{"cells":[{"cell_type":"markdown","id":"10-setup-hdr","metadata":{"papermill":{"duration":0.007721,"end_time":"2026-09-12T01:51:21.531281+00:00","exception":false,"start_time":"2026-09-12T01:51:21.52356+00:00","status":"completed"},"tags":[]},"source":["# 1.0 — Setup & Imports\n","\n","Mount Google Drive when running in Colab, import all dependencies, fix random seeds,\n","declare project paths, and set global training hyperparameters."]},{"cell_type":"markdown","id":"11-imports-hdr","metadata":{"papermill":{"duration":0.006124,"end_time":"2026-09-12T01:51:21.544169+00:00","exception":false,"start_time":"2026-09-12T01:51:21.538045+00:00","status":"completed"},"tags":[]},"source":["## 1.1 — Library Imports & Environment\n","\n","Standard-library → third-party → deep-learning imports, then Colab Drive mount and\n","global random-seed fixation so experiments are reproducible."]},{"cell_type":"code","execution_count":1,"id":"11-imports-code","metadata":{"execution":{"iopub.execute_input":"2026-09-12T01:51:21.558594Z","iopub.status.busy":"2026-09-12T01:51:21.558123Z","iopub.status.idle":"2026-09-12T01:51:50.89631Z","shell.execute_reply":"2026-09-12T01:51:50.895188Z"},"papermill":{"duration":29.347704,"end_time":"2026-09-12T01:51:50.898085+00:00","exception":false,"start_time":"2026-09-12T01:51:21.550381+00:00","status":"completed"},"tags":[]},"outputs":[{"name":"stdout","output_type":"stream","text":["GPU Active: 1 device(s) detected: ['/physical_device:GPU:0']\n","Keras unsafe deserialization enabled.\n","Python : 3.12.13\n","TensorFlow: 2.20.0\n","Keras : 3.13.2\n"]}],"source":["import os\n","import sys\n","import io\n","import json\n","import math\n","import gc\n","import random\n","import shutil\n","import time\n","import subprocess\n","from pathlib import Path\n","\n","from scipy.io import loadmat\n","\n","# ── Colab environment fallback ─────────────────────────────────────────────\n","if 'google.colab' in sys.modules:\n"," from google.colab import drive\n"," drive.mount('/content/drive', force_remount=False)\n","\n","# ── Dynamic dependency checks/imports (openpyxl, pillow, etc.) ─────────────\n","for pkg in ['openpyxl', 'pillow']:\n"," try:\n"," __import__(pkg)\n"," except ImportError:\n"," subprocess.run([sys.executable, '-m', 'pip', 'install', '-q', pkg], check=True)\n","\n","# ── Third-party ────────────────────────────────────────────────────────────\n","import numpy as np\n","import pandas as pd\n","import seaborn as sns\n","import matplotlib.pyplot as plt\n","from matplotlib import gridspec\n","from openpyxl import Workbook, load_workbook\n","from openpyxl.drawing.image import Image as XLImage\n","from openpyxl.utils.dataframe import dataframe_to_rows\n","from sklearn.model_selection import train_test_split\n","from sklearn.metrics import confusion_matrix, accuracy_score, cohen_kappa_score, classification_report\n","\n","# ── TensorFlow / Keras ─────────────────────────────────────────────────────\n","import tensorflow as tf\n","import tensorflow_probability as tfp\n","from tensorflow import keras\n","from tensorflow.keras import backend as K, Sequential, layers, activations, optimizers\n","from tensorflow.keras.layers import (\n"," Input, Add, Multiply, Reshape, Dense, Activation, BatchNormalization,\n"," Flatten, Dropout, concatenate, Lambda,\n",")\n","from tensorflow.keras.layers import (\n"," Conv2D, AveragePooling2D, MaxPooling2D, GlobalAveragePooling2D,\n"," GlobalAvgPool2D, DepthwiseConv2D, SeparableConv2D, MaxPool2D, UpSampling2D,\n",")\n","from tensorflow.keras.layers import (\n"," Conv2DTranspose, add, multiply, LayerNormalization,\n",")\n","from tensorflow.python.util.tf_export import keras_export\n","from tensorflow.python.ops import array_ops\n","from tensorflow.python.keras.utils import control_flow_util\n","from tensorflow.keras.callbacks import ModelCheckpoint\n","from tensorflow.keras.models import load_model, Model\n","from tensorflow.keras.utils import plot_model\n","from keras.regularizers import l2\n","\n","# ── GPU memory growth configuration ────────────────────────────────────────\n","gpus = tf.config.list_physical_devices('GPU')\n","if gpus:\n"," for gpu in gpus:\n"," try:\n"," tf.config.experimental.set_memory_growth(gpu, True)\n"," except RuntimeError as e:\n"," print(f\"Memory growth error: {e}\")\n"," print(f\"GPU Active: {len(gpus)} device(s) detected: {[g.name for g in gpus]}\")\n","else:\n"," print(\"WARNING: No physical GPU detected!\")\n","\n","# ── Keras unsafe deserialization configuration ─────────────────────────────\n","try:\n"," keras.config.enable_unsafe_deserialization()\n"," print(\"Keras unsafe deserialization enabled.\")\n","except Exception as e:\n"," print(f\"Notice: keras.config.enable_unsafe_deserialization() not available or failed: {e}\")\n","\n","print(f\"Python : {sys.version.split()[0]}\")\n","print(f\"TensorFlow: {tf.