#7853 语音识别阶段出错[Nemotron-3.5-asr-0.6b(内置)] CUDA error: out of memoryCompile with `TORCH_USE_CUDA_DSA` to enable device-side ass

语音识别阶段出错[Nemotron-3.5-asr-0.6b(内置)] CUDA error: out of memory
Compile with TORCH_USE_CUDA_DSA to enable device-side assertions.
:Traceback (most recent call last):
File "videotrans\process\stt_nemotron.py", line 41, in nemotron_asr
File "E:\sp\win-pyvideotrans-v4.14\_internal\transformers\models\nemotron3_5_asr\generation_nemotron3_5_asr.py", line 35, in generate

return super().generate(inputs=inputs, generation_config=generation_config, **kwargs)

File "E:\sp\win-pyvideotrans-v4.14\_internal\transformers\models\nemotron_asr_streaming\generation_nemotron_asr_streaming.py", line 228, in generate

outputs = super().generate(inputs=inputs, generation_config=generation_config, **kwargs)

File "E:\sp\win-pyvideotrans-v4.14\_internal\transformers\models\parakeet\generation_parakeet.py", line 256, in generate

outputs = super().generate(inputs=inputs, generation_config=generation_config, **kwargs)

File "E:\sp\win-pyvideotrans-v4.14\_internal\torch\utils\_contextl
......
es\module.py", line 1762, in _call_impl

return forward_call(*args, **kwargs)

File "E:\sp\win-pyvideotrans-v4.14\_internal\transformers\models\nemotron_asr_streaming\modeling_nemotron_asr_streaming.py", line 730, in forward

hidden_states = self.act_fn(_mask_subsampled_frames(hidden_states, lengths))

File "E:\sp\win-pyvideotrans-v4.14\_internal\torch\nn\modules\module.py", line 1751, in _wrapped_call_impl

return self._call_impl(*args, **kwargs)

File "E:\sp\win-pyvideotrans-v4.14\_internal\torch\nn\modules\module.py", line 1762, in _call_impl

return forward_call(*args, **kwargs)

File "E:\sp\win-pyvideotrans-v4.14\_internal\torch\nn\modules\activation.py", line 133, in forward

return F.relu(input, inplace=self.inplace)

File "E:\sp\win-pyvideotrans-v4.14\_internal\torch\nn\functional.py", line 1704, in relu

result = torch.relu(input)

RuntimeError: CUDA error: out of memory
Compile with TORCH_USE_CUDA_DSA to enable device-side assertions.

self.cfg=[TaskCfgVTT]当前工作模式: 转录并翻译字幕
原始输入文件名: G:/YTB/[Ep.913] 한공이들 마음 한 조각을 가져갈 앳하트💘 💌찐디의 아이돌 한국어 사랑방 💌-mp4/[Ep.913] 한공이들 마음 한 조각을 가져갈 앳하트 찐디의 아이돌 한국어 사랑방.mp4,
输出结果保存到文件夹: G:/YTB/[Ep.913] 한공이들 마음 한 조각을 가져갈 앳하트💘 💌찐디의 아이돌 한국어 사랑방 💌-mp4/[Ep.913] 한공이들 마음 한 조각을 가져갈 앳하트 찐디의 아이돌 한국어 사랑방-mp4,
临时文件夹: E:/sp/win-pyvideotrans-v4.14/tmp/81112/58d78a9ac8
未选 清理已存在
已启用CUDA加速
未选 降噪
语音识别:Nemotron-3.5-asr-0.6b(内置), , 发音语言: 韩语, 断句方式:默认断句
翻译渠道:Gemini AI,原始语言:韩语,目标语言:简体中文, 未选 发送完整字幕
输出字幕格式: 目标语言在下双字幕
代理地址:http://127.0.0.1:7897