{"metadata":{"kernelspec":{"language":"python","display_name":"Python 3","name":"python3"},"language_info":{"pygments_lexer":"ipython3","nbconvert_exporter":"python","version":"3.6.4","file_extension":".py","codemirror_mode":{"name":"ipython","version":3},"name":"python","mimetype":"text/x-python"}},"nbformat_minor":4,"nbformat":4,"cells":[{"cell_type":"markdown","source":"# SETI PyTorch XLA/TPU starter\n![image.png](attachment:image.png)\n\n### If you found this helpful, please give it an upvote!\n\n## Introduction\n\n[PyTorch XLA](https://pytorch.org/xla/master) is a PyTorch library for XLA support. XLA (Accelerated Linear Algebra) is a domain-specific compiler that was originally meant for compiling and accelerating TensorFlow models. However, other packages, like JAX and now PyTorch XLA can compile program with XLA to accelerate code. TPUs can be programmed with XLA programs and PyTorch XLA provides this interface with TPUs by compiling our PyTorch code as XLA programs to run on TPU devices.\n\nIn this kernel, I provide an in-depth look into how you can use PyTorch XLA to **train a PyTorch model on the TPU** for the SETI Breakthrough Listen - E.T. Signal Search competition.","metadata":{},"attachments":{"image.png":{"image/png":"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"}}},{"cell_type":"markdown","source":"## Installs & Imports\n\nThe below cell will install the PyTorch XLA package.","metadata":{}},{"cell_type":"code","source":"!curl https://raw.githubusercontent.com/pytorch/xla/master/contrib/scripts/env-setup.py -o pytorch-xla-env-setup.py  > /dev/null\n!python pytorch-xla-env-setup.py --version 1.8.1 > /dev/null","metadata":{"_uuid":"8f2839f25d086af736a60e9eeb907d3b93b6e0e5","_cell_guid":"b1076dfc-b9ad-4769-8c92-a6c4dae69d19","execution":{"iopub.status.busy":"2021-06-07T06:29:58.221049Z","iopub.execute_input":"2021-06-07T06:29:58.221484Z","iopub.status.idle":"2021-06-07T06:31:07.508621Z","shell.execute_reply.started":"2021-06-07T06:29:58.221384Z","shell.execute_reply":"2021-06-07T06:31:07.507593Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"The below cell will install the [timm]() library, which is what we will use to define our models and get pretrained weights.","metadata":{}},{"cell_type":"code","source":"!pip install timm  > /dev/null","metadata":{"execution":{"iopub.status.busy":"2021-06-07T06:31:07.510645Z","iopub.execute_input":"2021-06-07T06:31:07.511066Z","iopub.status.idle":"2021-06-07T06:31:14.750421Z","shell.execute_reply.started":"2021-06-07T06:31:07.511012Z","shell.execute_reply":"2021-06-07T06:31:14.749121Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"Here are all of our imports!","metadata":{}},{"cell_type":"code","source":"import gc\nimport os\nimport time\nimport torch\nimport random\nimport albumentations\n\nimport numpy as np\nimport pandas as pd\n\nimport cv2\nfrom PIL import Image\n\nimport torch.nn as nn\nfrom sklearn import metrics\nfrom sklearn import model_selection\nfrom torch.nn import functional as F\nfrom torch.optim import Adam\n\nimport torch_xla.core.xla_model as xm\nimport torch_xla.distributed.parallel_loader as pl\nimport torch_xla.distributed.xla_multiprocessing as xmp\nimport torch_xla.utils.serialization as xser\n\n\nimport timm\n\nimport warnings\nwarnings.filterwarnings(\"ignore\")\n\n\nos.environ['XLA_USE_BF16']=\"1\"\nos.environ['XLA_TENSOR_ALLOCATOR_MAXSIZE'] = '100000000'","metadata":{"execution":{"iopub.status.busy":"2021-06-07T06:31:14.752330Z","iopub.execute_input":"2021-06-07T06:31:14.752604Z","iopub.status.idle":"2021-06-07T06:31:19.091036Z","shell.execute_reply.started":"2021-06-07T06:31:14.752571Z","shell.execute_reply":"2021-06-07T06:31:19.089375Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"## Definitions\n\nNow let's define the necessary functions and variables needed for training.","metadata":{}},{"cell_type":"markdown","source":"These are the flags for training. When you fork (after you upvote of course, 😉), feel free to play around with these flags!","metadata":{}},{"cell_type":"code","source":"FLAGS = {\n    \n    'fold': 0,\n    'seed': 999,\n    'model': 'resnext50_32x4d',\n    'pretrained': True,\n    'batch_size': 64,\n    'num_workers': 8,\n    'lr': 3e-4,\n    'epochs': 