__version__}\")\n","print(f\"Keras : {keras.__version__}\")\n","\n","# ── Global styling & reproducibility seeds ─────────────────────────────────\n","sns.set()\n","np.random.seed(1337)\n","random.seed(1337)\n","tf.random.set_seed(1337)\n","\n","# ── Dynamic callable references for static analysis parity ─────────────────\n","new_layer = None\n","lyr = None\n"]},{"cell_type":"markdown","id":"12-paths-hdr","metadata":{"papermill":{"duration":0.006173,"end_time":"2026-09-12T01:51:50.910518+00:00","exception":false,"start_time":"2026-09-12T01:51:50.904345+00:00","status":"completed"},"tags":[]},"source":["## 1.2 — Project Paths & Training Hyperparameters\n","\n","Resolve the project root on Drive, create output directories, and define all\n","training constants (patch size, epoch count, batch size, dropout, image dimensions)."]},{"cell_type":"code","execution_count":2,"id":"12-paths-code","metadata":{"execution":{"iopub.execute_input":"2026-09-12T01:51:50.924553Z","iopub.status.busy":"2026-09-12T01:51:50.923921Z","iopub.status.idle":"2026-09-12T01:51:50.950872Z","shell.execute_reply":"2026-09-12T01:51:50.949988Z"},"papermill":{"duration":0.035586,"end_time":"2026-09-12T01:51:50.952415+00:00","exception":false,"start_time":"2026-09-12T01:51:50.916829+00:00","status":"completed"},"tags":[]},"outputs":[{"name":"stdout","output_type":"stream","text":["Environment: Kaggle\n","Data directory: /kaggle/input/datasets/daneshselwal/dias-dataset-k01\n","Data file: /kaggle/input/datasets/daneshselwal/dias-dataset-k01/k001.mat (exists: True)\n","Label file: /kaggle/input/datasets/daneshselwal/dias-dataset-k01/k002.mat (exists: True)\n","Output directory: /kaggle/working/results\n","Saved model dir: /kaggle/working/results/models\n","Training workbook: /kaggle/working/results/model_training_multihead_results.xlsx\n","Model registry: /kaggle/working/results/model_registry_multihead.json\n"]}],"source":["# ── Project directory structure & environment detection ─────────────────────\n","IN_KAGGLE = Path('/kaggle').exists() or Path('/kaggle/working').exists()\n","IN_COLAB = 'google.colab' in sys.modules\n","\n","if IN_KAGGLE:\n"," # Look for raw DIAS data files in Kaggle input\n"," k001_primary = Path('/kaggle/input/dias-dataset-k01/k001.mat')\n"," k002_primary = Path('/kaggle/input/dias-dataset-k01/k002.mat')\n","\n"," if k001_primary.exists() and k002_primary.exists():\n"," DATA_FILE = k001_primary\n"," LABEL_FILE = k002_primary\n"," DATA_DIR = k001_primary.parent\n"," else:\n"," # Fallback dynamic glob search\n"," dias_candidates = (\n"," list(Path('/kaggle/input').glob('**/k001.mat')) or\n"," list(Path('/kaggle/input').glob('**/DIAS.mat'))\n"," )\n"," ref_candidates = (\n"," list(Path('/kaggle/input').glob('**/k002.mat')) or\n"," list(Path('/kaggle/input').glob('**/DIAS_ref.mat'))\n"," )\n"," DATA_FILE = dias_candidates[0] if dias_candidates else k001_primary\n"," LABEL_FILE = ref_candidates[0] if ref_candidates else k002_primary\n"," DATA_DIR = DATA_FILE.parent\n","\n"," OUTPUT_ROOT = Path('/kaggle/working/results')\n"," SAVED_MODEL_DIR = OUTPUT_ROOT / 'models'\n"," TRAINING_RESULTS_DIR = OUTPUT_ROOT\n","elif IN_COLAB:\n"," PROJECT_ROOT = Path('/content/drive/My Drive/Classification')\n"," DATA_DIR = PROJECT_ROOT / 'data'\n"," DATA_FILE = DATA_DIR / 'DIAS.mat'\n"," LABEL_FILE = DATA_DIR / 'DIAS_ref.mat'\n"," OUTPUT_ROOT = PROJECT_ROOT / 'multicp' / 'results'\n"," SAVED_MODEL_DIR = OUTPUT_ROOT / 'models'\n"," TRAINING_RESULTS_DIR = OUTPUT_ROOT\n","else:\n"," PROJECT_ROOT = Path.cwd()\n"," DATA_DIR = Path('data') if Path('data').exists() else PROJECT_ROOT / 'data'\n"," DATA_FILE = DATA_DIR / 'DIAS.mat' if (DATA_DIR / 'DIAS.mat').exists() else DATA_DIR / 'k001.mat'\n"," LABEL_FILE = DATA_DIR / 'DIAS_ref.mat' if (DATA_DIR / 'DIAS_ref.mat').exists() else DATA_DIR / 'k002.mat'\n"," OUTPUT_ROOT = Path('multicp/results')\n"," SAVED_MODEL_DIR = OUTPUT_ROOT / 'models'\n"," TRAINING_RESULTS_DIR = OUTPUT_ROOT\n","\n","SAVED_MODEL_DIR.mkdir(parents=True, exist_ok=True)\n","TRAINING_RESULTS_DIR.mkdir(parents=True, exist_ok=True)\n","\n","TRAINING_WORKBOOK = OUTPUT_ROOT / 'model_training_multihead_results.xlsx'\n","MODEL_REGISTRY_PATH = OUTPUT_ROOT / 'model_registry_multihead.json'\n","\n","# ── Global training hyperparameters ────────────────────────────────────────\n","Data_to_use = '6_bands'\n","use_pearson_corr = False\n","P_S = 9 # patch size (pixels)\n","Targeted_accuracy = 0.985\n","Min_trainable_epoch = 20\n","train_percent = 75\n","epoch = 100\n","BATCH_SIZE = 128 # batch size\n","dropout_rate = 0.25\n","shifts = int(1 / dropout_rate) # number of staged-dropout shifts = 4\n","H, W, B = None, None, None\n","PERCENTILE_CLIP = (2.0, 98.0) # robust per-band stretch range (2nd-98th percentile)\n","\n","# ── Output folder paths for figure and model export ────────────────────────\n","folder_path = str(OUTPUT_ROOT) + os.sep\n","os.makedirs(os.path.join(folder_path, 'Results'), exist_ok=True)\n","os.makedirs(os.path.join(folder_path, 'Trained models'), exist_ok=True)\n","\n","print('Environment: ', 'Kaggle' if IN_KAGGLE else ('Colab' if IN_COLAB else 'Local'))\n","print('Data directory: ', DATA_DIR)\n","print('Data file: ', DATA_FILE, f'(exists: {DATA_FILE.exists()})')\n","print('Label file: ', LABEL_FILE, f'(exists: {LABEL_FILE.exists()})')\n","print('Output directory: ', OUTPUT_ROOT)\n","print('Saved model dir: ', SAVED_MODEL_DIR)\n","print('Training workbook: ', TRAINING_WORKBOOK)\n","print('Model registry: ', MODEL_REGISTRY_PATH)\n"]},{"cell_type":"markdown","id":"20-data-hdr","metadata":{"papermill":{"duration":0.006046,"end_time":"2026-09-12T01:51:50.964558+00:00","exception":false,"start_time":"2026-09-12T01:51:50.958512+00:00","status":"completed"},"tags":[]},"source":["# 2.0 — Data Loading & Patch Preparation\n","\n","Load multispectral data and reference labels, per-band min-max normalise the image,\n","extract labeled spatial patches, visualise class balance, and produce the\n","stratified train/test split."]},{"cell_type":"markdown","id":"21-load-hdr","metadata":{"papermill":{"duration":0.006534,"end_time":"2026-09-12T01:51:50.977168+00:00","exception":false,"start_time":"2026-09-12T01:51:50.970634+00:00","status":"completed"},"tags":[]},"source":["## 2.1 — Load & Normalise\n","\n","Read the hyperspectral/multispectral scene (H×W×Bands) and labels (H×W) from .mat files.\n","Each spectral band is robustly stretched to [0, 1] using a percentile clip."]},{"cell_type":"code","execution_count":3,"id":"21-load-code","metadata":{"execution":{"iopub.execute_input":"2026-09-12T01:51:50.991468Z","iopub.status.busy":"2026-09-12T01:51:50.99103Z","iopub.status.idle":"2026-09-12T01:51:51.724124Z","shell.execute_reply":"2026-09-12T01:51:51.723156Z"},"papermill":{"duration":0.742513,"end_time":"2026-09-12T01:51:51.725988+00:00","exception":false,"start_time":"2026-09-12T01:51:50.983475+00:00","status":"completed"},"tags":[]},"outputs":[{"name":"stdout","output_type":"stream","text":["shape of x: (512, 512, 65)\n","shape of y: (512, 512)\n"]}],"source":["def _first_mat_array(mat_dict, key=None):\n"," \"\"\"Return the named variable, or the first non-metadata array in the .mat file.\"\"\"\n"," if key is not None:\n"," return mat_dict[key]\n"," candidates = [v for k, v in mat_dict.items() if not k.startswith('__')]\n"," if not candidates:\n"," raise ValueError(f'No data arrays found in {mat_dict.keys()}')\n"," return candidates[0]\n","\n","data_mat = loadmat(DATA_FILE)\n","label_mat = loadmat(LABEL_FILE)\n","\n","x = _first_mat_array(data_mat).astype(np.float32)\n","y = _first_mat_array(label_mat).astype(np.int32)\n","\n","if x.ndim != 3:\n"," raise ValueError(f'Expected a (H, W, Bands) cube, got shape {x.shape}')\n","if y.shape != x.shape[:2]:\n"," raise ValueError(f'Label shape {y.shape} does not match image shape {x.shape[:2]}')\n","\n","H, W, B = x.shape\n","\n","# Robust per-band stretch: clip to the PERCENTILE_CLIP range before min-max\n","# scaling, instead of literal min/max. A handful of saturated/no-data pixels\n","# (e.g. sensor clipping at the int8 range limits -128/127) would otherwise\n","# dominate the stretch and crush the dynamic range of the rest of the scene.\n","for band_idx in range(B):\n"," band = x[:, :, band_idx]\n"," lo = np.percentile(band, PERCENTILE_CLIP[0])\n"," hi = np.percentile(band, PERCENTILE_CLIP[1])\n"," denom = max(hi - lo, 1e-8)\n"," x[:, :, band_idx] = np.clip((band - lo) / denom, 0.0, 1.0)\n","\n","print('shape of x:', x.shape)\n","print('shape of y:', y.shape)"]},{"cell_type":"markdown","id":"22-patch-hdr","metadata":{"papermill":{"duration":0.008141,"end_time":"2026-09-12T01:51:51.740912+00:00","exception":false,"start_time":"2026-09-12T01:51:51.732771+00:00","status":"completed"},"tags":[]},"source":["## 2.2 — Patch Extraction\n","\n","Edge-pad the image by `(P_S - 1) / 2` pixels, then slide over every labeled pixel\n","to collect `P_S × P_S × B` patches. Background pixels (label == 0) are skipped."]