# Mask2Former: CUDA training **URL:** \ **Category:** Models\ **Created:** [July 17, 2023, 11:32am UTC](https://discuss.huggingface.co/t/mask2former-cuda-training/47072 "2023-07-17T11:32:18Z")\ **Posts on this page:** 6\ **Page:** 1
**Author:** ![homo-luden](https://avatars.discourse-cdn.com/v4/letter/h/439d5e/32.png) [@homo-luden](https://discuss.huggingface.co/u/homo-luden)\ **Post date:** [July 17, 2023, 11:32am UTC](https://discuss.huggingface.co/t/mask2former-cuda-training/47072/1 "2023-07-17T11:32:18Z")
I am trying to train Mask2Former with CUDA enabled, but I am encountering the following error: “RuntimeError: Input type (torch.cuda.FloatTensor) and weight type (torch.FloatTensor) should be the same.” However, it works fine when CUDA is disabled. Can someone please help me fix this issue? Thank you in advance! ```auto config = Mask2FormerConfig(feature_size=CFG.img_size[0], mask_feature_size=CFG.img_size[0]) image_processor = Mask2FormerImageProcessor(config) model = Mask2FormerForUniversalSegmentation(config).to(device) criterion = nn.CrossEntropyLoss() torch.cuda.empty_cache() epochs = 1000 scaler = GradScaler() valid_fold = 3 train_df = df[df['fold'] != valid_fold] valid_df = df[df['fold'] == valid_fold] train_dataset = HubmapDataset(train_df, config) valid_dataset = HubmapDataset(valid_df, config) train_dataloader = DataLoader(train_dataset, batch_size=CFG.batch_size, shuffle=True) valid_dataloader = DataLoader(valid_dataset, batch_size=2 * CFG.batch_size, shuffle=False) optimizer, scheduler = get_optimizer_and_scheduler(model, train_dataloader, "adamw") for epoch in range(epochs): model.train() total_train_loss = 0 total_test_loss = 0 pbar = tqdm(train_dataloader, desc=f"Train: Epoch {epoch + 1}", total=len(train_dataloader), mininterval=5) for inputs in pbar: optimizer.zero_grad() imgs = inputs[0] imgs = image_processor(list(imgs), return_tensors="pt", size=(512,512)) #imgs = imgs.to(device) ings = image_processor.encode_inputs(pixel_values_list=imgs['pixel_values'], task_inputs=['instance'], segmentation_maps=inputs[1], ignore_index=0, return_tensors='pt') ings = ings.convert_to_tensors() for k in ings.keys(): try: ings[k] = ings[k].to(device) except AttributeError: ings[k] = torch.stack(ings[k]).to(device) outputs = model(**ings) loss = outputs.loss scaler.scale(loss).backward() scaler.step(optimizer) scale = scaler.get_scale() scaler.update() skip_lr_scheduler = scale > scaler.get_scale() if scheduler is not None and not skip_lr_scheduler: scheduler.step() lr = scheduler.get_last_lr()[0] if scheduler else CFG.one_cycle_max_lr loss = loss.item() pbar.set_postfix({"loss": loss, "lr": lr}) total_train_loss += loss total_train_loss /= len(train_dataloader) model.eval() pbar = tqdm(valid_dataloader, desc=f"Validation: Epoch {epoch + 1}", total=len(valid_dataloader), mininterval=5) for inputs in pbar: imgs = inputs[0] imgs = image_processor(list(imgs), return_tensors="pt", size=(512,512)) #imgs = imgs.to(device) ings = image_processor.encode_inputs(pixel_values_list=imgs['pixel_values'], task_inputs=['instance'], segmentation_maps=inputs[1], ignore_index=0, return_tensors='pt') ings = ings.convert_to_tensors() outputs = model(inputs) loss = outputs.item() pbar.set_postfix({"loss": loss}) total_test_loss += loss total_test_loss /= len(valid_dataloader) print(f'TOTAL TRAIN LOSS: {total_train_loss} | TOTAL VALID LOSS: {total_test_loss}') ``` ---
**Author:** ![nielsr](https://sea2.discourse-cdn.com/hellohellohello/user_avatar/discuss.huggingface.co/nielsr/32/39617_2.png) [@nielsr](https://discuss.huggingface.co/u/nielsr)\ **Post date:** [July 17, 2023, 2:28pm UTC](https://discuss.huggingface.co/t/mask2former-cuda-training/47072/2 "2023-07-17T14:28:10Z")
Hi, The error > RuntimeError: Input type (torch.cuda.FloatTensor) and weight type (torch.FloatTensor) should be the same. suggests that your inputs are on the GPU, but your model is not. So I’d double check whether your model is effectively on the GPU. ---
**Author:** ![homo-luden](https://avatars.discourse-cdn.com/v4/letter/h/439d5e/32.png) [@homo-luden](https://discuss.huggingface.co/u/homo-luden)\ **Post date:** [July 19, 2023, 8:56pm UTC](https://discuss.huggingface.co/t/mask2former-cuda-training/47072/4 "2023-07-19T20:56:40Z")
In the second row I move my model to the device, which is ‘cuda’, but looks like it doen’t move properly. So how can I check if the model is effectively on the GPU? ---
**Author:** ![nielsr](https://sea2.discourse-cdn.com/hellohellohello/user_avatar/discuss.huggingface.co/nielsr/32/39617_2.png) [@nielsr](https://discuss.huggingface.co/u/nielsr)\ **Post date:** [July 22, 2023, 5:13pm UTC](https://discuss.huggingface.co/t/mask2former-cuda-training/47072/5 "2023-07-22T17:13:58Z")
You can check by typing `nvidia-smi` in a terminal to see whether memory is being occupied. ---
**Author:** ![chokevin8](https://avatars.discourse-cdn.com/v4/letter/c/6f9a4e/32.png) [@chokevin8](https://discuss.huggingface.co/u/chokevin8)\ **Post date:** [July 28, 2023, 12:17am UTC](https://discuss.huggingface.co/t/mask2former-cuda-training/47072/6 "2023-07-28T00:17:38Z")
@homo-luden @nielsr I have the same exact problem. Did anyone one of you guys figure it out? I have the inputs in the GPU but it keeps saying model is in CPU when I clearly call model= model.to(device). My nvidia-smi shows GPU memory usage as well. This is perhaps a bug? I’ve tried downgrading to transformers to version 4.27.0 to no success as well. I am using python 3.10. I would appreciate some help, thank you! ---
**Author:** ![nielsr](https://sea2.discourse-cdn.com/hellohellohello/user_avatar/discuss.huggingface.co/nielsr/32/39617_2.png) [@nielsr](https://discuss.huggingface.co/u/nielsr)\ **Post date:** [July 30, 2023, 12:36pm UTC](https://discuss.huggingface.co/t/mask2former-cuda-training/47072/7 "2023-07-30T12:36:05Z")
Hi, I’m not able to reproduce this issue. The model is correctly placed on the GPU as shown in this notebook: [Google Colab](https://colab.research.google.com/drive/1lMbaBlL98nk0FB-0VFiGpkozuP_J-RUE?usp=sharing).