AssetCardFieldNotFoundError: The asset card 'seamlessM4T_v2_large' must have a field named 'model_type'.

[CaiJichang212]

import yaml

model_yaml = './src/seamless_communication/cards/seamlessM4T_v2_large_local.yaml'
vocoder_yaml = './src/seamless_communication/cards/vocoder_v2_local.yaml'

# 打开YAML文件并加载内容
with open(model_yaml, 'r') as f1, open(vocoder_yaml, 'r') as f2:
    model_metadata = yaml.safe_load(f1)
    vocoder_metadata = yaml.safe_load(f2)
    
from fairseq2.assets.card import AssetCard

model_card = AssetCard(model_metadata)
vocoder_card = AssetCard(vocoder_metadata)

translator = Translator(
    model_card,
    vocoder_card,
    # device=torch.device("cuda:0"),
    device=torch.device("cpu"),
    dtype=torch.float16,
)

AssetCardFieldNotFoundError: The asset card 'seamlessM4T_v2_large' must have a field named 'model_type'.
The weight of the model downloaded to the local is used by AssetCard. An error occurs during the loading process, so it is necessary to set the seamlessM4T model_type. I found that none of the seamlessM4T series models has set model_type, and the corresponding model_type has been set in other models. Looking forward to your answer.
通过AssetCard的方式使用下载到本地的模型权重，加载过程中出现报错，需要制定seamlessM4T的model_type，我发现seamlessM4T系列模型都没有设置model_type，其他模型都设置了对应的model_type?期待你的解答。