},{"cell_type":"code","execution_count":4,"id":"22-patch-code","metadata":{"execution":{"iopub.execute_input":"2026-09-12T01:51:51.754726Z","iopub.status.busy":"2026-09-12T01:51:51.754415Z","iopub.status.idle":"2026-09-12T01:51:52.495963Z","shell.execute_reply":"2026-09-12T01:51:52.495013Z"},"papermill":{"duration":0.750606,"end_time":"2026-09-12T01:51:52.497852+00:00","exception":false,"start_time":"2026-09-12T01:51:51.747246+00:00","status":"completed"},"tags":[]},"outputs":[{"name":"stdout","output_type":"stream","text":["shape of padded_x: (520, 520, 65)\n","20.5585% of data is labeled with 8 classes\n"]}],"source":["y_shape = y.shape[0] * y.shape[1]\n","pad_width = int((P_S - 1) / 2)\n","padded_x = np.pad(x, [(pad_width, pad_width), (pad_width, pad_width), (0, 0)], 'edge')\n","\n","X, Y = [], []\n","for row_idx in range(H): # renamed a → row_idx (single-letter variable)\n"," for col_idx in range(W): # renamed b → col_idx (single-letter variable)\n"," if y[row_idx][col_idx] != 0:\n"," patch = padded_x[row_idx:row_idx + P_S, col_idx:col_idx + P_S, :]\n"," X.append(patch)\n"," Y.append(y[row_idx][col_idx] - 1) # zero-index labels\n","\n","X = np.array(X, dtype=np.float32)\n","Y = np.array(Y, dtype=np.int32)\n","num_classes = len(np.unique(y)) - 1 # exclude background class 0\n","\n","print('shape of padded_x:', padded_x.shape)\n","print(f'{len(Y) * 100 / y_shape:.4f}% of data is labeled with {num_classes} classes')"]},{"cell_type":"markdown","id":"23-split-hdr","metadata":{"papermill":{"duration":0.006723,"end_time":"2026-09-12T01:51:52.511485+00:00","exception":false,"start_time":"2026-09-12T01:51:52.504762+00:00","status":"completed"},"tags":[]},"source":["## 2.3 — Class Distribution & Train/Test Split\n","\n","Display a bar chart of per-class pixel counts, then create a stratified split\n","keeping `train_percent`% for training."]},{"cell_type":"code","execution_count":5,"id":"23-split-code","metadata":{"execution":{"iopub.execute_input":"2026-09-12T01:51:52.526079Z","iopub.status.busy":"2026-09-12T01:51:52.525788Z","iopub.status.idle":"2026-09-12T01:51:53.149712Z","shell.execute_reply":"2026-09-12T01:51:53.148781Z"},"papermill":{"duration":0.633426,"end_time":"2026-09-12T01:51:53.151488+00:00","exception":false,"start_time":"2026-09-12T01:51:52.518062+00:00","status":"completed"},"tags":[]},"outputs":[{"data":{"image/png":"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\n","text/plain":["
"]},"metadata":{},"output_type":"display_data"},{"name":"stdout","output_type":"stream","text":["x_train: (40419, 9, 9, 65)\n","x_test: (13474, 9, 9, 65)\n","y_train: (40419,)\n","y_test: (13474,)\n"]}],"source":["# Approximate RGB visualisation (middle three bands)\n","Approximate_rgb_img = x[:, :, [B // 2 - 1, B // 2, B // 2 + 1]] * 255\n","\n","class_labels, value_counts = np.unique(y.reshape(-1, 1), return_counts=True)\n","plt.figure(figsize=(15, 5))\n","plt.bar(class_labels[1:] - 1, value_counts[1:])\n","plt.xticks(class_labels, rotation=0)\n","plt.xlabel('classes')\n","plt.ylabel('Number of pixels')\n","plt.title('Distribution Plot')\n","plt.show()\n","\n","x_train, x_test, y_train, y_test = train_test_split(\n"," X, Y, train_size=(train_percent / 100), stratify=Y, random_state=10\n",")\n","print(\n"," f'x_train: {x_train.shape}\\n'\n"," f'x_test: {x_test.shape}\\n'\n"," f'y_train: {y_train.shape}\\n'\n"," f'y_test: {y_test.shape}'\n",")"]},{"cell_type":"markdown","id":"30-shared-hdr","metadata":{"papermill":{"duration":0.00737,"end_time":"2026-09-12T01:51:53.16614+00:00","exception":false,"start_time":"2026-09-12T01:51:53.15877+00:00","status":"completed"},"tags":[]},"source":["# 3.0 — Shared Metrics, Layers, and Callbacks\n","\n","Define visualisation helpers, classification metric reporting, the Pearson-correlation\n","attention layer, a staged-dropout layer, and the custom training callback used by all\n","three models."]},{"cell_type":"markdown","id":"31-viz-hdr","metadata":{"papermill":{"duration":0.006842,"end_time":"2026-09-12T01:51:53.179838+00:00","exception":false,"start_time":"2026-09-12T01:51:53.172996+00:00","status":"completed"},"tags":[]},"source":["## 3.1 — Visualisation & Metric Helpers\n","\n","`plot_accuracy_loss_curve` overlays accuracy and loss on dual axes.\n","`performance_meausures` (name preserved as-is for backward compatibility) renders a\n","4-panel figure: classification report, confusion matrix, score summary, and parameter\n","count — and now correctly returns the figure for downstream export."]},{"cell_type":"code","execution_count":6,"id":"31-viz-code","metadata":{"execution":{"iopub.execute_input":"2026-09-12T01:51:53.195088Z","iopub.status.busy":"2026-09-12T01:51:53.194576Z","iopub.status.idle":"2026-09-12T01:51:53.20921Z","shell.execute_reply":"2026-09-12T01:51:53.208366Z"},"papermill":{"duration":0.024387,"end_time":"2026-09-12T01:51:53.210943+00:00","exception":false,"start_time":"2026-09-12T01:51:53.186556+00:00","status":"completed"},"tags":[]},"outputs":[],"source":["def plot_accuracy_loss_curve(history, use_pearson_corr=None, folder_path=None):\n"," \"\"\"Plot and optionally save training accuracy and loss curves.