虽然生成了缓存，但是训练的时候就报错，缓存的时候就提示vae报错，vae没问题啊，都能出图。而且重新下载的也是不行。
`2025-09-27 22:42:32 开始执行预缓存...
2025-09-27 22:42:32 开始执行预缓存 1/2 ...
2025-09-27 22:42:32 I:\AIProject\Loratrain\LoRAMaster.venv\Scripts\python.exe I:\AIProject\Loratrain\LoRAMaster\musubi-tuner\src\musubi_tuner\qwen_image_cache_latents.py --dataset_config I:\AIProject\Loratrain\LoRAMaster\samples\toml\qwen_image_saiboguofengji.toml --vae E:\ComfyUI_windows_portable\ComfyUI\models\vae\qwen_image_vae_t.safetensors
2025-09-27 22:42:39 INFO:main:Load dataset config from I:\AIProject\Loratrain\LoRAMaster\samples\toml\qwen_image_saiboguofengji.toml
2025-09-27 22:42:39 INFO:musubi_tuner.dataset.image_video_dataset:glob images in I:\LoRA_train\saiboguofengji\images
2025-09-27 22:42:39 INFO:musubi_tuner.dataset.image_video_dataset:found 35 images
2025-09-27 22:42:39 INFO:musubi_tuner.dataset.config_utils:[Dataset 0]
2025-09-27 22:42:39 is_image_dataset: True
2025-09-27 22:42:39 resolution: (1024, 1024)
2025-09-27 22:42:39 batch_size: 1
2025-09-27 22:42:39 num_repeats: 2
2025-09-27 22:42:39 caption_extension: ".txt"
2025-09-27 22:42:39 enable_bucket: False
2025-09-27 22:42:39 bucket_no_upscale: False
2025-09-27 22:42:39 cache_directory: "I:\LoRA_train\saiboguofengji\qwen_image_cache"
2025-09-27 22:42:39 debug_dataset: False
2025-09-27 22:42:39 image_directory: "I:\LoRA_train\saiboguofengji\images"
2025-09-27 22:42:39 image_jsonl_file: "None"
2025-09-27 22:42:39 fp_latent_window_size: 9
2025-09-27 22:42:39 fp_1f_clean_indices: None
2025-09-27 22:42:39 fp_1f_target_index: None
2025-09-27 22:42:39 fp_1f_no_post: False
2025-09-27 22:42:39 flux_kontext_no_resize_control: False
2025-09-27 22:42:39 qwen_image_edit_no_resize_control: False
2025-09-27 22:42:39 qwen_image_edit_control_resolution: None
2025-09-27 22:42:39
2025-09-27 22:42:39
2025-09-27 22:42:39 INFO:main:Loading VAE model from E:\ComfyUI_windows_portable\ComfyUI\models\vae\qwen_image_vae_t.safetensors
2025-09-27 22:42:39 INFO:musubi_tuner.qwen_image.qwen_image_utils:Initializing VAE
2025-09-27 22:42:40 INFO:musubi_tuner.qwen_image.qwen_image_utils:Loading VAE from E:\ComfyUI_windows_portable\ComfyUI\models\vae\qwen_image_vae_t.safetensors
2025-09-27 22:42:40 Trying to import sageattention
2025-09-27 22:42:40 Failed to import sageattention
2025-09-27 22:42:40 Traceback (most recent call last):
2025-09-27 22:42:40 File "I:\AIProject\Loratrain\LoRAMaster\musubi-tuner\src\musubi_tuner\qwen_image_cache_latents.py", line 123, in
2025-09-27 22:42:40 main()
2025-09-27 22:42:40 File "I:\AIProject\Loratrain\LoRAMaster\musubi-tuner\src\musubi_tuner\qwen_image_cache_latents.py", line 111, in main
2025-09-27 22:42:40 vae = qwen_image_utils.load_vae(args.vae, device=device, disable_mmap=True)
2025-09-27 22:42:40 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
2025-09-27 22:42:40 File "I:\AIProject\Loratrain\LoRAMaster\musubi-tuner\src\musubi_tuner\qwen_image\qwen_image_utils.py", line 548, in load_vae
2025-09-27 22:42:40 info = vae.load_state_dict(state_dict, strict=True, assign=True)
2025-09-27 22:42:40 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
2025-09-27 22:42:40 File "I:\AIProject\Loratrain\LoRAMaster.venv\Lib\site-packages\torch\nn\modules\module.py", line 2624, in load_state_dict
2025-09-27 22:42:40 raise RuntimeError(
2025-09-27 22:42:40 RuntimeError: Error(s) in loading state_dict for AutoencoderKLQwenImage:
2025-09-27 22:42:40 Missing key(s) in state_dict: "encoder.conv_in.weight", "encoder.conv_in.bias", "encoder.down_blocks.0.norm1.gamma", "encoder.down_blocks.0.conv1.weight", "encoder.down_blocks.0.conv1.bias", "encoder.down_blocks.0.norm2.gamma", "encoder.down_blocks.0.conv2.weight", "encoder.down_blocks.0.conv2.bias", "encoder.down_blocks.1.norm1.gamma", "encoder.down_blocks.1.conv1.weight", "encoder.down_blocks.1.conv1.bias", "encoder.down_blocks.1.norm2.gamma", "encoder.down_blocks.1.conv2.weight", "encoder.down_blocks.1.conv2.bias", "encoder.down_blocks.2.resample.1.weight", "encoder.down_blocks.2.resample.1.bias", "encoder.down_blocks.3.norm1.gamma", "encoder.down_blocks.3.conv1.weight", "encoder.down_blocks.3.conv1.bias", "encoder.down_blocks.3.norm2.gamma", "encoder.down_blocks.3.conv2.weight", "encoder.down_blocks.3.conv2.bias", "encoder.down_blocks.3.conv_shortcut.weight", "encoder.down_blocks.3.conv_shortcut.bias", "encoder.down_blocks.4.norm1.gamma", "encoder.down_blocks.4.conv1.weight", "encoder.down_blocks.4.conv1.bias", "encoder.down_blocks.4.norm2.gamma", "encoder.down_blocks.4.conv2.weight", "encoder.down_blocks.4.conv2.bias", "encoder.down_blocks.5.resample.1.weight", "encoder.down_blocks.5.resample.1.bias", "encoder.down_blocks.5.time_conv.weight", "encoder.down_blocks.5.time_conv.bias", "encoder.down_blocks.6.norm1.gamma", "encoder.down_blocks.6.conv1.weight", "encoder.down_blocks.6.conv1.bias", "encoder.down_blocks.6.norm2.gamma", "encoder.down_blocks.6.conv2.weight", "encoder.down_blocks.6.conv2.bias", "encoder.down_blocks.6.conv_shortcut.weight", "encoder.down_blocks.6.conv_shortcut.bias", "encoder.down_blocks.7.norm1.gamma", "encoder.down_blocks.7.conv1.weight", "encoder.down_blocks.7.conv1.bias", "encoder.down_blocks.7.norm2.gamma", "encoder.down_blocks.7.conv2.weight", "encoder.down_blocks.7.conv2.bias", "encoder.down_blocks.8.resample.1.weight", "encoder.down_blocks.8.resample.1.bias", "encoder.down_blocks.8.time_conv.weight", "encoder.down_blocks.8.time_conv.bias", "encoder.down_blocks.9.norm1.gamma", "encoder.down_blocks.9.conv1.weight", "encoder.down_blocks.9.conv1.bias", "encoder.down_blocks.9.norm2.gamma", "encoder.down_blocks.9.conv2.weight", "encoder.down_blocks.9.conv2.bias", "encoder.down_blocks.10.norm1.gamma", "encoder.down_blocks.10.conv1.weight", "encoder.down_blocks.10.conv1.bias", "encoder.down_blocks.10.norm2.gamma", "encoder.down_blocks.10.conv2.weight", "encoder.down_blocks.10.conv2.bias", "encoder.mid_block.attentions.0.norm.gamma", "encoder.mid_block.attentions.0.to_qkv.weight", "encoder.mid_block.attentions.0.to_qkv.bias", "encoder.mid_block.attentions.0.proj.weight", "encoder.mid_block.attentions.0.proj.bias", "encoder.mid_block.resnets.0.norm1.gamma", "encoder.mid_block.resnets.0.conv1.weight", "encoder.mid_block.resnets.0.conv1.bias", "encoder.mid_block.resnets.0.norm2.gamma", "encoder.mid_block.resnets.0.conv2.weight", "encoder.mid_block.resnets.0.conv2.bias", "encoder.mid_block.resnets.1.norm1.gamma", "encoder.mid_block.resnets.1.conv1.weight", "encoder.mid_block.resnets.1.conv1.bias", "encoder.mid_block.resnets.1.norm2.gamma", "encoder.mid_block.resnets.1.conv2.weight", "encoder.mid_block.resnets.1.conv2.bias", "encoder.norm_out.gamma", "encoder.conv_out.weight", "encoder.conv_out.bias", "quant_conv.weight", "quant_conv.bias", "post_quant_conv.weight", "post_quant_conv.bias", "decoder.conv_in.weight", "decoder.conv_in.bias", "decoder.mid_block.attentions.0.norm.gamma", "decoder.mid_block.attentions.0.to_qkv.weight", "decoder.mid_block.attentions.0.to_qkv.bias", "decoder.mid_block.attentions.0.proj.weight", "decoder.mid_block.attentions.0.proj.bias", "decoder.mid_block.resnets.0.norm1.gamma", "decoder.mid_block.resnets.0.conv1.weight", "decoder.mid_block.resnets.0.conv1.bias", "decoder.mid_block.resnets.0.norm2.gamma", "decoder.mid_block.resnets.0.conv2.weight", "decoder.mid_block.resnets.0.conv2.bias", "decoder.mid_block.resnets.1.norm1.gamma", "decoder.mid_block.resnets.1.conv1.weight", "decoder.mid_block.resnets.1.conv1.bias", "decoder.mid_block.resnets.1.norm2.gamma", "decoder.mid_block.resnets.1.conv2.weight", "decoder.mid_block.resnets.1