")
     args = parser.parse_args()
 
     # ========== 1. 初始化环境和随机种子 ==========
-    local_rank = init_distributed_mode()
-    if dist.is_initialized(): args.device = f"cuda:{local_rank}"
+    local_rank = init_distributed_mode("nccl" if torch.cuda.is_available() else "hccl")
+    if dist.is_initialized(): args.device = f"cuda:{local_rank}" if torch.cuda.is_available() else f"npu:{local_rank}"
     setup_seed(42 + (dist.get_rank() if dist.is_initialized() else 0))
     
     # ========== 2. 配置目录、模型参数、检查ckp ==========
@@ -117,7 +124,7 @@ if __name__ == "__main__":
     ckp_data = lm_checkpoint(lm_config, weight=args.save_weight, save_dir='../checkpoints') if args.from_resume==1 else None
     
     # ========== 3. 设置混合精度 ==========
-    device_type = "cuda" if "cuda" in args.device else "cpu"
+    device_type = "cuda" if "cuda" in args.device else ("npu" if "npu" in args.device else "cpu")
     dtype = torch.bfloat16 if args.dtype == "bfloat16" else torch.float16
     autocast_ctx = nullcontext() if device_type == "cpu" else torch.cuda.amp.autocast(dtype=dtype)
     
diff --git a/trainer/train_grpo.py b/trainer/train_grpo.py
index 52f7c7d..2cea524 100755
--- a/trainer/train_grpo.py
+++ b/trainer/train_grpo.py
@@ -10,6 +10,10 @@ import re
 import gc
 import warnings
 import torch
+try:
+    import torch_npu
+except:
+    torch_npu = None
 import torch.nn.functional as F
 import torch.distributed as dist
 from transformers import AutoTokenizer
@@ -209,7 +213,6 @@ if __name__ == "__main__":
     parser.add_argument("--epochs", type=int, default=1, help="训练轮数")
     parser.add_argument("--batch_size", type=int, default=2, help="batch size")
     parser.add_argument("--learning_rate", type=float, default=3e-7, help="初始学习率")
-    parser.add_argument("--device", type=str, default="cuda:0" if torch.cuda.is_available() else "cpu", help="训练设备")
     parser.add_argument("--dtype", type=str, default="bfloat16", help="混合精度类型")
     parser.add_argument("--num_workers", type=int, default=8, help="数据加载线程数")
     parser.add_argument("--accumulation_steps", type=int, default=1, help="梯度累积步数")
@@ -240,11 +243,15 @@ if __name__ == "__main__":
     parser.add_argument("--sglang_base_url", type=str, default="http://localhost:8998", help="SGLang服务器URL")
     parser.add_argument("--sglang_model_path", type=str, default="../model", help="SGLang tokenizer路径")
     parser.add_argument("--sglang_shared_path", type=str, default="./sglang_ckpt_grpo", help="SGLang共享存储路径")
+    parser.add_argument("--device",
+                        type=str,
+                        default="cuda:0" if torch.cuda.is_available() else ("npu:0" if torch_npu and torch_npu.npu.is_available() else "cpu"),
+                        help="训练设备")
     args = parser.parse_args()
 
     # ========== 1. 初始化环境和随机种子 ==========
-    local_rank = init_distributed_mode()
-    if dist.is_initialized(): args.device = f"cuda:{local_rank}"
+    local_rank = init_distributed_mode("nccl" if torch.cuda.is_available() else "hccl")
+    if dist.is_initialized(): args.device = f"cuda:{local_rank}" if torch.cuda.is_available() else f"npu:{local_rank}"
     setup_seed(42 + (dist.get_rank() if dist.is_initialized() else 0))
     
     # ========== 2. 配置目录、模型参数、检查ckp ==========
@@ -254,7 +261,7 @@ if __name__ == "__main__":
     ckp_data = lm_checkpoint(lm_config, weight=args.save_weight, save_dir='../checkpoints') if args.from_resume==1 else None
     
     # ========== 3. 设置混合精度 ==========
-    device_type = "cuda" if "cuda" in args.device else "cpu"
+    device_type = "cuda" if "cuda" in args.device else ("npu" if "npu" in args.device else "cpu")
     dtype = torch.bfloat16 if args.dtype == "bfloat16" else torch.float16
     autocast_ctx = nullcontext() if device_type == "cpu" else torch.cuda.amp.autocast(dtype=dtype)
     