3\n}","metadata":{"execution":{"iopub.status.busy":"2021-06-07T06:31:19.094059Z","iopub.execute_input":"2021-06-07T06:31:19.094536Z","iopub.status.idle":"2021-06-07T06:31:19.099545Z","shell.execute_reply.started":"2021-06-07T06:31:19.094481Z","shell.execute_reply":"2021-06-07T06:31:19.098457Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"def seed_everything(seed):\n    random.seed(seed)\n    os.environ['PYTHONHASHSEED'] = str(seed)\n    np.random.seed(seed)\n    torch.manual_seed(seed)\n    torch.cuda.manual_seed(seed)\n    torch.backends.cudnn.deterministic = True\n\nseed_everything(FLAGS['seed'])","metadata":{"execution":{"iopub.status.busy":"2021-06-07T06:31:19.101118Z","iopub.execute_input":"2021-06-07T06:31:19.101507Z","iopub.status.idle":"2021-06-07T06:31:19.123801Z","shell.execute_reply.started":"2021-06-07T06:31:19.101476Z","shell.execute_reply":"2021-06-07T06:31:19.122593Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"Here, I define a model class for the timm models.","metadata":{}},{"cell_type":"code","source":"# Using Ross Wightman's timm package\nclass TimmModels(nn.Module):\n    def __init__(self, model_name,pretrained=True, num_classes=1, inp_chan=1):\n        super(TimmModels, self).__init__()\n        self.m = timm.create_model(model_name,pretrained=pretrained,in_chans=inp_chan)\n        model_list = list(self.m.children())\n        model_list[-1] = nn.Linear(\n            in_features=model_list[-1].in_features, \n            out_features=num_classes, \n            bias=True\n        )\n        self.m = nn.Sequential(*model_list)\n        \n    def forward(self, image):\n        out = self.m(image)\n        return out","metadata":{"execution":{"iopub.status.busy":"2021-06-07T06:31:19.125181Z","iopub.execute_input":"2021-06-07T06:31:19.125666Z","iopub.status.idle":"2021-06-07T06:31:19.134062Z","shell.execute_reply.started":"2021-06-07T06:31:19.125620Z","shell.execute_reply":"2021-06-07T06:31:19.133004Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"Here, I define a class for the PyTorch Dataset (taken from my [earlier notebook](https://www.kaggle.com/tanlikesmath/seti-et-signal-detection-a-simple-cnn-starter/notebook)).","metadata":{}},{"cell_type":"code","source":"class SETIDataset:\n    def __init__(self, df, spatial=True, sixchan=True):\n        self.df = df\n        self.spatial = spatial\n        self.sixchan = sixchan\n        \n        \n    def __len__(self):\n        return len(self.df)\n\n    def __getitem__(self, index):\n        label = self.df.iloc[index].target\n        filename = self.df.iloc[index].path\n        data = np.load(filename).astype(np.float32)\n        if not self.sixchan: data = data[::2].astype(np.float32)\n        if self.spatial:\n            data = np.vstack(data).transpose((1, 0))\n            data = cv2.resize(data, dsize=(256,256))     \n            data_tensor = torch.tensor(data).float().unsqueeze(0)\n        else:\n            data = np.transpose(data, (1,2,0))\n            data = cv2.resize(data, dsize=(256,256))     \n            data = np.transpose(data, (2, 0, 1)).astype(np.float32)\n            data_tensor = torch.tensor(data).float()\n            \n        \n\n        return (data_tensor, torch.tensor(label))","metadata":{"execution":{"iopub.status.busy":"2021-06-07T06:31:19.136002Z","iopub.execute_input":"2021-06-07T06:31:19.137338Z","iopub.status.idle":"2021-06-07T06:31:19.155108Z","shell.execute_reply.started":"2021-06-07T06:31:19.137290Z","shell.execute_reply":"2021-06-07T06:31:19.152809Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"I now create my folds. Of course, you can replace with your own CV setup here.","metadata":{}},{"cell_type":"code","source":"# create folds\ndf = pd.read_csv(\"../input/seti-breakthrough-listen/train_labels.csv\")\ndf['path'] = df['id'].apply(lambda x: '../input/seti-breakthrough-listen/'+'train/'+x[0]+'/'+x+'.npy') #adding the path for each id for easier processing\ndf[\"kfold\"] = -1    \ndf = df.sample(frac=1).reset_index(drop=True)\ny = df.target.values\nkf = model_selection.StratifiedKFold(n_splits=5)\n\nfor f, (t_, v_) in enumerate(kf.split(X=df, y=y)):\n    df.loc[v_, 'kfold'] = f\n\ndf.to_csv(\"train_folds.csv\", index=False)","metadata":{"execution":{"iopub.status.busy":"2021-06-07T06:31:19.159494Z","iopub.execute_input":"2021-06-07T06:31:19.159843Z","iopub.status.idle":"2021-06-07T06:31:19.628316Z","shell.execute_reply.started":"2021-06-07T06:31:19.159812Z","shell.execute_reply":"2021-06-07T06:31:19.627565Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"## Training