\"\"\"\n"," train_loss = history.history['loss']\n"," val_loss = history.history['val_loss']\n"," train_accuracy = history.history['accuracy']\n"," val_accuracy = history.history['val_accuracy']\n","\n"," fig = plt.figure(figsize=(24, 8))\n"," ax = plt.subplot(1, 1, 1)\n"," ax2 = ax.twinx()\n","\n"," ax.plot(train_accuracy, color='blue', marker='o', linewidth=1.5, markersize=2, label='train_accuracy')\n"," ax.plot(val_accuracy, color='green', marker='o', linewidth=1.5, markersize=2, label='val_accuracy')\n"," ax.grid()\n"," ax.legend()\n"," plt.xlabel('no. of epoches')\n"," plt.ylabel('accuracy')\n","\n"," ax2.plot(train_loss, color='black', marker='o', linewidth=1.5, markersize=2, label='train_loss')\n"," ax2.plot(val_loss, color='red', marker='o', linewidth=1.5, markersize=2, label='val_loss')\n"," ax2.grid()\n"," ax2.legend()\n"," plt.xlabel('no. of epoches')\n"," plt.ylabel('loss')\n"," plt.title('accuracy and loss plot for model performance')\n"," plt.show()\n","\n"," if folder_path:\n"," results_dir = os.path.join(folder_path, \"Results\")\n"," os.makedirs(results_dir, exist_ok=True)\n"," prefix = \"Pearson_Corr \" if use_pearson_corr else \"\"\n"," path = os.path.join(results_dir, f\"{prefix}{train_percent}% ps_{P_S} accuracy_loss.png\")\n"," fig.savefig(path)\n","\n","\n","def performance_meausures(y_test, y_pred, tt, *parameters_summary, folder_path=None):\n"," \"\"\"Compute and visualize classification metrics; returns the figure for export.\n","\n"," NOTE: function name has a typo ('meausures') — preserved for backward compatibility.\n"," \"\"\"\n"," Total_params, Trainable_params, Non_trainable_params = parameters_summary\n"," accuracy = accuracy_score(y_test, y_pred)\n"," kappa = cohen_kappa_score(y_test, y_pred)\n"," cm = confusion_matrix(y_test, y_pred).astype('int32')\n"," cr = classification_report(y_test, y_pred, output_dict=True)\n"," df_cr = pd.DataFrame(cr).T\n"," df_score = pd.DataFrame({\n"," 'accuracy score: ': [accuracy],\n"," 'Cohen_Kappa score: ': [kappa],\n"," \"Training Time: \": [tt],\n"," }).T\n"," df_summary = pd.DataFrame({\n"," 'Total_params: ': [Total_params],\n"," 'Trainable_params: ': [Trainable_params],\n"," 'Non_trainable_params: ': [Non_trainable_params],\n"," }).T\n","\n"," spec = gridspec.GridSpec(\n"," ncols=2, nrows=2, width_ratios=[1, 3],\n"," wspace=0.5, hspace=0.5, height_ratios=[7, 1],\n"," )\n"," fig = plt.figure(figsize=(24, 10))\n","\n"," ax1 = fig.add_subplot(spec[0])\n"," ax1.set_title('classification report')\n"," sns.heatmap(df_cr, cmap='Blues', cbar=False, annot=True, fmt=' .5g', ax=ax1)\n","\n"," ax2 = fig.add_subplot(spec[1])\n"," ax2.set_title('confusion matrix')\n"," ax2.set_xlabel('predicted class')\n"," ax2.set_ylabel('actual class')\n"," sns.heatmap(cm, cmap='Blues', cbar=False, annot=True, fmt=' .5g', ax=ax2)\n","\n"," ax3 = fig.add_subplot(spec[2])\n"," sns.heatmap(df_score, cmap='Blues', cbar=False, annot=True, fmt=' .5g', ax=ax3)\n"," ax3.set_xticks([])\n","\n"," ax4 = fig.add_subplot(spec[3])\n"," sns.heatmap(df_summary, cmap=\"Blues\", cbar=False, annot=True, fmt=' .10g', ax=ax4)\n"," ax4.set_xticks([])\n","\n"," if folder_path:\n"," path = folder_path + \"Results/\" + str(train_percent) + \"% ps_\" + str(P_S) + \" Performance Measure.png\"\n"," fig.savefig(path)\n","\n"," return fig # FIX: was missing; run_training used plt.gcf() as a workaround"]},{"cell_type":"markdown","id":"32-layers-hdr","metadata":{"papermill":{"duration":0.006595,"end_time":"2026-09-12T01:51:53.224314+00:00","exception":false,"start_time":"2026-09-12T01:51:53.217719+00:00","status":"completed"},"tags":[]},"source":["## 3.2 — Custom Keras Layers\n","\n","`Pearson_correlation_masked` computes a spatial Pearson-correlation attention mask\n","relative to the central pixel. `Dropout_Train` implements deterministic channel-shift\n","dropout, cycling through channel groups across training shifts rather than random zeroing."]},{"cell_type":"code","execution_count":7,"id":"32-layers-code","metadata":{"execution":{"iopub.execute_input":"2026-09-12T01:51:53.239212Z","iopub.status.busy":"2026-09-12T01:51:53.238863Z","iopub.status.idle":"2026-09-12T01:51:53.25705Z","shell.execute_reply":"2026-09-12T01:51:53.256166Z"},"papermill":{"duration":0.028038,"end_time":"2026-09-12T01:51:53.258917+00:00","exception":false,"start_time":"2026-09-12T01:51:53.230879+00:00","status":"completed"},"tags":[]},"outputs":[],"source":["class Pearson_correlation_masked(layers.Layer):\n"," \"\"\"Apply a Pearson-correlation attention mask to image patches.\"\"\"\n","\n"," def __init__(self, P_S=9, **kwargs):\n"," \"\"\"Initialize with patch size P_S.\"\"\"\n"," super(Pearson_correlation_masked, self).__init__(**kwargs)\n"," self.P_S = P_S\n","\n"," def call(self, inputs):\n"," \"\"\"Compute per-pixel Pearson correlation with the central patch pixel and apply as attention.