.conv2.bias", "decoder.up_blocks.0.resnets.0.norm1.gamma", "decoder.up_blocks.0.resnets.0.conv1.weight", "decoder.up_blocks.0.resnets.0.conv1.bias", "decoder.up_blocks.0.resnets.0.norm2.gamma", "decoder.up_blocks.0.resnets.0.conv2.weight", "decoder.up_blocks.0.resnets.0.conv2.bias", "decoder.up_blocks.0.resnets.1.norm1.gamma", "decoder.up_blocks.0.resnets.1.conv1.weight", "decoder.up_blocks.0.resnets.1.conv1.bias", "decoder.up_blocks.0.resnets.1.norm2.gamma", "decoder.up_blocks.0.resnets.1.conv2.weight", "decoder.up_blocks.0.resnets.1.conv2.bias", "decoder.up_blocks.0.resnets.2.norm1.gamma", "decoder.up_blocks.0.resnets.2.conv1.weight", "decoder.up_blocks.0.resnets.2.conv1.bias", "decoder.up_blocks.0.resnets.2.norm2.gamma", "decoder.up_blocks.0.resnets.2.conv2.weight", "decoder.up_blocks.0.resnets.2.conv2.bias", "decoder.up_blocks.0.upsamplers.0.resample.1.weight", "decoder.up_blocks.0.upsamplers.0.resample.1.bias", "decoder.up_blocks.0.upsamplers.0.time_conv.weight", "decoder.up_blocks.0.upsamplers.0.time_conv.bias", "decoder.up_blocks.1.resnets.0.norm1.gamma", "decoder.up_blocks.1.resnets.0.conv1.weight", "decoder.up_blocks.1.resnets.0.conv1.bias", "decoder.up_blocks.1.resnets.0.norm2.gamma", "decoder.up_blocks.1.resnets.0.conv2.weight", "decoder.up_blocks.1.resnets.0.conv2.bias", "decoder.up_blocks.1.resnets.0.conv_shortcut.weight", "decoder.up_blocks.1.resnets.0.conv_shortcut.bias", "decoder.up_blocks.1.resnets.1.norm1.gamma", "decoder.up_blocks.1.resnets.1.conv1.weight", "decoder.up_blocks.1.resnets.1.conv1.bias", "decoder.up_blocks.1.resnets.1.norm2.gamma", "decoder.up_blocks.1.resnets.1.conv2.weight", "decoder.up_blocks.1.resnets.1.conv2.bias", "decoder.up_blocks.1.resnets.2.norm1.gamma", "decoder.up_blocks.1.resnets.2.conv1.weight", "decoder.up_blocks.1.resnets.2.conv1.bias", "decoder.up_blocks.1.resnets.2.norm2.gamma", "decoder.up_blocks.1.resnets.2.conv2.weight", "decoder.up_blocks.1.resnets.2.conv2.bias", "decoder.up_blocks.1.upsamplers.0.resample.1.weight", "decoder.up_blocks.1.upsamplers.0.resample.1.bias", "decoder.up_blocks.1.upsamplers.0.time_conv.weight", "decoder.up_blocks.1.upsamplers.0.time_conv.bias", "decoder.up_blocks.2.resnets.0.norm1.gamma", "decoder.up_blocks.2.resnets.0.conv1.weight", "decoder.up_blocks.2.resnets.0.conv1.bias", "decoder.up_blocks.2.resnets.0.norm2.gamma", "decoder.up_blocks.2.resnets.0.conv2.weight", "decoder.up_blocks.2.resnets.0.conv2.bias", "decoder.up_blocks.2.resnets.1.norm1.gamma", "decoder.up_blocks.2.resnets.1.conv1.weight", "decoder.up_blocks.2.resnets.1.conv1.bias", "decoder.up_blocks.2.resnets.1.norm2.gamma", "decoder.up_blocks.2.resnets.1.conv2.weight", "decoder.up_blocks.2.resnets.1.conv2.bias", "decoder.up_blocks.2.resnets.2.norm1.gamma", "decoder.up_blocks.2.resnets.2.conv1.weight", "decoder.up_blocks.2.resnets.2.conv1.bias", "decoder.up_blocks.2.resnets.2.norm2.gamma", "decoder.up_blocks.2.resnets.2.conv2.weight", "decoder.up_blocks.2.resnets.2.conv2.bias", "decoder.up_blocks.2.upsamplers.0.resample.1.weight", "decoder.up_blocks.2.upsamplers.0.resample.1.bias", "decoder.up_blocks.3.resnets.0.norm1.gamma", "decoder.up_blocks.3.resnets.0.conv1.weight", "decoder.up_blocks.3.resnets.0.conv1.bias", "decoder.up_blocks.3.resnets.0.norm2.gamma", "decoder.up_blocks.3.resnets.0.conv2.weight", "decoder.up_blocks.3.resnets.0.conv2.bias", "decoder.up_blocks.3.resnets.1.norm1.gamma", "decoder.up_blocks.3.resnets.1.conv1.weight", "decoder.up_blocks.3.resnets.1.conv1.bias", "decoder.up_blocks.3.resnets.1.norm2.gamma", "decoder.up_blocks.3.resnets.1.conv2.weight", "decoder.up_blocks.3.resnets.1.conv2.bias", "decoder.up_blocks.3.resnets.2.norm1.gamma", "decoder.up_blocks.3.resnets.2.conv1.weight", "decoder.up_blocks.3.resnets.2.conv1.bias", "decoder.up_blocks.3.resnets.2.norm2.gamma", "decoder.up_blocks.3.resnets.2.conv2.weight", "decoder.up_blocks.3.resnets.2.conv2.bias", "decoder.norm_out.gamma", "decoder.conv_out.weight", "decoder.conv_out.bias".
2025-09-27 22:42:40 Unexpected key(s) in state_dict: "conv1.bias", "conv1.weight", "conv2.bias", "conv2.weight", "encoder.conv1.bias", "encoder.conv1.weight", "encoder.downsamples.0.residual.0.gamma", "encoder.downsamples.0.residual.2.bias", "encoder.downsamples.0.residual.2.weight", "encoder.downsamples.0.residual.3.gamma", "encoder.downsamples.0.residual.6.bias", "encoder.downsamples.0.residual.6.weight", "encoder.downsamples.1.residual.0.gamma", "encoder.downsamples.1.residual.2.bias", "encoder.downsamples.1.residual.2.weight", "encoder.downsamples.1.residual.3.gamma", "encoder.downsamples.1.residual.6.bias", "encoder.downsamples.1.residual.6.weight", "encoder.downsamples.10.residual.0.gamma", "encoder.downsamples.10.residual.2.bias", "encoder.downsamples.10.residual.2.weight", "encoder.downsamples.10.residual.3.gamma", "encoder.downsamples.10.residual.6.bias", "encoder.downsamples.10.residual.6.weight", "encoder.downsamples.2.resample.1.bias", "encoder.downsamples.2.resample.1.weight", "encoder.downsamples.3.residual.0.gamma", "encoder.downsamples.3.residual.2.bias", "encoder.downsamples.3.residual.2.weight", "encoder.downsamples.3.residual.3.gamma", "encoder.downsamples.3.residual.6.bias", "encoder.downsamples.3.residual.6.weight", "encoder.downsamples.3.shortcut.bias", "encoder.downsamples.3.shortcut.weight", "encoder.downsamples.4.residual.0.gamma", "encoder.downsamples.4.residual.2.bias", "encoder.downsamples.4.residual.2.weight", "encoder.downsamples.4.residual.3.gamma", "encoder.downsamples.4.residual.6.bias", "encoder.downsamples.4.residual.6.weight", "encoder.downsamples.5.resample.1.bias", "encoder.downsamples.5.resample.1.weight", "encoder.downsamples.5.time_conv.bias", "encoder.downsamples.5.time_conv.weight", "encoder.downsamples.6.residual.0.gamma", "encoder.downsamples.6.residual.2.bias", "encoder.downsamples.6.residual.2.weight", "encoder.downsamples.6.residual.3.gamma", "encoder.downsamples.6.residual.6.bias", "encoder.downsamples.6.residual.6.weight", "encoder.downsamples.6.shortcut.bias", "encoder.downsamples.6.shortcut.weight", "encoder.downsamples.7.residual.0.gamma", "encoder.downsamples.7.residual.2.bias", "encoder.downsamples.7.residual.2.weight", "encoder.downsamples.7.residual.3.gamma", "encoder.downsamples.7.residual.6.bias", "encoder.downsamples.7.residual.6.weight", "encoder.downsamples.8.resample.1.bias", "encoder.downsamples.8.resample.1.weight", "encoder.downsamples.8.time_conv.bias", "encoder.downsamples.8.time_conv.weight", "encoder.downsamples.9.residual.0.gamma", "encoder.downsamples.9.residual.2.bias", "encoder.downsamples.9.residual.2.weight", "encoder.downsamples.9.residual.3.gamma", "encoder.downsamples.9.residual.6.bias", "encoder.downsamples.9.residual.6.weight", "encoder.head.0.gamma", "encoder.head.2.bias", "encoder.head.2.weight", "encoder.middle.0.residual.0.gamma", "encoder.middle.0.residual.2.bias", "encoder.middle.0.residual.2.weight", "encoder.middle.0.residual.3.gamma", "encoder.middle.0.residual.6.bias", "encoder.middle.0.residual.6.weight", "encoder.middle.1.norm.gamma", "encoder.middle.1.proj.bias", "encoder.middle.1.proj.weight", "encoder.middle.1.to_qkv.bias", "encoder.middle.1.to_qkv.weight", "encoder.middle.2.residual.0.gamma", "encoder.middle.2.residual.2.bias", "encoder.middle.2.residual.2.weight", "encoder.middle.2.residual.3.gamma", "encoder.middle.2.residual.6.bias", "encoder.middle.2.residual.6.weight", "decoder.conv1.bias", "decoder.conv1.weight", "decoder.head.0.gamma", "decoder.head.2.bias", "decoder.head.2.weight", "decoder.middle.0.residual.0.gamma", "decoder.middle.0.residual.2.bias", "decoder.middle.0.residual.2.weight", "decoder.middle.0.residual.3.gamma", "decoder.middle.0.residual.6.bias", "decoder.middle.0.residual.6.weight", "decoder.middle.1.norm.gamma", "decoder.middle.1.proj.bias", "decoder.middle.1.proj.weight", "decoder.middle.1.to_qkv.bias", "decoder.middle.1.to_qkv.weight", "decoder.middle.2.residual.0.gamma", "decoder.middle.2.residual.2.bias", "decoder.middle.2.residual.2.weight", "decoder.middle.2.residual.3.gamma", "decoder.middle.2.residual.6.bias", "decoder.middle.2.residual.6.weight", "decoder.upsamples.0.residual.0.gamma", "decoder.upsamples.0.residual.2.bias", "decoder.upsamples.0.residual.2.weight", "decoder.upsamples.0.residual.3.gamma", "decoder.upsamples.0.residual.6.bias", "decoder.upsamples.0.residual.6.weight", "decoder.upsamples.1.residual.0.gamma", "decoder.upsamples.1.residual.2.bias", "decoder.upsamples.1.residual.2.weight", "decoder.upsamples.1.residual.3.gamma", "decoder.upsamples.1.residual.6.bias", "decoder.upsamples.1.residual.6.weight", "decoder.upsamples.10.residual.0.gamma", "decoder.upsamples.10.residual.2.bias", "decoder.upsamples.10.residual.2.weight", "decoder.upsamples.10.residual.3.gamma", "decoder.upsamples.10.residual.6.bias", "decoder.upsamples.10.residual.6.weight", "decoder.upsamples.11.resample.1.bias", "decoder.upsamples.11.resample.1.weight", "decoder.upsamples.12.residual.0.gamma", "decoder.upsamples.12.residual.2.bias", "decoder.upsamples.12.residual.2.weight", "decoder.upsamples.12.residual.3.gamma", "decoder.upsamples.12.residual.6.bias", "decoder.upsamples.12.residual.6.weight", "decoder.upsamples.13.residual.0.gamma", "decoder.upsamples.13.residual.2.bias", "decoder.upsamples.13.residual.2.weight", "decoder.upsamples.13.residual.3.gamma", "decoder.upsamples.13.residual.6.bias", "decoder.upsamples.13.residual.6.weight", "decoder.upsamples.14.residual.0.gamma", "decoder.upsamples.14.residual.2.bias", "decoder.upsamples.14.residual.2.weight", "decoder.upsamples.14.residual.3.gamma", "decoder.upsamples.14.residual.6.bias", "decoder.upsamples.14.residual.6.weight", "decoder.upsamples.2.residual.0.gamma", "decoder.upsamples.2.residual.2.bias", "decoder.upsamples.2.residual.2.weight", "decoder.upsamples.2.residual.3.gamma", "decoder.upsamples.2.residual.6.bias", "decoder.upsamples.2.residual.6.weight", "decoder.upsamples.3.resample.1.bias", "decoder.upsamples.3.resample.1.weight", "decoder.upsamples.3.time_conv.bias", "decoder.upsamples.3.time_conv.weight", "decoder.upsamples.4.residual.0.gamma", "decoder.upsamples.4.residual.2.bias", "decoder.upsamples.4.residual.2.weight", "decoder.upsamples.4.residual.3.gamma", "decoder.upsamples.4.residual.6.bias", "decoder.upsamples.4.residual.6.weight", "decoder.upsamples.4.shortcut.bias", "decoder.upsamples.4.shortcut.weight", "decoder.upsamples.5.residual.0.gamma", "decoder.upsamples.5.residual.2.bias", "decoder.upsamples.5.residual.2.weight", "decoder.upsamples.5.residual.3.gamma", "decoder.upsamples.5.residual.6.bias", "decoder.upsamples.5.residual.6.weight", "decoder.upsamples.6.residual.0.gamma", "decoder.upsamples.6.residual.2.bias", "decoder.upsamples.6.residual.2.weight", "decoder.upsamples.6.residual.3.gamma", "decoder.upsamples.6.residual.6.bias", "decoder.upsamples.6.residual.6.weight", "decoder.upsamples.7.resample.1.bias", "decoder.upsamples.7.resample.1.weight", "decoder.upsamples.7.time_conv.bias", "decoder.upsamples.7.time_conv.weight", "decoder.upsamples.8.residual.0.gamma", "decoder.upsamples.8.residual.2.bias", "decoder.upsamples.8.residual.2.weight", "decoder.upsamples.8.residual.3.gamma", "decoder.upsamples.8.residual.6.bias", "decoder.upsamples.8.residual.6.weight", "decoder.upsamples.9.residual.0.gamma", "decoder.upsamples.9.residual.2.bias", "decoder.upsamples.9.residual.2.weight", "decoder.upsamples.9.residual.3.gamma", "decoder.upsamples.9.residual.6.bias", "decoder.upsamples.9.residual.6.weight".
2025-09-27 22:42:41
[ERROR] 命令执行失败，返回码: 1
2025-09-27 22:42:41 预缓存 1/2 完成!
2025-09-27 22:42:41 开始执行预缓存2/2...
2025-09-27 22:42:41 I:\AIProject\Loratrain\LoRAMaster.venv\Scripts\python.exe I:\AIProject\Loratrain\LoRAMaster\musubi-tuner\src\musubi_tuner\qwen_image_cache_text_encoder_outputs.py --dataset_config I:\AIProject\Loratrain\LoRAMaster\samples\toml\qwen_image_saiboguofengji.toml --text_encoder E:\ComfyUI_windows_portable\ComfyUI\models\clip\qwen_2.5_vl_7b.safetensors --batch_size 1 --fp8_vl
2025-09-27 22:42:47 INFO:main:Load dataset config from I:\AIProject\Loratrain\LoRAMaster\samples\toml\qwen_image_saiboguofengji.toml
2025-09-27 22:42:47 INFO:musubi_tuner.dataset.image_video_dataset:glob images in I:\LoRA_train\saiboguofengji\images
2025-09-27 22:42:47 INFO:musubi_tuner.dataset.image_video_dataset:found 35 images
2025-09-27 22:42:47 INFO:musubi_tuner.dataset.config_utils:[Dataset 0]
2025-09-27 22:42:47 is_image_dataset: True
2025-09-27 22:42:47 resolution: (1024, 1024)
2025-09-27 22:42:47 batch_size: 1
2025-09-27 22:42:47 num_repeats: 2
2025-09-27 22:42:47 caption_extension: ".txt"
2025-09-27 22:42:47 enable_bucket: False
2025-09-27 22:42:47 bucket_no_upscale: False
2025-09-27 22:42:47 cache_directory: "I:\LoRA_train\saiboguofengji\qwen_image_cache"
2025-09-27 22:42:47 debug_dataset: False
2025-09-27 22:42:47 image_directory: "I:\LoRA_train\saiboguofengji\images"
2025-09-27 22:42:47 image_jsonl_file: "None"
2025-09-27 22:42:47 fp_latent_window_size: 9
2025-09-27 22:42:47 fp_1f_clean_indices: None
2025-09-27 22:42:47 fp_1f_target_index: None
2025-09-27 22:42:47 fp_1f_no_post: False
2025-09-27 22:42:47 flux_kontext_no_resize_control: False
2025-09-27 22:42:47 qwen_image_edit_no_resize_control: False
2025-09-27 22:42:47 qwen_image_edit_control_resolution: None
2025-09-27 22:42:47
2025-09-27 22:42:47
2025-09-27 22:42:47 INFO:main:Loading Qwen2.5-VL: E:\ComfyUI_windows_portable\ComfyUI\models\clip\qwen_2.5_vl_7b.safetensors
2025-09-27 22:42:47 INFO:musubi_tuner.qwen_image.qwen_image_utils:Loading state dict from E:\ComfyUI_windows_portable\ComfyUI\models\clip\qwen_2.5_vl_7b.safetensors
2025-09-27 22:42:54 INFO:musubi_tuner.qwen_image.qwen_image_utils:Loaded Qwen2.5-VL:
2025-09-27 22:42:54 INFO:musubi_tuner.qwen_image.qwen_image_utils:prepare Qwen2.5-VL for fp8: set to torch.float8_e4m3fn from torch.bfloat16
2025-09-27 22:42:54 INFO:musubi_tuner.qwen_image.qwen_image_utils:Loading tokenizer from Qwen/Qwen-Image
2025-09-27 22:42:57 INFO:main:Encoding with Qwen2.5-VL
2025-09-27 22:42:57 INFO:musubi_tuner.cache_text_encoder_outputs:Encoding dataset [0]
2025-09-27 22:42:57 Trying to import sageattention
2025-09-27 22:42:57 Failed to import sageattention
2025-09-27 22:42:57
2025-09-27 22:42:58 0it [00:00, ?it/s]
2025-09-27 22:42:58 1it [00:00, 1.02it/s]
2025-09-27 22:42:58 3it [00:01, 3.27it/s]
2025-09-27 22:42:58 5it [00:01, 5.45it/s]
2025-09-27 22:42:58 7it [00:01, 7.45it/s]
2025-09-27 22:42:59 9it [00:01, 9.20it/s]
2025-09-27 22:42:59 11it [00:01, 10.65it/s]
2025-09-27 22:42:59 13it [00:01, 11.75it/s]
2025-09-27 22:42:59 15it [00:01, 12.58it/s]
2025-09-27 22:42:59 17it [00:02, 13.19it/s]
2025-09-27 22:42:59 19it [00:02, 11.89it/s]
2025-09-27 22:42:59 21it [00:02, 12.57it/s]
2025-09-27 22:43:00 23it [00:02, 13.13it/s]
2025-09-27 22:43:00 25it [00:02, 13.35it/s]
2025-09-27 22:43:00 27it [00:02, 13.66it/s]
2025-09-27 22:43:00 29it [00:02, 13.83it/s]
2025-09-27 22:43:00 31it [00:03, 11.57it/s]
2025-09-27 22:43:00 33it [00:03, 12.34it/s]
2025-09-27 22:43:00 35it [00:03, 12.99it/s]
2025-09-27 22:43:00 35it [00:03, 10.06it/s]
2025-09-27 22:43:01 预缓存 2/2 完成!