code","metadata":{}},{"cell_type":"markdown","source":"Let's start training! To do so, we start by initializing the model. Let's make sure we initialize with the correct number of classes and input channels. We use the `xmp.MpModelWrapper` provided by PyTorch XLA to save memory when initializing the model.","metadata":{}},{"cell_type":"code","source":"MX = xmp.MpModelWrapper(TimmModels(FLAGS['model'],pretrained=FLAGS['pretrained'], num_classes=1, inp_chan=1))","metadata":{"execution":{"iopub.status.busy":"2021-06-07T06:31:19.630318Z","iopub.execute_input":"2021-06-07T06:31:19.630642Z","iopub.status.idle":"2021-06-07T06:31:22.362134Z","shell.execute_reply.started":"2021-06-07T06:31:19.630578Z","shell.execute_reply":"2021-06-07T06:31:22.359500Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"Let's now define our training and validationfunctions. ","metadata":{}},{"cell_type":"code","source":"def train_loop_fn(data_loader, loss_fn, model, optimizer, device, scheduler=None):\n    model.train() # put model in training mode\n    for bi, d in enumerate(data_loader): # enumerate through the dataloader\n        \n        images = d[0] # obtain the ids\n        targets = d[1] # obtain the target\n\n        # pass image to model\n        optimizer.zero_grad()\n        outputs = model(images)\n        \n        # calculate loss\n        loss = loss_fn(outputs, targets.unsqueeze(1).float())\n        \n        # backpropagate\n        loss.backward()\n        \n        # Use PyTorch XLA optimizer stepping\n        xm.optimizer_step(optimizer)\n        \n        # Step the scheduler\n        if scheduler is not None: scheduler.step()\n    \n    # since the loss is on all 8 cores, reduce the loss values and print the average\n    loss_reduced = xm.mesh_reduce('loss_reduce',loss, lambda x: sum(x) / len(x)) \n    # master_print will only print once (not from all 8 cores)\n    xm.master_print(f'bi={bi}, train loss={loss_reduced}')\n        \n    model.eval() # put model in eval mode for later use\n    \ndef eval_loop_fn(data_loader, loss_fn, model, device):\n    fin_targets = []\n    fin_outputs = []\n    for bi, d in enumerate(data_loader): # enumerate through dataloader\n        \n        images = d[0] # obtain the ids\n        targets = d[1]# # obtain the targets\n        \n\n        # pass image to model\n        with torch.no_grad(): outputs = model(images)\n\n        # Add the outputs and targets to a list \n        targets_np = targets.cpu().detach().numpy().tolist()\n        outputs_np = outputs.cpu().detach().numpy().tolist()\n        fin_targets.extend(targets_np)\n        fin_outputs.extend(outputs_np)    \n        del targets_np, outputs_np\n        gc.collect() # delete for memory conservation\n                \n    o,t = np.array(fin_outputs), np.array(fin_targets)\n    \n    # calculate loss\n    loss = loss_fn(torch.tensor(o), torch.tensor(t).unsqueeze(1).float())\n    # since the loss is on all 8 cores, reduce the loss values and print the average\n    loss_reduced = xm.mesh_reduce('loss_reduce',loss, lambda x: sum(x) / len(x)) \n    # master_print will only print once (not from all 8 cores)\n    xm.master_print(f'val. loss={loss_reduced}')\n    \n    # since the output/target values are on all 8 cores, reduce/gather the values for metric calculation\n    o_reduced = xm.mesh_reduce('o_reduce', torch.tensor(o).to(device), torch.cat)\n    t_reduced = xm.mesh_reduce('t_reduce', torch.tensor(t).to(device), torch.cat)\n    \n    # metric calculation\n    auc = metrics.roc_auc_score(t_reduced.cpu(),o_reduced.cpu())\n        \n    xm.master_print(f'val. auc = {auc}')","metadata":{"execution":{"iopub.status.busy":"2021-06-07T06:31:22.364711Z","iopub.execute_input":"2021-06-07T06:31:22.365001Z","iopub.status.idle":"2021-06-07T06:31:22.405173Z","shell.execute_reply.started":"2021-06-07T06:31:22.364973Z","shell.execute_reply":"2021-06-07T06:31:22.401941Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"Finally, we define a main function that we will run on each of the 8 cores of the TPU.","metadata":{}},{"cell_type":"code","source":"def run(rank, flags):\n    global FLAGS\n    