\"\"\"\n"," loc = self.P_S // 2\n"," channels = inputs.shape[-1]\n","\n"," # Mean of each pixel's spectral vector across channels\n"," x_mean = tf.repeat(tf.math.reduce_mean(inputs, axis=-1, keepdims=True), repeats=channels, axis=-1)\n","\n"," # Central pixel, tiled to match spatial extent\n"," y = tf.repeat(\n"," tf.repeat(inputs[:, loc:loc+1, loc:loc+1, :], repeats=self.P_S, axis=-2),\n"," repeats=self.P_S, axis=-3,\n"," )\n"," y_mean = tf.repeat(tf.math.reduce_mean(y, axis=-1, keepdims=True), repeats=channels, axis=-1)\n","\n"," a = tf.math.subtract(inputs, x_mean)\n"," b = tf.math.subtract(y, y_mean)\n"," ab = tf.math.multiply(a, b)\n","\n"," num = tf.math.reduce_sum(ab, axis=-1, keepdims=True)\n"," a_sq = tf.math.reduce_sum(tf.math.multiply(a, a), axis=-1, keepdims=True)\n"," b_sq = tf.math.reduce_sum(tf.math.multiply(b, b), axis=-1, keepdims=True)\n"," deno = tf.math.sqrt(tf.math.multiply(a_sq, b_sq))\n"," corr = tf.math.divide(num, deno)\n","\n"," # Threshold at mean correlation; apply mask\n"," thresh = tf.math.reduce_mean(corr)\n"," mask = tf.cast(corr > thresh, corr.dtype)\n"," masked_corr = tf.math.multiply(mask, corr)\n"," attention_wts = tf.repeat(masked_corr, repeats=channels, axis=-1)\n","\n"," return multiply([inputs, attention_wts])\n","\n"," def get_config(self):\n"," \"\"\"Return serializable Keras configuration.\"\"\"\n"," config = super(Pearson_correlation_masked, self).get_config()\n"," config.update({\"P_S\": self.P_S})\n"," return config\n","\n","\n","@keras_export('keras.layers.Dropout')\n","class Dropout_Train(layers.Layer):\n"," \"\"\"Deterministic channel-shift dropout applied during training only.\n","\n"," Instead of random zeroing, successive shifts zero out a different contiguous\n"," slice of channels, ensuring every channel is trained without being dropped.\n"," \"\"\"\n","\n"," def __init__(self, rate, shift=1, noise_shape=None, seed=None, **kwargs):\n"," \"\"\"Initialize with dropout rate and current shift index.\"\"\"\n"," super(Dropout_Train, self).__init__(**kwargs)\n"," if isinstance(rate, (int, float)) and not 0 <= rate <= 1:\n"," raise ValueError(f\"Invalid value {rate} received for `rate`, expected a value between 0 and 1.\")\n"," if not isinstance(shift, int):\n"," raise TypeError(f\"Invalid dtype {type(shift)} found for `shift`. It must be an integer\")\n"," if shift * rate > 1.0:\n"," raise ValueError(\n"," f\"Invalid value {shift} received for `shift`, \"\n"," f\"expected an integer value less than or equal to {int(1 / rate)}\"\n"," )\n"," self.rate = rate\n"," self.shift = shift\n"," self.noise_shape = noise_shape\n"," self.seed = seed\n"," self.supports_masking = True\n","\n"," def _get_noise_shape(self, inputs):\n"," \"\"\"Resolve the concrete dropout noise shape from symbolic noise_shape.\"\"\"\n"," if self.noise_shape is None:\n"," return None\n"," concrete_inputs_shape = array_ops.shape(inputs)\n"," noise_shape = []\n"," for i, value in enumerate(self.noise_shape):\n"," noise_shape.append(concrete_inputs_shape[i] if value is None else value)\n"," return tf.convert_to_tensor(noise_shape)\n","\n"," def call(self, inputs, training=None):\n"," \"\"\"Apply channel-shift dropout when training; identity otherwise.\"\"\"\n"," if self.rate == 0:\n"," return tf.identity(inputs)\n"," if training is None:\n"," training = K.learning_phase()\n","\n"," def dropped_inputs():\n"," \"\"\"Build and apply the binary channel mask for this shift.\"\"\"\n"," input_shape = inputs.shape\n"," range_0 = int(self.rate * (self.shift - 1) * input_shape[-1])\n"," range_1 = (\n"," int(self.rate * self.shift * input_shape[-1])\n"," if self.shift * self.rate < 1.0\n"," else None\n"," )\n"," multiplier = np.ones(input_shape[-1])\n"," multiplier[range_0:range_1] = 0.0\n"," multiplier = tf.constant(multiplier)\n"," return Multiply()([inputs, multiplier])\n","\n"," return control_flow_util.smart_cond(training, dropped_inputs, lambda: array_ops.identity(inputs))\n","\n"," def compute_output_shape(self, input_shape):\n"," \"\"\"Output shape is unchanged from input shape.\"\"\"\n"," return input_shape\n","\n"," def get_config(self):\n"," \"\"\"Return serializable Keras configuration.\"\"\"\n"," config = super(Dropout_Train, self).get_config()\n"," config.update({\n"," \"rate\": self.rate,\n"," \"shift\": self.shift,\n"," \"noise_shape\": self.noise_shape,\n"," \"seed\": self.seed,\n"," \"supports_masking\": self.supports_masking,\n"," })\n"," return config"]},{"cell_type":"markdown","id":"33-callbacks-hdr","metadata":{"papermill":{"duration":0.006872,"end_time":"2026-09-12T01:51:53.27279+00:00","exception":false,"start_time":"2026-09-12T01:51:53.265918+00:00","status":"completed"},"tags":[]},"source":["## 3.3 — Model Modifier & Training Callback\n","\n","`modified_model` rebuilds a Keras model, replacing layers whose name contains\n","`layer_name` with a new layer instance (used to advance the dropout shift stage).