diff --git a/trainer/train_lora.py b/trainer/train_lora.py
index 974bf37..ba33fc4 100644
--- a/trainer/train_lora.py
+++ b/trainer/train_lora.py
@@ -8,6 +8,10 @@ import argparse
 import time
 import warnings
 import torch
+try:
+    import torch_npu
+except:
+    torch_npu = None
 import torch.distributed as dist
 from contextlib import nullcontext
 from torch import optim, nn
@@ -81,7 +85,6 @@ if __name__ == "__main__":
     parser.add_argument("--epochs", type=int, default=10, help="训练轮数")
     parser.add_argument("--batch_size", type=int, default=32, help="batch size")
     parser.add_argument("--learning_rate", type=float, default=1e-4, help="初始学习率")
-    parser.add_argument("--device", type=str, default="cuda:0" if torch.cuda.is_available() else "cpu", help="训练设备")
     parser.add_argument("--dtype", type=str, default="bfloat16", help="混合精度类型")
     parser.add_argument("--num_workers", type=int, default=8, help="数据加载线程数")
     parser.add_argument("--accumulation_steps", type=int, default=1, help="梯度累积步数")
@@ -98,11 +101,15 @@ if __name__ == "__main__":
     parser.add_argument("--use_wandb", action="store_true", help="是否使用wandb")
     parser.add_argument("--wandb_project", type=str, default="MiniMind-LoRA", help="wandb项目名")
     parser.add_argument("--use_compile", default=0, type=int, choices=[0, 1], help="是否使用torch.compile加速（0=否，1=是）")
+    parser.add_argument("--device",
+                        type=str,
+                        default="cuda:0" if torch.cuda.is_available() else ("npu:0" if torch_npu and torch_npu.npu.is_available() else "cpu"),
+                        help="训练设备")
     args = parser.parse_args()
 
     # ========== 1. 初始化环境和随机种子 ==========
-    local_rank = init_distributed_mode()
-    if dist.is_initialized(): args.device = f"cuda:{local_rank}"
+    local_rank = init_distributed_mode("nccl" if torch.cuda.is_available() else "hccl")
+    if dist.is_initialized(): args.device = f"cuda:{local_rank}" if torch.cuda.is_available() else f"npu:{local_rank}"
     setup_seed(42 + (dist.get_rank() if dist.is_initialized() else 0))
     
     # ========== 2. 配置目录、模型参数、检查ckp ==========
@@ -111,7 +118,7 @@ if __name__ == "__main__":
     ckp_data = lm_checkpoint(lm_config, weight=args.lora_name, save_dir='../checkpoints') if args.from_resume==1 else None
     
     # ========== 3. 设置混合精度 ==========
-    device_type = "cuda" if "cuda" in args.device else "cpu"
+    device_type = "cuda" if "cuda" in args.device else ("npu" if "npu" in args.device else "cpu")
     dtype = torch.bfloat16 if args.dtype == "bfloat16" else torch.float16
     autocast_ctx = nullcontext() if device_type == "cpu" else torch.cuda.amp.autocast(dtype=dtype)
     
diff --git a/trainer/train_ppo.py b/trainer/train_ppo.py
index 228fb68..500aac5 100644
--- a/trainer/train_ppo.py
+++ b/trainer/train_ppo.py
@@ -9,6 +9,10 @@ import math
 import re
 import warnings
 import torch
+try:
+    import torch_npu
+except:
+    torch_npu = None
 import torch.distributed as dist
 import torch.nn.functional as F
 from transformers import AutoTokenizer
@@ -300,7 +304,6 @@ if __name__ == "__main__":
     parser.add_argument("--batch_size", type=int, default=2, help="batch size")
     parser.add_argument("--learning_rate", type=float, default=3e-7, help="Actor学习率")
     parser.add_argument("--critic_learning_rate", type=float, default=5e-7, help="Critic学习率")
-    parser.add_argument("--device", type=str, default="cuda:0" if torch.cuda.is_available() else "cpu", help="训练设备")
     parser.add_argument("--dtype", type=str, default="bfloat16", help="混合精度类型")
     parser.add_argument("--num_workers", type=int, default=8, help="数据加载线程数")
     parser.add_argument("--accumulation_steps", type=int, default=1, help="梯度累积步数")
@@ -335,11 +338,15 @@ if __name__ == "__main__":
     parser.add_argument("--sglang_base_url", type=str, default="http://localhost:8998", help="SGLang服务器URL")
     parser.add_argument("--sglang_model_path", type=str, default="../model", help="SGLang tokenizer路径")
     parser.add_argument("--sglang_shared_path", type=str, default="./sglang_ckpt_ppo", help="SGLang共享存储路径")
+    parser.add_argument("--device",
+                        type=str,
+                        default="cuda:0" if torch.cuda.is_available() else ("npu:0" if torch_npu and torch_npu.npu.is_available() else "cpu"),
+                        help="训练设备")
     args = parser.parse_args()
 
     # ========== 1. 初始化环境和随机种子 ==========
-    local_rank = init_distributed_mode()
-    if dist.is_initialized(): args.device = f"cuda:{local_rank}"
+    local_rank = init_distributed_mode("nccl" if torch.cuda.is_available() else "hccl")
+    if dist.is_initialized(): args.device = f"cuda:{local_rank}" if torch.cuda.is_available() else f"npu:{local_rank}"
     setup_seed(42 + (dist.get_rank() if dist.is_initialized() else 0))
     
