This is for bugs only
Did you already ask in the discord?
Yes/No
You verified that this is a bug and not a feature request or question by asking in the discord?
Yes/No
Describe the bug
{
"type": "diffusion_trainer",
"training_folder": "D:\\ai-toolkit\\output",
"sqlite_db_path": "D:\\ai-toolkit\\aitk_db.db",
"device": "cuda",
"trigger_word": "wys",
"performance_log_every": 10,
"network": {
"type": "lora",
"linear": 32,
"linear_alpha": 32,
"conv": 16,
"conv_alpha": 16,
"lokr_full_rank": true,
"lokr_factor": -1,
"network_kwargs": {
"ignore_if_contains": []
}
},
"save": {
"dtype": "fp16",
"save_every": 250,
"max_step_saves_to_keep": 4,
"save_format": "diffusers",
"push_to_hub": false
},
"datasets": [
{
"folder_path": "D:\\ai-toolkit\\datasets/wys",
"mask_path": null,
"mask_min_value": 0.1,
"default_caption": "",
"caption_ext": "txt",
"caption_dropout_rate": 0.05,
"cache_latents_to_disk": false,
"is_reg": false,
"network_weight": 1,
"resolution": [
1024
],
"controls": [],
"shrink_video_to_frames": true,
"num_frames": 1,
"do_i2v": true,
"flip_x": false,
"flip_y": false
}
],
"train": {
"batch_size": 1,
"bypass_guidance_embedding": false,
"steps": 3000,
"gradient_accumulation": 1,
"train_unet": true,
"train_text_encoder": false,
"gradient_checkpointing": true,
"noise_scheduler": "flowmatch",
"optimizer": "adamw8bit",
"timestep_type": "weighted",
"content_or_style": "balanced",
"optimizer_params": {
"weight_decay": 0.0001
},
"unload_text_encoder": false,
"cache_text_embeddings": false,
"lr": 0.0001,
"ema_config": {
"use_ema": false,
"ema_decay": 0.99
},
"skip_first_sample": false,
"force_first_sample": false,
"disable_sampling": false,
"dtype": "fp16",
"diff_output_preservation": false,
"diff_output_preservation_multiplier": 1,
"diff_output_preservation_class": "person",
"switch_boundary_every": 1,
"loss_type": "mse"
},
"model": {
"name_or_path": "Tongyi-MAI/Z-Image-Turbo",
"quantize": false,
"qtype": "qfloat8",
"quantize_te": false,
"qtype_te": "qfloat8",
"arch": "zimage:turbo",
"low_vram": false,
"model_kwargs": {},
"layer_offloading": false,
"layer_offloading_text_encoder_percent": 1,
"layer_offloading_transformer_percent": 1,
"assistant_lora_path": "ostris/zimage_turbo_training_adapter/zimage_turbo_training_adapter_v1.safetensors"
},
"sample": {
"sampler": "flowmatch",
"sample_every": 250,
"width": 1024,
"height": 1024,
"samples": [
{
"prompt": "wys\u62ff\u7740\u4e66\u5750\u5728\u516c\u56ed\u7684\u957f\u6905\u4e0a\u9605\u8bfb\uff0c\u6709\u6ed1\u6ed1\u68af"
}
],
"neg": "",
"seed": 42,
"walk_seed": true,
"guidance_scale": 1,
"sample_steps": 8,
"num_frames": 1,
"fps": 1
}
}
Using SQLite database at D:\ai-toolkit\aitk_db.db
Job ID: "46a61b96-a273-4b9b-a0b9-fe87b54a1874"
#############################################
# Running job: wys_lora
#############################################
Running 1 process
Loading ZImage model
Loading transformer
Fetching 3 files: 100%|##########| 3/3 [00:00, ?it/s]
Loading checkpoint shards: 100%|##########| 3/3 [00:20
Are all the sampled images completely black?
On my 3080 laptop graphics card also has this issue when training with the official model.
It might be caused by this reason. https://github.com/Tongyi-MAI/Z-Image/issues/15
I can use this model to sample normal images and train LoRA with this model. https://huggingface.co/dimitribarbot/Z-Image-Turbo-BF16
Are all the sampled images completely black? <img alt="Image" width="760" height="520" src="https://private-user-
During the training process, I did not set sampling and found that the model fails to generate the expected images when used in Comfy UI. After investigation, I discovered the issue was due to the model being corrupted. I re-downloaded the model from https://www.modelscope.cn/models/Tongyi-MAI/Z-Image-Turbo, and the problem was resolved. Now the model can train normally and display the loss rate.
Are all the sampled images completely black? <img alt="Image" width="760" height="520" src="https://private-user-
During the training process, I did not set sampling and found that the model fails to generate the expected images when used in Comfy UI. After investigation, I discovered the issue was due to the model being corrupted. I re-downloaded the model from https://www.modelscope.cn/models/Tongyi-MAI/Z-Image-Turbo, and the problem was resolved. Now the model can train normally and display the loss rate.
Could it be that the model I downloaded is also corrupted? I'll try re-downloading it as well.
ostris
