From ecc653871d305c79002d2630c7cf0d0e1d6bf1ed Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?=E5=BE=90=E7=BB=9C=E6=BA=9F?= <xjy46566696@gmail.com>
Date: Fri, 10 Jan 2025 19:26:36 +0800
Subject: [PATCH] Fix: export-onnx.py(expected all tensors to be on the same
 device) (#1699)
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

由于SenseVoiceSmall.from_pretrained()
    调用的funasr.auto.auto_model.AutoModel.build_model()默认device是cuda
    (在cuda available的环境中)
    ```py
    device = kwargs.get("device", "cuda")
    if not torch.cuda.is_available() or kwargs.get("ngpu", 1) == 0:
        device = "cpu"
        kwargs["batch_size"] = 1
    kwargs["device"] = device
    ```
    而export-onnx.py里的tensor默认都是cpu, 导致
    RuntimeError: Expected all tensors to be on the same device, but found at least two devices, cuda:0 and cpu
    所以直接在加载model的时候指定cpu
---
 scripts/sense-voice/export-onnx.py | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/scripts/sense-voice/export-onnx.py b/scripts/sense-voice/export-onnx.py
index 97d9a506..48b68636 100755
--- a/scripts/sense-voice/export-onnx.py
+++ b/scripts/sense-voice/export-onnx.py
@@ -119,7 +119,7 @@ def display_params(params):
 
 
 def main():
-    model, params = SenseVoiceSmall.from_pretrained(model="iic/SenseVoiceSmall")
+    model, params = SenseVoiceSmall.from_pretrained(model="iic/SenseVoiceSmall", device="cpu")
     display_params(params)
 
     generate_tokens(params)