     # ========== 2. 配置目录、模型参数、检查ckp ==========
@@ -348,7 +355,7 @@ if __name__ == "__main__":
     ckp_data = lm_checkpoint(lm_config, weight=args.save_weight, save_dir='../checkpoints') if args.from_resume==1 else None
     
     # ========== 3. 设置混合精度 ==========
-    device_type = "cuda" if "cuda" in args.device else "cpu"
+    device_type = "cuda" if "cuda" in args.device else ("npu" if "npu" in args.device else "cpu")
     dtype = torch.bfloat16 if args.dtype == "bfloat16" else torch.float16
     autocast_ctx = nullcontext() if device_type == "cpu" else torch.cuda.amp.autocast(dtype=dtype)
     
diff --git a/trainer/train_pretrain.py b/trainer/train_pretrain.py
index 5d33de4..19d6e2a 100644
--- a/trainer/train_pretrain.py
+++ b/trainer/train_pretrain.py
@@ -8,6 +8,10 @@ import argparse
 import time
 import warnings
 import torch
+try:
+    import torch_npu
+except:
+    torch_npu = None
 import torch.distributed as dist
 from contextlib import nullcontext
 from torch import optim, nn
@@ -86,7 +90,6 @@ if __name__ == "__main__":
     parser.add_argument("--epochs", type=int, default=2, help="训练轮数")
     parser.add_argument("--batch_size", type=int, default=32, help="batch size")
     parser.add_argument("--learning_rate", type=float, default=5e-4, help="初始学习率")
-    parser.add_argument("--device", type=str, default="cuda:0" if torch.cuda.is_available() else "cpu", help="训练设备")
     parser.add_argument("--dtype", type=str, default="bfloat16", help="混合精度类型")
     parser.add_argument("--num_workers", type=int, default=8, help="数据加载线程数")
     parser.add_argument("--accumulation_steps", type=int, default=8, help="梯度累积步数")
@@ -103,11 +106,14 @@ if __name__ == "__main__":
     parser.add_argument("--use_wandb", action="store_true", help="是否使用wandb")
     parser.add_argument("--wandb_project", type=str, default="MiniMind-Pretrain", help="wandb项目名")
     parser.add_argument("--use_compile", default=0, type=int, choices=[0, 1], help="是否使用torch.compile加速（0=否，1=是）")
+    parser.add_argument("--device", type=str,
+                        default="cuda:0" if torch.cuda.is_available() else ("npu:0" if torch_npu and torch_npu.npu.is_available() else "cpu"),
+                        help="训练设备")
     args = parser.parse_args()
 
     # ========== 1. 初始化环境和随机种子 ==========
-    local_rank = init_distributed_mode()
-    if dist.is_initialized(): args.device = f"cuda:{local_rank}"
+    local_rank = init_distributed_mode("nccl" if torch.cuda.is_available() else "hccl")
+    if dist.is_initialized(): args.device = f"cuda:{local_rank}" if torch.cuda.is_available() else f"npu:{local_rank}"
     setup_seed(42 + (dist.get_rank() if dist.is_initialized() else 0))
     
     # ========== 2. 配置目录、模型参数、检查ckp ==========
@@ -116,7 +122,7 @@ if __name__ == "__main__":
     ckp_data = lm_checkpoint(lm_config, weight=args.save_weight, save_dir='../checkpoints') if args.from_resume==1 else None
     
     # ========== 3. 设置混合精度 ==========
-    device_type = "cuda" if "cuda" in args.device else "cpu"
+    device_type = "cuda" if "cuda" in args.device else ("npu" if "npu" in args.device else "cpu")
     dtype = torch.bfloat16 if args.dtype == "bfloat16" else torch.float16
     autocast_ctx = nullcontext() if device_type == "cpu" else torch.cuda.amp.autocast(dtype=dtype)
     
diff --git a/trainer/trainer_utils.py b/trainer/trainer_utils.py
index b6211d5..8689525 100644
--- a/trainer/trainer_utils.py
+++ b/trainer/trainer_utils.py
@@ -41,11 +41,11 @@ def get_lr(current_step, total_steps, lr):
     return lr*(0.1 + 0.45*(1 + math.cos(math.pi * current_step / total_steps)))
 
 
-def init_distributed_mode():
+def init_distributed_mode(backend="nccl"):
     if int(os.environ.get("RANK", -1)) == -1:
         return 0  # 非DDP模式
 
-    dist.init_process_group(backend="nccl")
+    dist.init_process_group(backend=backend)
     local_rank = int(os.environ["LOCAL_RANK"])
     torch.cuda.set_device(local_rank)
     return local_rank