Can't load config for ‘"XXXXX"

[jiyilin]

When I used the AI toolkit to train the Lora model of z-image, I found that the pre trained model could not be successfully loaded

The environment is AWS g6e.2xlarge in us-west-2b
The AI toolkit version is pull requests=feature: support nested dataset folders

I checked the bumping face cache and the data inside is as follows
├── models--Tongyi-MAI--Z-Image-Turbo
│   ├── blobs
│   │   └── c579190a5e03d602a2fd9647221c9d0d9441f150.incomplete
│   ├── refs
│   │   └── main
│   └── snapshots
│   └── 5f4b9cbb80cc95ba44fe6667dfd75710f7db2947
│   └── transformer
└── models--ostris--Z-Image-De-Turbo
├── blobs
│   └── f0b861072ec14990e36d7cf01ff62c418a2152e4.incomplete
├── refs
│   └── main
└── snapshots
└── 9da355082d6374634080361054e712a08cc54af1
└── transformer

13 directories, 4 files

I tried manually pulling Tongyi MAI/Z Image Turbo to the bumping face cache hub, but I found that this would cause [Errno 95] Operation not supported

My job Model Name or Path = ostris/Z-Image-De-Turbo, Options are all turned off, Quantization uses the default float8, Target Type = LoRA, Linear Rank = 32

[jiyilin]

my os system is Ubuntu 2404, based on Deep Learning OSS Nvidia Driver AMI GPU PyTorch 2.9 (Ubuntu 24.04) AMI