torch.set_default_tensor_type('torch.FloatTensor')\n    \n    df = pd.read_csv(\"/kaggle/working/train_folds.csv\") #read train csv created earlier\n    device = xm.xla_device() #device, will be different for each core on the TPU\n    xm.set_rng_state(FLAGS['seed'], device)\n    epochs = FLAGS['epochs']\n    fold = FLAGS['fold']\n    \n    df_train = df[df.kfold != fold].reset_index(drop=True)\n    df_valid = df[df.kfold == fold].reset_index(drop=True)\n    \n    \n    train_dataset = SETIDataset(df_train)    \n    valid_dataset = SETIDataset(df_valid)\n\n    \n    # special sampler needed for distributed/multi-core (divides dataset among the replicas/cores/devices)\n    train_sampler = torch.utils.data.distributed.DistributedSampler(\n        train_dataset,\n        num_replicas=xm.xrt_world_size(), #divide dataset among this many replicas\n        rank=xm.get_ordinal(), #which replica/device/core\n        shuffle=True)\n    \n    # define DataLoader with the defined sampler\n    train_loader = torch.utils.data.DataLoader(\n        train_dataset,\n        batch_size=FLAGS['batch_size'],\n        sampler=train_sampler,\n        num_workers=FLAGS['num_workers'],\n        drop_last=True)\n    \n    # same as train but with valid data\n    valid_sampler = torch.utils.data.distributed.DistributedSampler(\n        valid_dataset,\n        num_replicas=xm.xrt_world_size(),\n        rank=xm.get_ordinal(),\n        shuffle=False)\n\n    valid_loader = torch.utils.data.DataLoader(\n        valid_dataset,\n        batch_size=FLAGS['batch_size'],\n        sampler=valid_sampler,\n        num_workers=FLAGS['num_workers'],\n        drop_last=False)\n    \n    \n    \n    train_loader = pl.MpDeviceLoader(train_loader, device) # puts the train data onto the current TPU core\n    valid_loader = pl.MpDeviceLoader(valid_loader, device) # puts the valid data onto the current TPU core\n    \n\n    model = MX.to(device) # put model onto the current TPU core\n    loss_fn = nn.BCEWithLogitsLoss()\n    optimizer = Adam(model.parameters(), lr=FLAGS['lr']*xm.xrt_world_size()) # often a good idea to scale the learning rate by number of cores\n    scheduler = torch.optim.lr_scheduler.CosineAnnealingLR(optimizer, len(train_loader)*FLAGS['epochs']) #let's use a scheduler\n\n    gc.collect()\n    \n    xm.master_print(f'========== training fold {FLAGS[\"fold\"]} for {FLAGS[\"epochs\"]} epochs ==========')\n    for i in range(FLAGS['epochs']):\n        xm.master_print(f'EPOCH {i}:')\n        # train one epoch\n        train_loop_fn(train_loader, loss_fn, model, optimizer, device, scheduler)\n                \n        # validation one epoch\n        eval_loop_fn(valid_loader, loss_fn, model, device)\n\n        gc.collect()\n    \n    xm.rendezvous('save_model')\n    \n    xm.master_print('save model')\n    \n    xm.save(model.state_dict(), f'xla_trained_model_{FLAGS[\"epochs\"]}_epochs_fold_{FLAGS[\"fold\"]}.pth')","metadata":{"execution":{"iopub.status.busy":"2021-06-07T06:31:22.410640Z","iopub.execute_input":"2021-06-07T06:31:22.412566Z","iopub.status.idle":"2021-06-07T06:31:22.466795Z","shell.execute_reply.started":"2021-06-07T06:31:22.412504Z","shell.execute_reply":"2021-06-07T06:31:22.464394Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"Let's train all 5 folds!","metadata":{}},{"cell_type":"code","source":"for i in range(5):\n    FLAGS['fold'] = i\n    start_time = time.time()\n    xmp.spawn(run, args=(FLAGS,), nprocs=8, start_method='fork')\n    print('time taken: ', time.time()-start_time)\n    print('==============================================================================')","metadata":{"execution":{"iopub.status.busy":"2021-06-07T06:31:22.470276Z","iopub.execute_input":"2021-06-07T06:31:22.470890Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"#### Now, **WE ARE DONE!**\n\nIf you enjoyed this kernel, please give it an upvote. If you have any questions or suggestions, please leave a comment!\n\nFeel free to check out my [EDA notebook](https://www.kaggle.com/tanlikesmath/seti-simple-eda-to-help-you-get-started) to learn more about the SETI E.T. Signal competition.\n\nAlso, check out my [related notebook](https://www.kaggle.com/tanlikesmath/the-ultimate-pytorch-tpu-tutorial-jigsaw-xlm-r) with more detailed information on PyTorch XLA/TPU training.","metadata":{}}]}