jurgens1206
commented
2 weeks ago Hello.
I was also having trouble because the same error occurred, but when I reviewed the image data I was trying to learn, I realized that grayscale was mixed in, and I converted it to RGB, and it worked properly.
I will post the address of the blog referenced below.
Well then.
The following is the text after the TRAINTRAIN run.
Start Training! Loading weights [5353d90e0c] from /content/drive/MyDrive/sd/stable-diffusion-webui-forge/models/Stable-diffusion/anyloraCheckpoint_bakedvaeBlessedFp16.safetensors model_type EPS UNet ADM Dimension 0 Using xformers attention in VAE Working with z of shape (1, 4, 32, 32) = 4096 dimensions. Using xformers attention in VAE extra {'cond_stage_model.clip_l.text_projection', 'cond_stage_model.clip_l.logit_scale'} left over keys: dict_keys(['alphas_cumprod', 'alphas_cumprod_prev', 'betas', 'log_one_minus_alphas_cumprod', 'posterior_log_variance_clipped', 'posterior_mean_coef1', 'posterior_mean_coef2', 'posterior_variance', 'sqrt_alphas_cumprod', 'sqrt_one_minus_alphas_cumprod', 'sqrt_recip_alphas_cumprod', 'sqrt_recipm1_alphas_cumprod']) loaded straight to GPU To load target model BaseModel Begin to load 1 model [Memory Management] Current Free GPU Memory (MB) = 13313.25634765625 [Memory Management] Model Memory (MB) = 0.00762939453125 [Memory Management] Minimal Inference Memory (MB) = 1024.0 [Memory Management] Estimated Remaining GPU Memory (MB) = 12289.248718261719 Moving model(s) has taken 0.02 seconds To load target model SD1ClipModel Begin to load 1 model [Memory Management] Current Free GPU Memory (MB) = 13313.24755859375 [Memory Management] Model Memory (MB) = 454.2076225280762 [Memory Management] Minimal Inference Memory (MB) = 1024.0 [Memory Management] Estimated Remaining GPU Memory (MB) = 11835.039936065674 Moving model(s) has taken 0.11 seconds Model loaded in 12.4s (unload existing model: 2.9s, load weights from disk: 0.4s, forge load real models: 8.8s, calculate empty prompt: 0.1s). Preparing the Model... Enabling Xformers Preparing image latents and text-conditional... max bucket sizes : [(512, 512)] sub bucket sizes : [] Traceback (most recent call last): File "/content/drive/MyDrive/sd/stable-diffusion-webui-forge/extensions/sd-webui-traintrain/trainer/train.py", line 195, in train_main result = train_lora(t) File "/content/drive/MyDrive/sd/stable-diffusion-webui-forge/extensions/sd-webui-traintrain/trainer/train.py", line 226, in train_lora dataloaders = dataset.make_dataloaders(t) File "/content/drive/MyDrive/sd/stable-diffusion-webui-forge/extensions/sd-webui-traintrain/trainer/dataset.py", line 15, in make_dataloaders load_resize_image_and_text(t) #画像を読み込み、画像サイズごとに振り分け、リサイズ、テキストの読み込み File "/content/drive/MyDrive/sd/stable-diffusion-webui-forge/extensions/sd-webui-traintrain/trainer/dataset.py", line 241, in load_resize_image_and_text resized, alpha_mask = resize_and_crop(ar_error, image, *max, t.image_disable_upscale) File "/content/drive/MyDrive/sd/stable-diffusion-webui-forge/extensions/sd-webui-traintrain/trainer/dataset.py", line 231, in resize_and_crop tensor = torch.from_numpy(np.array(image)).permute(2, 0, 1).float() / 255.0 RuntimeError: permute(sparse_coo): number of dimensions in the tensor input does not match the length of the desired ordering of dimensions i.e. input.dim() = 2 is not equal to len(dims) = 3
Loading weights [9807d1172b] from /content/drive/MyDrive/sd/stable-diffusion-webui-forge/models/Stable-diffusion/anyloracleanlinearmix_v10.safetensors model_type EPS UNet ADM Dimension 0 Using xformers attention in VAE Working with z of shape (1, 4, 32, 32) = 4096 dimensions. Using xformers attention in VAE extra {'cond_stage_model.clip_l.text_projection', 'cond_stage_model.clip_l.logit_scale'} left over keys: dict_keys(['model_ema.decay', 'model_ema.num_updates']) loaded straight to GPU To load target model BaseModel Begin to load 1 model [Memory Management] Current Free GPU Memory (MB) = 13294.35791015625 [Memory Management] Model Memory (MB) = 0.00762939453125 [Memory Management] Minimal Inference Memory (MB) = 1024.0 [Memory Management] Estimated Remaining GPU Memory (MB) = 12270.350280761719 Moving model(s) has taken 0.02 seconds To load target model SD1ClipModel Begin to load 1 model [Memory Management] Current Free GPU Memory (MB) = 13294.34912109375 [Memory Management] Model Memory (MB) = 454.2076225280762 [Memory Management] Minimal Inference Memory (MB) = 1024.0 [Memory Management] Estimated Remaining GPU Memory (MB) = 11816.141498565674 Moving model(s) has taken 0.10 seconds Model loaded in 21.2s (load weights from disk: 1.2s, forge load real models: 19.6s, calculate empty prompt: 0.2s).
That's all. Reinstalling traintrain did not solve the problem and searching did not find a solution.
How can I solve this problem?