I:\AIProject\Loratrain\LoRAMaster.venv\Scripts\python.exe I:\AIProject\Loratrain\LoRAMaster\musubi-tuner\src\musubi_tuner\qwen_image_train_network.py
2025-09-27 22:46:11 [INFO] LoRAMaster: 开始运行 Wan LoRA训练命令...
2025-09-27 22:46:11 [INFO] LoRAMaster: I:\AIProject\Loratrain\LoRAMaster.venv\Scripts\python.exe -m accelerate.commands.launch --num_cpu_threads_per_process 1 --mixed_precision bf16 --num_processes 100 --gpu_ids 0 I:\AIProject\Loratrain\LoRAMaster\musubi-tuner\src\musubi_tuner\qwen_image_train_network.py --dit E:\ComfyUI_windows_portable\ComfyUI\models\diffusion_models\qwen_image_bf16.safetensors --dataset_config I:\AIProject\Loratrain\LoRAMaster\samples\toml\qwen_image_saiboguofengji.toml --vae E:\ComfyUI_windows_portable\ComfyUI\models\vae\qwen_image_vae_t.safetensors --text_encoder E:\ComfyUI_windows_portable\ComfyUI\models\clip\qwen_2.5_vl_7b.safetensors --mixed_precision bf16 --optimizer_type adamw8bit --learning_rate 5e-5 --gradient_checkpointing --gradient_accumulation_steps=1 --max_data_loader_n_workers 2 --persistent_data_loader_workers --network_module networks.lora_qwen_image --network_dim 32 --timestep_sampling shift --discrete_flow_shift 2.2 --max_train_epochs 200 --save_every_n_epochs 10 --save_every_n_steps 100 --output_dir ./output --output_name tiquhuawen_v7_x --seed 42 --log_with tensorboard --lr_scheduler constant --sdpa --blocks_to_swap 20 --fp8_base --logging_dir ./logs
2025-09-27 22:46:20 [INFO] LoRAMaster: INFO:musubi_tuner.hv_train_network:Load dataset config from I:\AIProject\Loratrain\LoRAMaster\samples\toml\qwen_image_saiboguofengji.toml
2025-09-27 22:46:20 [INFO] LoRAMaster: INFO:musubi_tuner.dataset.image_video_dataset:glob images in I:\LoRA_train\saiboguofengji\images
2025-09-27 22:46:20 [INFO] LoRAMaster: INFO:musubi_tuner.dataset.image_video_dataset:found 35 images
2025-09-27 22:46:20 [INFO] LoRAMaster: INFO:musubi_tuner.dataset.config_utils:[Dataset 0]
2025-09-27 22:46:20 [INFO] LoRAMaster: is_image_dataset: True
2025-09-27 22:46:20 [INFO] LoRAMaster: resolution: (1024, 1024)
2025-09-27 22:46:20 [INFO] LoRAMaster: batch_size: 1
2025-09-27 22:46:20 [INFO] LoRAMaster: num_repeats: 2
2025-09-27 22:46:20 [INFO] LoRAMaster: caption_extension: ".txt"
2025-09-27 22:46:20 [INFO] LoRAMaster: enable_bucket: False
2025-09-27 22:46:20 [INFO] LoRAMaster: bucket_no_upscale: False
2025-09-27 22:46:20 [INFO] LoRAMaster: cache_directory: "I:\LoRA_train\saiboguofengji\qwen_image_cache"
2025-09-27 22:46:20 [INFO] LoRAMaster: debug_dataset: False
2025-09-27 22:46:20 [INFO] LoRAMaster: image_directory: "I:\LoRA_train\saiboguofengji\images"
2025-09-27 22:46:20 [INFO] LoRAMaster: image_jsonl_file: "None"
2025-09-27 22:46:20 [INFO] LoRAMaster: fp_latent_window_size: 9
2025-09-27 22:46:20 [INFO] LoRAMaster: fp_1f_clean_indices: None
2025-09-27 22:46:20 [INFO] LoRAMaster: fp_1f_target_index: None
2025-09-27 22:46:20 [INFO] LoRAMaster: fp_1f_no_post: False
2025-09-27 22:46:20 [INFO] LoRAMaster: flux_kontext_no_resize_control: False
2025-09-27 22:46:20 [INFO] LoRAMaster: qwen_image_edit_no_resize_control: False
2025-09-27 22:46:20 [INFO] LoRAMaster: qwen_image_edit_control_resolution: None
2025-09-27 22:46:20 [INFO] LoRAMaster:
2025-09-27 22:46:20 [INFO] LoRAMaster:
2025-09-27 22:46:20 [INFO] LoRAMaster: INFO:musubi_tuner.dataset.image_video_dataset:total batches: 0
2025-09-27 22:46:20 [INFO] LoRAMaster: Trying to import sageattention
2025-09-27 22:46:20 [INFO] LoRAMaster: Failed to import sageattention
2025-09-27 22:46:20 [INFO] LoRAMaster: Traceback (most recent call last):
2025-09-27 22:46:20 [INFO] LoRAMaster: File "I:\AIProject\Loratrain\LoRAMaster\musubi-tuner\src\musubi_tuner\qwen_image_train_network.py", line 433, in
2025-09-27 22:46:20 [INFO] LoRAMaster: main()
2025-09-27 22:46:20 [INFO] LoRAMaster: File "I:\AIProject\Loratrain\LoRAMaster\musubi-tuner\src\musubi_tuner\qwen_image_train_network.py", line 429, in main
2025-09-27 22:46:20 [INFO] LoRAMaster: trainer.train(args)
2025-09-27 22:46:20 [INFO] LoRAMaster: File "I:\AIProject\Loratrain\LoRAMaster\musubi-tuner\src\musubi_tuner\hv_train_network.py", line 1667, in train
2025-09-27 22:46:20 [INFO] LoRAMaster: raise ValueError(
2025-09-27 22:46:20 [INFO] LoRAMaster: ValueError: No training items found in the dataset. Please ensure that the latent/Text Encoder cache has been created beforehand. / データセットに学習データがありません。latent/Text Encoderキャッシュを事前に作成したか確認してください
2025-09-27 22:46:20 [INFO] LoRAMaster: Traceback (most recent call last):
2025-09-27 22:46:20 [INFO] LoRAMaster: File "", line 198, in _run_module_as_main
2025-09-27 22:46:20 [INFO] LoRAMaster: File "", line 88, in _run_code
2025-09-27 22:46:20 [INFO] LoRAMaster: File "I:\AIProject\Loratrain\LoRAMaster.venv\Lib\site-packages\accelerate\commands\launch.py", line 1223, in
2025-09-27 22:46:20 [INFO] LoRAMaster: main()
2025-09-27 22:46:20 [INFO] LoRAMaster: File "I:\AIProject\Loratrain\LoRAMaster.venv\Lib\site-packages\accelerate\commands\launch.py", line 1219, in main
2025-09-27 22:46:20 [INFO] LoRAMaster: launch_command(args)
2025-09-27 22:46:20 [INFO] LoRAMaster: File "I:\AIProject\Loratrain\LoRAMaster.venv\Lib\site-packages\accelerate\commands\launch.py", line 1213, in launch_command
2025-09-27 22:46:20 [INFO] LoRAMaster: simple_launcher(args)
2025-09-27 22:46:20 [INFO] LoRAMaster: File "I:\AIProject\Loratrain\LoRAMaster.venv\Lib\site-packages\accelerate\commands\launch.py", line 795, in simple_launcher
2025-09-27 22:46:20 [INFO] LoRAMaster: raise subprocess.CalledProcessError(returncode=process.returncode, cmd=cmd)
2025-09-27 22:46:20 [INFO] LoRAMaster: subprocess.CalledProcessError: Command '['I:\AIProject\Loratrain\LoRAMaster\.venv\Scripts\python.exe', 'I:\AIProject\Loratrain\LoRAMaster\musubi-tuner\src\musubi_tuner\qwen_image_train_network.py', '--dit', 'E:\ComfyUI_windows_portable\ComfyUI\models\diffusion_models\qwen_image_bf16.safetensors', '--dataset_config', 'I:\AIProject\Loratrain\LoRAMaster\samples\toml\qwen_image_saiboguofengji.toml', '--vae', 'E:\ComfyUI_windows_portable\ComfyUI\models\vae\qwen_image_vae_t.safetensors', '--text_encoder', 'E:\ComfyUI_windows_portable\ComfyUI\models\clip\qwen_2.5_vl_7b.safetensors', '--mixed_precision', 'bf16', '--optimizer_type', 'adamw8bit', '--learning_rate', '5e-5', '--gradient_checkpointing', '--gradient_accumulation_steps=1', '--max_data_loader_n_workers', '2', '--persistent_data_loader_workers', '--network_module', 'networks.lora_qwen_image', '--network_dim', '32', '--timestep_sampling', 'shift', '--discrete_flow_shift', '2.2', '--max_train_epochs', '200', '--save_every_n_epochs', '10', '--save_every_n_steps', '100', '--output_dir', './output', '--output_name', 'tiquhuawen_v7_x', '--seed', '42', '--log_with', 'tensorboard', '--lr_scheduler', 'constant', '--sdpa', '--blocks_to_swap', '20', '--fp8_base', '--logging_dir', './logs']' returned non-zero exit status 1.
