24G显存出现CUDA error: out of memory正常吗？

from pathlib import Path
import importlib.util
import torch
import torchaudio
from transformers import AutoModel, AutoProcessor
# Disable the broken cuDNN SDPA backend
torch.backends.cuda.enable_cudnn_sdp(False)
# Keep these enabled as fallbacks
torch.backends.cuda.enable_flash_sdp(True)
torch.backends.cuda.enable_mem_efficient_sdp(True)
torch.backends.cuda.enable_math_sdp(True)


pretrained_model_name_or_path = "OpenMOSS-Team/MOSS-TTS-v1.5"
device = "cuda" if torch.cuda.is_available() else "cpu"
dtype = torch.bfloat16 if device == "cuda" else torch.float32

def resolve_attn_implementation() -> str:
   # Prefer FlashAttention 2 when package + device conditions are met.
   if (
       device == "cuda"
       and importlib.util.find_spec("flash_attn") is not None
       and dtype in {torch.float16, torch.bfloat16}
   ):
       major, _ = torch.cuda.get_device_capability()
       if major >= 8:
           return "flash_attention_2"

   # CUDA fallback: use PyTorch SDPA kernels.
   if device == "cuda":
       return "sdpa"

   # CPU fallback.
   return "eager"


attn_implementation = resolve_attn_implementation()
print(f"[INFO] Using attn_implementation={attn_implementation}")

processor = AutoProcessor.from_pretrained(
   pretrained_model_name_or_path,
   trust_remote_code=True,
)
processor.audio_tokenizer = processor.audio_tokenizer.to(device)

text_1 = "法国天文学家希望与巴黎连接的重要地点之一是乌拉尼堡，即16世纪先驱天文学家第谷·布拉赫的旧天文台。"

# Use audio from ./assets/audio to avoid downloading from the cloud.
ref_audio_1 = "/data/code/python/django/dub_sys/media/878/index-tts/101.wav"


conversations = [
   # Direct TTS (no reference). Language tags are recommended in v1.5.
   [processor.build_user_message(text=text_1)],
  
   # Voice cloning (with reference)
   [processor.build_user_message(text=text_1, reference=[ref_audio_1])],
]

model = AutoModel.from_pretrained(
   pretrained_model_name_or_path,
   trust_remote_code=True,
   attn_implementation=attn_implementation,
   torch_dtype=dtype,
).to(device)
model.eval()

batch_size = 1

save_dir = Path("inference_root")
save_dir.mkdir(exist_ok=True, parents=True)
sample_idx = 0
with torch.no_grad():
   for start in range(0, len(conversations), batch_size):
       batch_conversations = conversations[start : start + batch_size]
       batch = processor(batch_conversations, mode="generation")
       input_ids = batch["input_ids"].to(device)
       attention_mask = batch["attention_mask"].to(device)

       outputs = model.generate(
           input_ids=input_ids,
           attention_mask=attention_mask,
           max_new_tokens=4096,
       )

       for message in processor.decode(outputs):
           audio = message.audio_codes_list[0]
           out_path = save_dir / f"sample{sample_idx}.wav"
           sample_idx += 1
           torchaudio.save(out_path, audio.unsqueeze(0), processor.model_config.sampling_rate)

Start by running the provided Python inference snippet with the reported 24G CUDA setup and inspect where generation runs out of memory. Review the model loading, attention backend selection, batch size, and max_new_tokens settings shown in the issue. Done means establishing whether the failure is reproducible and documenting the relevant conditions or required configuration.