2025-09-27 22:46:21 [INFO] LoRAMaster:
[ERROR] 命令执行失败，返回码: 1
`
虽然生成了缓存，但是训练的时候就报错，缓存的时候就提示vae报错，vae没问题啊，都能出图。而且重新下载的也是不行。
`2025-09-27 22:42:32 开始执行预缓存...
2025-09-27 22:42:32 开始执行预缓存 1/2 ...
2025-09-27 22:42:32 I:\AIProject\Loratrain\LoRAMaster.venv\Scripts\python.exe I:\AIProject\Loratrain\LoRAMaster\musubi-tuner\src\musubi_tuner\qwen_image_cache_latents.py --dataset_config I:\AIProject\Loratrain\LoRAMaster\samples\toml\qwen_image_saiboguofengji.toml --vae E:\ComfyUI_windows_portable\ComfyUI\models\vae\qwen_image_vae_t.safetensors
2025-09-27 22:42:39 INFO:main:Load dataset config from I:\AIProject\Loratrain\LoRAMaster\samples\toml\qwen_image_saiboguofengji.toml
2025-09-27 22:42:39 INFO:musubi_tuner.dataset.image_video_dataset:glob images in I:\LoRA_train\saiboguofengji\images
2025-09-27 22:42:39 INFO:musubi_tuner.dataset.image_video_dataset:found 35 images
2025-09-27 22:42:39 INFO:musubi_tuner.dataset.config_utils:[Dataset 0]
2025-09-27 22:42:39 is_image_dataset: True
2025-09-27 22:42:39 resolution: (1024, 1024)
2025-09-27 22:42:39 batch_size: 1
2025-09-27 22:42:39 num_repeats: 2
2025-09-27 22:42:39 caption_extension: ".txt"
2025-09-27 22:42:39 enable_bucket: False
2025-09-27 22:42:39 bucket_no_upscale: False
2025-09-27 22:42:39 cache_directory: "I:\LoRA_train\saiboguofengji\qwen_image_cache"
2025-09-27 22:42:39 debug_dataset: False
2025-09-27 22:42:39 image_directory: "I:\LoRA_train\saiboguofengji\images"
2025-09-27 22:42:39 image_jsonl_file: "None"
2025-09-27 22:42:39 fp_latent_window_size: 9
2025-09-27 22:42:39 fp_1f_clean_indices: None
2025-09-27 22:42:39 fp_1f_target_index: None
2025-09-27 22:42:39 fp_1f_no_post: False
2025-09-27 22:42:39 flux_kontext_no_resize_control: False
2025-09-27 22:42:39 qwen_image_edit_no_resize_control: False
2025-09-27 22:42:39 qwen_image_edit_control_resolution: None
2025-09-27 22:42:39
2025-09-27 22:42:39
2025-09-27 22:42:39 INFO:main:Loading VAE model from E:\ComfyUI_windows_portable\ComfyUI\models\vae\qwen_image_vae_t.safetensors
2025-09-27 22:42:39 INFO:musubi_tuner.qwen_image.qwen_image_utils:Initializing VAE
2025-09-27 22:42:40 INFO:musubi_tuner.qwen_image.qwen_image_utils:Loading VAE from E:\ComfyUI_windows_portable\ComfyUI\models\vae\qwen_image_vae_t.safetensors
2025-09-27 22:42:40 Trying to import sageattention
2025-09-27 22:42:40 Failed to import sageattention
2025-09-27 22:42:40 Traceback (most recent call last):
2025-09-27 22:42:40 File "I:\AIProject\Loratrain\LoRAMaster\musubi-tuner\src\musubi_tuner\qwen_image_cache_latents.py", line 123, in
2025-09-27 22:42:40 main()
2025-09-27 22:42:40 File "I:\AIProject\Loratrain\LoRAMaster\musubi-tuner\src\musubi_tuner\qwen_image_cache_latents.py", line 111, in main
2025-09-27 22:42:40 vae = qwen_image_utils.load_vae(args.vae, device=device, disable_mmap=True)
2025-09-27 22:42:40 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
2025-09-27 22:42:40 File "I:\AIProject\Loratrain\LoRAMaster\musubi-tuner\src\musubi_tuner\qwen_image\qwen_image_utils.py", line 548, in load_vae
2025-09-27 22:42:40 info = vae.load_state_dict(state_dict, strict=True, assign=True)
2025-09-27 22:42:40 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
2025-09-27 22:42:40 File "I:\AIProject\Loratrain\LoRAMaster.venv\Lib\site-packages\torch\nn\modules\module.py", line 2624, in load_state_dict
2025-09-27 22:42:40 raise RuntimeError(
2025-09-27 22:42:40 RuntimeError: Error(s) in loading state_dict for AutoencoderKLQwenImage:
2025-09-27 22:42:40 Missing key(s) in state_dict: "encoder.conv_in.weight", "encoder.conv_in.bias", "encoder.down_blocks.0.norm1.gamma", "encoder.down_blocks.0.conv1.weight", "encoder.down_blocks.0.conv1.bias", "encoder.down_blocks.0.norm2.gamma", "encoder.down_blocks.0.conv2.weight", "encoder.down_blocks.0.conv2.bias", "encoder.down_blocks.1.norm1.gamma", "encoder.down_blocks.1.conv1.weight", "encoder.down_blocks.1.conv1.bias", "encoder.down_blocks.1.norm2.gamma", "encoder.down_blocks.1.conv2.weight", "encoder.down_blocks.1.conv2.bias", "encoder.down_blocks.2.resample.1.weight", "encoder.down_blocks.2.resample.1.bias", "encoder.down_blocks.3.norm1.gamma", "encoder.down_blocks.3.conv1.weight", "encoder.down_blocks.3.conv1.bias", "encoder.down_blocks.3.norm2.gamma", "encoder.down_blocks.3.conv2.weight", "encoder.down_blocks.3.conv2.bias", "encoder.down_blocks.3.conv_shortcut.weight", "encoder.down_blocks.3.conv_shortcut.bias", "encoder.down_blocks.4.norm1.gamma", "encoder.down_blocks.4.conv1.weight", "encoder.down_blocks.4.conv1.bias", "encoder.down_blocks.4.norm2.gamma", "encoder.down_blocks.4.conv2.weight", "encoder.down_blocks.4.conv2.bias", "encoder.down_blocks.5.resample.1.weight", "encoder.down_blocks.5.resample.1.bias", "encoder.down_blocks.5.time_conv.weight", "encoder.down_blocks.5.time_conv.bias", "encoder.down_blocks.6.norm1.gamma", "encoder.down_blocks.6.conv1.weight", "encoder.down_blocks.6.conv1.bias", "encoder.down_blocks.6.norm2.gamma", "encoder.down_blocks.6.conv2.weight", "encoder.down_blocks.6.conv2.bias", "encoder.down_blocks.6.conv_shortcut.weight", "encoder.down_blocks.6.conv_shortcut.bias", "encoder.down_blocks.7.norm1.gamma", "encoder.down_blocks.7.conv1.weight", "encoder.down_blocks.7.conv1.bias", "encoder.down_blocks.7.norm2.gamma", "encoder.down_blocks.7.conv2.weight", "encoder.down_blocks.7.conv2.bias", "encoder.down_blocks.8.resample.1.weight", "encoder.down_blocks.8.resample.1.bias", "encoder.down_blocks.8.time_conv.weight", "encoder.down_blocks.8.time_conv.bias", "encoder.down_blocks.9.norm1.gamma", "encoder.down_blocks.9.conv1.weight", "encoder.down_blocks.9.conv1.bias", "encoder.down_blocks.9.norm2.gamma", "encoder.down_blocks.9.conv2.weight", "encoder.down_blocks.9.conv2.bias", "encoder.down_blocks.10.norm1.gamma", "encoder.down_blocks.10.conv1.weight", "encoder.down_blocks.10.conv1.bias", "encoder.down_blocks.10.norm2.gamma", "encoder.down_blocks.10.conv2.weight", "encoder.down_blocks.10.conv2.bias", "encoder.mid_block.attentions.0.norm.gamma", "encoder.mid_block.attentions.0.to_qkv.weight", "encoder.mid_block.attentions.0.to_qkv.bias", "encoder.mid_block.attentions.0.proj.weight", "encoder.mid_block.attentions.0.proj.bias", "encoder.mid_block.resnets.0.norm1.gamma", "encoder.mid_block.resnets.0.conv1.weight", "encoder.mid_block.resnets.0.conv1.bias", "encoder.mid_block.resnets.0.norm2.gamma", "encoder.mid_block.resnets.0.conv2.weight", "encoder.mid_block.resnets.0.conv2.bias", "encoder.mid_block.resnets.1.norm1.gamma", "encoder.mid_block.resnets.1.conv1.weight", "encoder.mid_block.resnets.1.conv1.bias", "encoder.mid_block.resnets.1.norm2.gamma", "encoder.mid_block.resnets.1.conv2.weight", "encoder.mid_block.resnets.1.conv2.bias", "encoder.norm_out.gamma", "encoder.conv_out.weight", "encoder.conv_out.bias", "quant_conv.weight", "quant_conv.bias", "post_quant_conv.weight", "post_quant_conv.bias", "decoder.conv_in.weight", "decoder.conv_in.bias", "decoder.mid_block.attentions.0.norm.gamma", "decoder.mid_block.attentions.0.to_qkv.weight", "decoder.mid_block.attentions.0.to_qkv.bias", "decoder.mid_block.attentions.0.proj.weight", "decoder.mid_block.attentions.0.proj.bias", "decoder.mid_block.resnets.0.norm1.gamma", "decoder.mid_block.resnets.0.conv1.weight", "decoder.mid_block.resnets.0.conv1.bias", "decoder.mid_block.resnets.0.norm2.gamma", "decoder.mid_block.resnets.0.conv2.weight", "decoder.mid_block.resnets.0.conv2.bias", "decoder.mid_block.resnets.1.norm1.gamma", "decoder.mid_block.resnets.1.conv1.weight", "decoder.mid_block.resnets.1.conv1.bias", "decoder.mid_block.resnets.1.norm2.gamma", "decoder.mid_block.resnets.1.conv2.weight", "decoder.mid_block.resnets.1.conv2.bias", "decoder.up_blocks.0.resnets.0.norm1.gamma", "decoder.up_blocks.0.resnets.0.conv1.weight", "decoder.up_blocks.0.resnets.0.conv1.bias", "decoder.up_blocks.0.resnets.0.norm2.gamma", "decoder.up_blocks.0.resnets.0.conv2.weight", "decoder.up_blocks.0.resnets.0.conv2.bias", "decoder.up_blocks.0.resnets.1.norm1.gamma", "decoder.up_blocks.0.resnets.1.conv1.weight", "decoder.up_blocks.0.resnets.1.conv1.bias", "decoder.up_blocks.0.resnets.1.norm2.gamma", "decoder.up_blocks.0.resnets.1.conv2.weight", "decoder.up_blocks.0.resnets.1.conv2.bias", "decoder.up_blocks.0.resnets.2.norm1.gamma", "decoder.up_blocks.0.resnets.2.conv1.weight", "decoder.up_blocks.0.resnets.2.conv1.bias", "decoder.up_blocks.0.resnets.2.norm2.gamma", "decoder.up_blocks.0.resnets.2.conv2.weight", "decoder.up_blocks.0.resnets.2.conv2.bias", "decoder.up_blocks.0.upsamplers.0.resample.1.weight", "decoder.up_blocks.0.upsamplers.0.resample.1.bias", "decoder.up_blocks.0.upsamplers.0.time_conv.weight", "decoder.up_blocks.0.upsamplers.0.time_conv.bias", "decoder.up_blocks.1.resnets.0.norm1.gamma", "decoder.up_blocks.1.resnets.0.conv1.weight", "decoder.up_blocks.1.resnets.0.conv1.bias", "decoder.up_blocks.1.resnets.0.norm2.gamma", "decoder.up_blocks.1.resnets.0.conv2.weight", "decoder.up_blocks.1.resnets.0.conv2.bias", "decoder.up_blocks.1.resnets.0.conv_shortcut.weight", "decoder.up_blocks.1.resnets.0.conv_shortcut.bias", "decoder.up_blocks.1.resnets.1.norm1.gamma", "decoder.up_blocks.1.resnets.1.conv1.weight", "decoder.up_blocks.1.resnets.1.conv1.bias", "decoder.up_blocks.1.resnets.1.norm2.gamma", "decoder.up_blocks.1.resnets.1.conv2.weight", "decoder.up_blocks.1.resnets.1.conv2.bias", "decoder.up_blocks.1.resnets.2.norm1.gamma", "decoder.up_blocks.1.resnets.2.conv1.weight", "decoder.up_blocks.1.resnets.2.conv1.bias", "decoder.up_blocks.1.resnets.2.norm2.gamma", "decoder.up_blocks.1.resnets.2.conv2.weight", "decoder.up_blocks.1.resnets.2.conv2.bias", "decoder.up_blocks.1.upsamplers.0.resample.1.weight", "decoder.up_blocks.1.upsamplers.0.resample.1.bias", "decoder.up_blocks.1.upsamplers.0.time_conv.weight", "decoder.up_blocks.1.upsamplers.0.time_conv.bias", "decoder.up_blocks.2.resnets.0.norm1.gamma", "decoder.up_blocks.2.resnets.0.conv1.weight", "decoder.up_blocks.2.resnets.0.conv1.bias", "decoder.up_blocks.2.resnets.0.norm2.gamma", "decoder.up_blocks.2.resnets.0.conv2.weight", "decoder.up_blocks.2.resnets.0.conv2.bias", "decoder.up_blocks.2.resnets.1.norm1.gamma", "decoder.up_blocks.2.resnets.1.conv1.weight", "decoder.up_blocks.2.resnets.1.conv1.bias", "decoder.up_blocks.2.resnets.1.norm2.gamma", "decoder.up_blocks.2.resnets.1.conv2.weight", "decoder.up_blocks.2.resnets.1.conv2.bias", "decoder.up_blocks.2.resnets.2.norm1.gamma", "decoder.up_blocks.2.resnets.2.conv1.weight", "decoder.up_blocks.2.resnets.2.conv1.bias", "decoder.up_blocks.2.resnets.2.norm2.gamma", "decoder.up_blocks.2.resnets.2.conv2.weight", "decoder.up_blocks.2.resnets.2.conv2.bias", "decoder.up_blocks.2.upsamplers.0.resample.1.weight", "decoder.up_blocks.2.upsamplers.0.resample.1.bias", "decoder.up_blocks.3.resnets.0.norm1.gamma", "decoder.up_blocks.3.resnets.0.conv1.weight", "decoder.up_blocks.3.resnets.0.conv1.bias", "decoder.up_blocks.3.resnets.0.norm2.gamma", "decoder.up_blocks.3.resnets.0.conv2.weight", "decoder.up_blocks.3.resnets.0.conv2.bias", "decoder.up_blocks.3.resnets.1.norm1.gamma", "decoder.up_blocks.3.resnets.1.conv1.weight", "decoder.up_blocks.3.resnets.1.conv1.bias", "decoder.up_blocks.3.resnets.1.norm2.gamma", "decoder.up_blocks.3.resnets.1.conv2.weight", "decoder.up_blocks.3.resnets.1.conv2.bias", "decoder.up_blocks.3.resnets.2.norm1.gamma", "decoder.up_blocks.3.resnets.2.conv1.weight", "decoder.up_blocks.3.resnets.2.conv1.bias", "decoder.up_blocks.3.resnets.2.norm2.gamma", "decoder.up_blocks.3.resnets.2.conv2.weight", "decoder.up_blocks.3.resnets.2.conv2.bias", "decoder.norm_out.gamma", "decoder.conv_out.weight", "decoder.conv_out.bias".
2025-09-27 22:42:40 Unexpected key(s) in state_dict: "conv1.bias", "conv1.weight", "conv2.bias", "conv2.weight", "encoder.conv1.bias", "encoder.conv1.weight", "encoder.downsamples.0.residual.0.gamma", "encoder.downsamples.0.residual.2.bias", "encoder.downsamples.0.residual.2.weight", "encoder.downsamples.0.residual.3.gamma", "encoder.downsamples.0.residual.6.bias", "encoder.downsamples.0.residual.6.weight", "encoder.downsamples.1.residual.0.gamma", "encoder.downsamples.1.residual.2.bias", "encoder.downsamples.1.residual.2.weight", "encoder.downsamples.1.residual.3.gamma", "encoder.downsamples.1.residual.6.bias", "encoder.downsamples.1.residual.6.weight", "encoder.downsamples.10.residual.0.gamma", "encoder.downsamples.10.residual.2.bias", "encoder.downsamples.10.residual.2.weight", "encoder.downsamples.10.residual.3.gamma", "encoder.downsamples.10.residual.6.bias", "encoder.downsamples.10.residual.6.weight", "encoder.downsamples.2.resample.1.bias", "encoder.downsamples.2.resample.1.weight", "encoder.downsamples.3.residual.0.gamma", "encoder.downsamples.3.residual.2.bias", "encoder.downsamples.3.residual.2.weight", "encoder.downsamples.3.residual.3.gamma", "encoder.downsamples.3.residual.6.bias", "encoder.downsamples.3.residual.6.weight", "encoder.downsamples.3.shortcut.bias", "encoder.downsamples.3.shortcut.weight", "encoder.downsamples.4.residual.0.gamma", "encoder.downsamples.4.residual.2.bias", "encoder.downsamples.4.residual.2.weight", "encoder.downsamples.4.residual.3.gamma", "encoder.downsamples.4.residual.6.bias", "encoder.downsamples.4.residual.6.weight", "encoder.downsamples.5.resample.1.bias", "encoder.downsamples.5.resample.1.weight", "encoder.downsamples.5.time_conv.bias", "encoder.downsamples.5.time_conv.weight", "encoder.downsamples.6.residual.0.gamma", "encoder.downsamples.6.residual.2.bias", "encoder.downsamples.6.residual.2.weight", "encoder.downsamples.6.residual.3.gamma", "encoder.downsamples.6.residual.6.bias", "encoder.downsamples.6.residual.6.weight", "encoder.downsamples.6.shortcut.bias", "encoder.downsamples.6.shortcut.weight", "encoder.downsamples.7.residual.0.gamma", "encoder.downsamples.7.residual.2.bias", "encoder.downsamples.7.residual.2.weight", "encoder.downsamples.7.residual.3.gamma", "encoder.downsamples.7.residual.6.bias", "encoder.downsamples.7.residual.6.weight", "encoder.downsamples.8.resample.1.bias", "encoder.downsamples.8.resample.1.weight", "encoder.downsamples.8.time_conv.bias", "encoder.downsamples.8.time_conv.weight", "encoder.downsamples.9.residual.0.gamma", "encoder.downsamples.9.residual.2.bias", "encoder.downsamples.9.residual.2.weight", "encoder.downsamples.9.residual.3.gamma", "encoder.downsamples.9.residual.6.bias", "encoder.downsamples.9.residual.6.weight", "encoder.head.0.gamma", "encoder.head.2.bias", "encoder.head.2.weight", "encoder.middle.0.residual.0.gamma", "encoder.middle.0.residual.2.bias", "encoder.middle.0.residual.2.weight", "encoder.middle.0.residual.3.gamma", "encoder.middle.0.residual.6.bias", "encoder.middle.0.residual.6.weight", "encoder.middle.1.norm.gamma", "encoder.middle.1.proj.bias", "encoder.middle.1.proj.weight", "encoder.middle.1.to_qkv.bias", "encoder.middle.1.to_qkv.weight", "encoder.middle.2.residual.0.gamma", "encoder.middle.2.residual.2.bias", "encoder.middle.2.residual.2.weight", "encoder.middle.2.residual.3.gamma", "encoder.middle.2.residual.6.bias", "encoder.middle.2.residual.6.weight", "decoder.conv1.bias", "decoder.conv1.weight", "decoder.head.0.gamma", "decoder.head.2.bias", "decoder.head.2.weight", "decoder.middle.0.residual.0.gamma", "decoder.middle.0.residual.2.bias", "decoder.middle.0.residual.2.weight", "decoder.middle.0.residual.3.gamma", "decoder.middle.0.residual.6.bias", "decoder.middle.0.residual.6.weight", "decoder.middle.1.norm.gamma", "decoder.middle.1.proj.bias", "decoder.middle.1.proj.weight", "decoder.middle.1.to_qkv.bias", "decoder.middle.1.to_qkv.weight", "decoder.middle.2.residual.0.gamma", "decoder.middle.2.residual.2.bias", "decoder.middle.2.residual.2.weight", "decoder.middle.2.residual.3.gamma", "decoder.middle.2.residual.6.bias", "decoder.middle.2.residual.6.weight", "decoder.upsamples.0.residual.0.gamma", "decoder.upsamples.0.residual.2.bias", "decoder.upsamples.0.residual.2.weight", "decoder.upsamples.0.residual.3.gamma", "decoder.upsamples.0.residual.6.bias", "decoder.upsamples.0.residual.6.weight", "decoder.upsamples.1.residual.0.gamma", "decoder.upsamples.1.residual.2.bias", "decoder.upsamples.1.residual.2.weight", "decoder.upsamples.1.residual.3.gamma", "decoder.upsamples.1.residual.6.bias", "decoder.upsamples.1.residual.6.weight", "decoder.upsamples.10.residual.0.gamma", "decoder.upsamples.10.residual.2.bias", "decoder.upsamples.10.residual.2.weight", "decoder.upsamples.10.residual.3.gamma", "decoder.upsamples.10.residual.6.bias", "decoder.upsamples.10.residual.6.weight", "decoder.upsamples.11.resample.1.bias", "decoder.upsamples.11.resample.1.weight", "decoder.upsamples.12.residual.0.gamma", "decoder.upsamples.12.residual.2.bias", "decoder.upsamples.12.residual.2.weight", "decoder.upsamples.12.residual.3.gamma", "decoder.upsamples.12.residual.6.bias", "decoder.upsamples.12.residual.6.weight", "decoder.upsamples.13.residual.0.gamma", "decoder.upsamples.13.residual.2.bias", "decoder.upsamples.13.residual.2.weight", "decoder.upsamples.13.residual.3.gamma", "decoder.upsamples.13.residual.6.bias", "decoder.upsamples.13.residual.6.weight", "decoder.upsamples.14.residual.0.gamma", "decoder.upsamples.14.residual.2.bias", "decoder.upsamples.14.residual.2.weight", "decoder.upsamples.14.residual.3.gamma", "decoder.upsamples.14.residual.6.bias", "decoder.upsamples.14.residual.6.weight", "decoder.upsamples.2.residual.0.gamma", "decoder.upsamples.2.residual.2.bias", "decoder.upsamples.2.residual.2.weight", "decoder.upsamples.2.residual.3.gamma", "decoder.upsamples.2.residual.6.bias", "decoder.upsamples.2.residual.6.weight", "decoder.upsamples.3.resample.1.bias", "decoder.upsamples.3.resample.1.weight", "decoder.upsamples.3.time_conv.bias", "decoder.upsamples.3.time_conv.weight", "decoder.upsamples.4.residual.0.gamma", "decoder.upsamples.4.residual.2.bias", "decoder.upsamples.4.residual.2.weight", "decoder.upsamples.4.residual.3.gamma", "decoder.upsamples.4.residual.6.bias", "decoder.upsamples.4.residual.6.weight", "decoder.upsamples.4.shortcut.bias", "decoder.upsamples.4.shortcut.weight", "decoder.upsamples.5.residual.0.gamma", "decoder.upsamples.5.residual.2.bias", "decoder.upsamples.5.residual.2.weight", "decoder.upsamples.5.residual.3.gamma", "decoder.upsamples.5.residual.6.bias", "decoder.upsamples.5.residual.6.weight", "decoder.upsamples.6.residual.0.gamma", "decoder.upsamples.6.residual.2.bias", "decoder.upsamples.6.residual.2.weight", "decoder.upsamples.6.residual.3.gamma", "decoder.upsamples.6.residual.6.bias", "decoder.upsamples.6.residual.6.weight", "decoder.upsamples.7.resample.1.bias", "decoder.upsamples.7.resample.1.weight", "decoder.upsamples.7.time_conv.bias", "decoder.upsamples.7.time_conv.weight", "decoder.upsamples.8.residual.0.gamma", "decoder.upsamples.8.residual.2.bias", "decoder.upsamples.8.residual.2.weight", "decoder.upsamples.8.residual.3.gamma", "decoder.upsamples.8.residual.6.bias", "decoder.upsamples.8.residual.6.weight", "decoder.upsamples.9.residual.0.gamma", "decoder.upsamples.9.residual.2.bias", "decoder.upsamples.9.residual.2.weight", "decoder.upsamples.9.residual.3.gamma", "decoder.upsamples.9.residual.6.bias", "decoder.upsamples.9.residual.6.weight".
2025-09-27 22:42:41
[ERROR] 命令执行失败，返回码: 1
2025-09-27 22:42:41 预缓存 1/2 完成!
2025-09-27 22:42:41 开始执行预缓存2/2...
2025-09-27 22:42:41 I:\AIProject\Loratrain\LoRAMaster.venv\Scripts\python.exe I:\AIProject\Loratrain\LoRAMaster\musubi-tuner\src\musubi_tuner\qwen_image_cache_text_encoder_outputs.py --dataset_config I:\AIProject\Loratrain\LoRAMaster\samples\toml\qwen_image_saiboguofengji.toml --text_encoder E:\ComfyUI_windows_portable\ComfyUI\models\clip\qwen_2.5_vl_7b.safetensors --batch_size 1 --fp8_vl
2025-09-27 22:42:47 INFO:main:Load dataset config from I:\AIProject\Loratrain\LoRAMaster\samples\toml\qwen_image_saiboguofengji.toml
2025-09-27 22:42:47 INFO:musubi_tuner.dataset.image_video_dataset:glob images in I:\LoRA_train\saiboguofengji\images
2025-09-27 22:42:47 INFO:musubi_tuner.dataset.image_video_dataset:found 35 images
2025-09-27 22:42:47 INFO:musubi_tuner.dataset.config_utils:[Dataset 0]
2025-09-27 22:42:47 is_image_dataset: True
2025-09-27 22:42:47 resolution: (1024, 1024)
2025-09-27 22:42:47 batch_size: 1
2025-09-27 22:42:47 num_repeats: 2
2025-09-27 22:42:47 caption_extension: ".txt"
2025-09-27 22:42:47 enable_bucket: False
2025-09-27 22:42:47 bucket_no_upscale: False
2025-09-27 22:42:47 cache_directory: "I:\LoRA_train\saiboguofengji\qwen_image_cache"
2025-09-27 22:42:47 debug_dataset: False
2025-09-27 22:42:47 image_directory: "I:\LoRA_train\saiboguofengji\images"
2025-09-27 22:42:47 image_jsonl_file: "None"
2025-09-27 22:42:47 fp_latent_window_size: 9
2025-09-27 22:42:47 fp_1f_clean_indices: None
2025-09-27 22:42:47 fp_1f_target_index: None
2025-09-27 22:42:47 fp_1f_no_post: False
2025-09-27 22:42:47 flux_kontext_no_resize_control: False
2025-09-27 22:42:47 qwen_image_edit_no_resize_control: False
2025-09-27 22:42:47 qwen_image_edit_control_resolution: None
2025-09-27 22:42:47
2025-09-27 22:42:47
2025-09-27 22:42:47 INFO:main:Loading Qwen2.5-VL: E:\ComfyUI_windows_portable\ComfyUI\models\clip\qwen_2.5_vl_7b.safetensors
2025-09-27 22:42:47 INFO:musubi_tuner.qwen_image.qwen_image_utils:Loading state dict from E:\ComfyUI_windows_portable\ComfyUI\models\clip\qwen_2.5_vl_7b.safetensors
2025-09-27 22:42:54 INFO:musubi_tuner.qwen_image.qwen_image_utils:Loaded Qwen2.5-VL:
2025-09-27 22:42:54 INFO:musubi_tuner.qwen_image.qwen_image_utils:prepare Qwen2.5-VL for fp8: set to torch.float8_e4m3fn from torch.bfloat16
2025-09-27 22:42:54 INFO:musubi_tuner.qwen_image.qwen_image_utils:Loading tokenizer from Qwen/Qwen-Image
2025-09-27 22:42:57 INFO:main:Encoding with Qwen2.5-VL
2025-09-27 22:42:57 INFO:musubi_tuner.cache_text_encoder_outputs:Encoding dataset [0]
2025-09-27 22:42:57 Trying to import sageattention
2025-09-27 22:42:57 Failed to import sageattention
2025-09-27 22:42:57
2025-09-27 22:42:58 0it [00:00, ?it/s]
2025-09-27 22:42:58 1it [00:00, 1.02it/s]
2025-09-27 22:42:58 3it [00:01, 3.27it/s]
2025-09-27 22:42:58 5it [00:01, 5.45it/s]
2025-09-27 22:42:58 7it [00:01, 7.45it/s]
2025-09-27 22:42:59 9it [00:01, 9.20it/s]
2025-09-27 22:42:59 11it [00:01, 10.65it/s]
2025-09-27 22:42:59 13it [00:01, 11.75it/s]
2025-09-27 22:42:59 15it [00:01, 12.58it/s]
2025-09-27 22:42:59 17it [00:02, 13.19it/s]
2025-09-27 22:42:59 19it [00:02, 11.89it/s]
2025-09-27 22:42:59 21it [00:02, 12.57it/s]
2025-09-27 22:43:00 23it [00:02, 13.13it/s]
2025-09-27 22:43:00 25it [00:02, 13.35it/s]
2025-09-27 22:43:00 27it [00:02, 13.66it/s]
2025-09-27 22:43:00 29it [00:02, 13.83it/s]
2025-09-27 22:43:00 31it [00:03, 11.57it/s]
2025-09-27 22:43:00 33it [00:03, 12.34it/s]
2025-09-27 22:43:00 35it [00:03, 12.99it/s]
2025-09-27 22:43:00 35it [00:03, 10.06it/s]
2025-09-27 22:43:01 预缓存 2/2 完成!
I:\AIProject\Loratrain\LoRAMaster.venv\Scripts\python.exe I:\AIProject\Loratrain\LoRAMaster\musubi-tuner\src\musubi_tuner\qwen_image_train_network.py
2025-09-27 22:46:11 [INFO] LoRAMaster: 开始运行 Wan LoRA训练命令...
2025-09-27 22:46:11 [INFO] LoRAMaster: I:\AIProject\Loratrain\LoRAMaster.venv\Scripts\python.exe -m accelerate.commands.launch --num_cpu_threads_per_process 1 --mixed_precision bf16 --num_processes 100 --gpu_ids 0 I:\AIProject\Loratrain\LoRAMaster\musubi-tuner\src\musubi_tuner\qwen_image_train_network.py --dit E:\ComfyUI_windows_portable\ComfyUI\models\diffusion_models\qwen_image_bf16.safetensors --dataset_config I:\AIProject\Loratrain\LoRAMaster\samples\toml\qwen_image_saiboguofengji.toml --vae E:\ComfyUI_windows_portable\ComfyUI\models\vae\qwen_image_vae_t.safetensors --text_encoder E:\ComfyUI_windows_portable\ComfyUI\models\clip\qwen_2.5_vl_7b.safetensors --mixed_precision bf16 --optimizer_type adamw8bit --learning_rate 5e-5 --gradient_checkpointing --gradient_accumulation_steps=1 --max_data_loader_n_workers 2 --persistent_data_loader_workers --network_module networks.lora_qwen_image --network_dim 32 --timestep_sampling shift --discrete_flow_shift 2.2 --max_train_epochs 200 --save_every_n_epochs 10 --save_every_n_steps 100 --output_dir ./output --output_name tiquhuawen_v7_x --seed 42 --log_with tensorboard --lr_scheduler constant --sdpa --blocks_to_swap 20 --fp8_base --logging_dir ./logs
2025-09-27 22:46:20 [INFO] LoRAMaster: INFO:musubi_tuner.hv_train_network:Load dataset config from I:\AIProject\Loratrain\LoRAMaster\samples\toml\qwen_image_saiboguofengji.toml
2025-09-27 22:46:20 [INFO] LoRAMaster: INFO:musubi_tuner.dataset.image_video_dataset:glob images in I:\LoRA_train\saiboguofengji\images
2025-09-27 22:46:20 [INFO] LoRAMaster: INFO:musubi_tuner.dataset.image_video_dataset:found 35 images
2025-09-27 22:46:20 [INFO] LoRAMaster: INFO:musubi_tuner.dataset.config_utils:[Dataset 0]
2025-09-27 22:46:20 [INFO] LoRAMaster: is_image_dataset: True
2025-09-27 22:46:20 [INFO] LoRAMaster: resolution: (1024, 1024)
2025-09-27 22:46:20 [INFO] LoRAMaster: batch_size: 1
2025-09-27 22:46:20 [INFO] LoRAMaster: num_repeats: 2
2025-09-27 22:46:20 [INFO] LoRAMaster: caption_extension: ".txt"
2025-09-27 22:46:20 [INFO] LoRAMaster: enable_bucket: False
2025-09-27 22:46:20 [INFO] LoRAMaster: bucket_no_upscale: False
2025-09-27 22:46:20 [INFO] LoRAMaster: cache_directory: "I:\LoRA_train\saiboguofengji\qwen_image_cache"
2025-09-27 22:46:20 [INFO] LoRAMaster: debug_dataset: False
2025-09-27 22:46:20 [INFO] LoRAMaster: image_directory: "I:\LoRA_train\saiboguofengji\images"
2025-09-27 22:46:20 [INFO] LoRAMaster: image_jsonl_file: "None"
2025-09-27 22:46:20 [INFO] LoRAMaster: fp_latent_window_size: 9
2025-09-27 22:46:20 [INFO] LoRAMaster: fp_1f_clean_indices: None
2025-09-27 22:46:20 [INFO] LoRAMaster: fp_1f_target_index: None
2025-09-27 22:46:20 [INFO] LoRAMaster: fp_1f_no_post: False
2025-09-27 22:46:20 [INFO] LoRAMaster: flux_kontext_no_resize_control: False
2025-09-27 22:46:20 [INFO] LoRAMaster: qwen_image_edit_no_resize_control: False
2025-09-27 22:46:20 [INFO] LoRAMaster: qwen_image_edit_control_resolution: None
2025-09-27 22:46:20 [INFO] LoRAMaster:
2025-09-27 22:46:20 [INFO] LoRAMaster:
2025-09-27 22:46:20 [INFO] LoRAMaster: INFO:musubi_tuner.dataset.image_video_dataset:total batches: 0
2025-09-27 22:46:20 [INFO] LoRAMaster: Trying to import sageattention
2025-09-27 22:46:20 [INFO] LoRAMaster: Failed to import sageattention
2025-09-27 22:46:20 [INFO] LoRAMaster: Traceback (most recent call last):
2025-09-27 22:46:20 [INFO] LoRAMaster: File "I:\AIProject\Loratrain\LoRAMaster\musubi-tuner\src\musubi_tuner\qwen_image_train_network.py", line 433, in
2025-09-27 22:46:20 [INFO] LoRAMaster: main()
2025-09-27 22:46:20 [INFO] LoRAMaster: File "I:\AIProject\Loratrain\LoRAMaster\musubi-tuner\src\musubi_tuner\qwen_image_train_network.py", line 429, in main
2025-09-27 22:46:20 [INFO] LoRAMaster: trainer.train(args)
2025-09-27 22:46:20 [INFO] LoRAMaster: File "I:\AIProject\Loratrain\LoRAMaster\musubi-tuner\src\musubi_tuner\hv_train_network.py", line 1667, in train
2025-09-27 22:46:20 [INFO] LoRAMaster: raise ValueError(
2025-09-27 22:46:20 [INFO] LoRAMaster: ValueError: No training items found in the dataset. Please ensure that the latent/Text Encoder cache has been created beforehand. / データセットに学習データがありません。latent/Text Encoderキャッシュを事前に作成したか確認してください
2025-09-27 22:46:20 [INFO] LoRAMaster: Traceback (most recent call last):
2025-09-27 22:46:20 [INFO] LoRAMaster: File "", line 198, in _run_module_as_main
2025-09-27 22:46:20 [INFO] LoRAMaster: File "", line 88, in _run_code
2025-09-27 22:46:20 [INFO] LoRAMaster: File "I:\AIProject\Loratrain\LoRAMaster.venv\Lib\site-packages\accelerate\commands\launch.py", line 1223, in
2025-09-27 22:46:20 [INFO] LoRAMaster: main()
2025-09-27 22:46:20 [INFO] LoRAMaster: File "I:\AIProject\Loratrain\LoRAMaster.venv\Lib\site-packages\accelerate\commands\launch.py", line 1219, in main
2025-09-27 22:46:20 [INFO] LoRAMaster: launch_command(args)
2025-09-27 22:46:20 [INFO] LoRAMaster: File "I:\AIProject\Loratrain\LoRAMaster.venv\Lib\site-packages\accelerate\commands\launch.py", line 1213, in launch_command
2025-09-27 22:46:20 [INFO] LoRAMaster: simple_launcher(args)
2025-09-27 22:46:20 [INFO] LoRAMaster: File "I:\AIProject\Loratrain\LoRAMaster.venv\Lib\site-packages\accelerate\commands\launch.py", line 795, in simple_launcher
2025-09-27 22:46:20 [INFO] LoRAMaster: raise subprocess.CalledProcessError(returncode=process.returncode, cmd=cmd)
2025-09-27 22:46:20 [INFO] LoRAMaster: subprocess.CalledProcessError: Command '['I:\AIProject\Loratrain\LoRAMaster\.venv\Scripts\python.exe', 'I:\AIProject\Loratrain\LoRAMaster\musubi-tuner\src\musubi_tuner\qwen_image_train_network.py', '--dit', 'E:\ComfyUI_windows_portable\ComfyUI\models\diffusion_models\qwen_image_bf16.safetensors', '--dataset_config', 'I:\AIProject\Loratrain\LoRAMaster\samples\toml\qwen_image_saiboguofengji.toml', '--vae', 'E:\ComfyUI_windows_portable\ComfyUI\models\vae\qwen_image_vae_t.safetensors', '--text_encoder', 'E:\ComfyUI_windows_portable\ComfyUI\models\clip\qwen_2.5_vl_7b.safetensors', '--mixed_precision', 'bf16', '--optimizer_type', 'adamw8bit', '--learning_rate', '5e-5', '--gradient_checkpointing', '--gradient_accumulation_steps=1', '--max_data_loader_n_workers', '2', '--persistent_data_loader_workers', '--network_module', 'networks.lora_qwen_image', '--network_dim', '32', '--timestep_sampling', 'shift', '--discrete_flow_shift', '2.2', '--max_train_epochs', '200', '--save_every_n_epochs', '10', '--save_every_n_steps', '100', '--output_dir', './output', '--output_name', 'tiquhuawen_v7_x', '--seed', '42', '--log_with', 'tensorboard', '--lr_scheduler', 'constant', '--sdpa', '--blocks_to_swap', '20', '--fp8_base', '--logging_dir', './logs']' returned non-zero exit status 1.
2025-09-27 22:46:21 [INFO] LoRAMaster:
[ERROR] 命令执行失败，返回码: 1
`