\n","`Custom_callbacks` orchestrates staged training: it triggers `modified_model` each\n","time the validation accuracy threshold is met, cycling through all dropout shifts\n","before finalising and saving the best weights."]},{"cell_type":"code","execution_count":8,"id":"33-callbacks-code","metadata":{"execution":{"iopub.execute_input":"2026-09-12T01:51:53.28863Z","iopub.status.busy":"2026-09-12T01:51:53.288007Z","iopub.status.idle":"2026-09-12T01:51:53.305639Z","shell.execute_reply":"2026-09-12T01:51:53.304982Z"},"papermill":{"duration":0.027542,"end_time":"2026-09-12T01:51:53.307398+00:00","exception":false,"start_time":"2026-09-12T01:51:53.279856+00:00","status":"completed"},"tags":[]},"outputs":[],"source":["def modified_model(model, layer_name, rate, new_layer, shift, **kwargs):\n"," \"\"\"Rebuild `model`, replacing layers matching `layer_name` with `new_layer`.\n","\n"," layer_name: substring to match in layer names (case-insensitive)\n"," shift: int → Dropout_Train shift index; str → standard Dropout (e.g. 'Final')\n"," \"\"\"\n"," name = kwargs.get(\"name\", None)\n"," x = model.layers[0].output\n"," modification = False\n"," z = 0\n","\n"," for lyr in model.layers[1:]:\n"," if (layer_name in lyr.name or layer_name in lyr.name.upper()) and isinstance(shift, int):\n"," x = new_layer(rate=rate, shift=shift, name=f\"{layer_name}_{shift}_{z}\")(x)\n"," modification = True\n"," z += 1\n"," elif (layer_name in lyr.name or layer_name in lyr.name.upper()) and isinstance(shift, str):\n"," x = new_layer(rate=rate, name=f\"{layer_name}_{shift}_{z}\")(x)\n"," modification = True\n"," z += 1\n"," else:\n"," x = lyr(x)\n","\n"," if not modification:\n"," print(\"___________________________________Model has not been modified___________________________________\")\n"," return Model(inputs=model.layers[0].input, outputs=x, name=name)\n","\n","\n","class Custom_callbacks(tf.keras.callbacks.Callback):\n"," \"\"\"Manage staged dropout-shift training and best-weight restoration.\n","\n"," Advances the dropout shift whenever val_accuracy >= accuracy_score AND\n"," at least min_epochs epochs have been completed in the current shift.\n"," After all shifts, saves the model with the best observed val_accuracy.\n"," \"\"\"\n","\n"," def __init__(self, filepath, epochs, rate, new_layer=Dropout_Train,\n"," layer_name=\"DROPOUT\", accuracy_score=0.99, min_epochs=50):\n"," \"\"\"Configure staged training parameters.\"\"\"\n"," super(Custom_callbacks, self).__init__()\n"," self.filepath = filepath\n"," self.epochs = epochs\n"," self.new_layer = new_layer\n"," self.rate = rate\n"," self.best = 0.0\n"," self.epoch_num = 1\n"," self.layer_name = layer_name\n"," self.min_epochs = min_epochs # minimum epochs per shift before advancing\n"," self.accuracy_score = accuracy_score if accuracy_score <= 1.0 else accuracy_score / 100.0\n","\n"," def on_train_begin(self, logs=None):\n"," \"\"\"Initialise shift counter and apply the first shift to the model.\"\"\"\n"," print(self.epochs)\n"," self.shift = 1\n"," self.epoch_completed = 0\n"," print(f\"Model will be trained in {int(1 / self.rate)} shifts\")\n"," print(\"Starting training with 1st shift \\n\")\n"," self.model = modified_model(self.model, self.layer_name, self.rate, self.new_layer, self.shift)\n","\n"," def on_train_end(self, logs=None):\n"," \"\"\"Verify all shifts completed, restore best weights, and save the model.\"\"\"\n"," if self.shift <= int(1 / self.rate):\n"," raise NotImplementedError(\n"," f\"model has not trained fully in the available no. of epochs\\n\"\n"," f\" only {self.shift - 1} shifts completed out of {int(1 / self.rate)}\"\n"," )\n"," print(\"Model training completition \", \"███████████\" * self.shift, (self.rate * (self.shift - 1)) * 100, \"%\")\n"," print(f\"Model has been fully trained in {int(1 / self.rate)} shifts\")\n"," self.model.set_weights(self.best_weights)\n"," print(f\"\\nSaving best model to {self.filepath}\")\n"," self.model.save(self.filepath)\n","\n"," def on_epoch_end(self, epoch, logs=None):\n"," \"\"\"Advance the shift or record best weights depending on current training state.\"\"\"\n"," self.epoch_completed += 1\n"," self.epoch_num += 1\n"," total_shifts = int(1 / self.rate)\n"," acc = logs.get(\"val_accuracy\", 0.0)\n"," threshold_met = (acc >= self.accuracy_score) and (self.epoch_completed >= self.min_epochs)\n","\n"," if threshold_met and self.shift < total_shifts:\n"," # Advance to the next dropout shift\n"," print(\"\\nTargeted accuracy has been achieved\")\n"," print(\"Model training completition \", \"███████████\" * self.shift, (self.rate * self.shift) * 100, \"%\")\n"," self.shift += 1\n"," suffix = \"nd\" if self.shift == 2 else \"th\"\n"," print(f\"Modifying the model for {self.shift}{suffix} shift\")\n"," self.model = modified_model(self.model, self.layer_name, self.rate, self.new_layer, self.shift)\n"," self.epoch_completed = 0\n","\n"," elif threshold_met and self.shift == total_shifts:\n"," # All shifts done — switch to standard Dropout for the final phase\n"," print(\"\\nModel training completition \", \"███████████\" * self.shift, (self.rate * self.shift) * 100, \"%\")\n"," print(\"All shifting has been completed\\n\")\n"," print(\"██████████████████████===============> Now redefining the model to standard model <===============██████████████████████\")\n"," self.model = modified_model(self.model, self.layer_name, self.rate, self.new_layer, \"Final\", name=\"AlexNet\")\n"," self.shift += 1\n"," self.epoch_completed = 0\n","\n"," else:\n"," print(\", need more training\")\n"," # Record best weights in the final (post-shift) stage\n"," if self.shift >= total_shifts:\n"," current = logs.get(\"val_accuracy\")\n"," if not np.less(current, self.best) and (self.epoch_num >= self.epochs - 10):\n"," print(f\"val_accuracy improved from {self.best:.4f} to {current:.4f}\")\n"," self.best = current\n"," self.best_weights = self.model.get_weights()\n","\n"," def get_config(self):\n"," \"\"\"Return serializable Keras configuration.\"\"\"\n"," config = super(Custom_callbacks, self).get_config()\n"," config.update({\n"," \"filepath\": self.filepath,\n"," \"epochs\": self.epochs,\n"," \"new_layer\": self.new_layer,\n"," \"rate\": self.rate,\n"," \"best\": self.best,\n"," \"epoch_num\": self.epoch_num,\n"," \"layer_name\": self.layer_name,\n"," \"min_epochs\": self.min_epochs,\n"," \"accuracy_score\": self.accuracy_score,\n"," })\n"," return config"]},{"cell_type":"markdown","id":"40-alex-hdr","metadata":{"papermill":{"duration":0.006797,"end_time":"2026-09-12T01:51:53.321354+00:00","exception":false,"start_time":"2026-09-12T01:51:53.314557+00:00","status":"completed"},"tags":[]},"source":["# 4.0 — AlexNet Multi-Head Model\n","\n","A five-block convolutional backbone (AlexNet-style) followed by three dense layers,\n","each guarded by a `Dropout` layer named `TRAIN_DROPOUT_*` (targeted by `Custom_callbacks`).\n","The backbone fans out to `K_HEADS = 7` independent softmax output heads."]},{"cell_type":"code","execution_count":9,"id":"40-alex-code","metadata":{"execution":{"iopub.execute_input":"2026-09-12T01:51:53.336617Z","iopub.status.busy":"2026-09-12T01:51:53.336331Z","iopub.status.idle":"2026-09-12T01:51:53.345375Z","shell.execute_reply":"2026-09-12T01:51:53.344722Z"},"papermill":{"duration":0.018773,"end_time":"2026-09-12T01:51:53.346966+00:00","exception":false,"start_time":"2026-09-12T01:51:53.328193+00:00","status":"completed"},"tags":[]},"outputs":[],"source":["def AlexNet(input_shape, num_classes=13, use_pearson_corr=False, dropout_rate=0.5):\n"," \"\"\"Build the multi-head AlexNet-style convolutional classifier.\"\"\"\n"," K_HEADS = 7 # number of independent output heads\n","\n"," x_input = Input(input_shape)\n"," X = Pearson_correlation_masked(P_S)(x_input) if use_pearson_corr else x_input\n","\n"," # ── Convolutional backbone ────────────────────────────────────────────\n"," X = Conv2D(filters=96, kernel_size=(3, 3), activation='relu', strides=(1, 1), padding='same')(X)\n"," X = Conv2D(filters=256, kernel_size=(3, 3), activation='relu', strides=(1, 1), padding='same')(X)\n"," X = Conv2D(filters=384, kernel_size=(3, 3), activation='relu', strides=(1, 1), padding='same')(X)\n"," X = Conv2D(filters=384, kernel_size=(3, 3), activation='relu', strides=(1, 1), padding='same')(X)\n"," X = Conv2D(filters=256, kernel_size=(3, 3), activation='relu', strides=(1, 1), padding='same')(X)\n"," X = MaxPooling2D(pool_size=(2, 2), strides=(2, 2), padding='same')(X)\n","\n"," # ── Dense head ───────────────────────────────────────────────────────\n"," X = Flatten()(X)\n"," X = Dense(4096, activation='relu')(X)\n"," X = Dropout(dropout_rate, name=\"TRAIN_DROPOUT_1\")(X)\n"," X = Dense(1024, activation='relu')(X)\n"," X = Dropout(dropout_rate, name=\"TRAIN_DROPOUT_2\")(X)\n"," X = Dense(256, activation='relu')(X)\n"," X = Dropout(dropout_rate, name=\"TRAIN_DROPOUT_3\")(X)\n"," X = Dense(32, activation='relu')(X)\n","\n"," # ── Multi-head outputs ────────────────────────────────────────────────\n"," output_heads = [\n"," Dense(num_classes, activation='softmax', dtype='float32', name=f'head_{i+1}')(X)\n"," for i in range(K_HEADS)\n"," ]\n"," return Model(inputs=x_input, outputs=output_heads, name=\"MultiHead_AlexNet\")"]},{"cell_type":"markdown","id":"50-gfnet-hdr","metadata":{"papermill":{"duration":0.00697,"end_time":"2026-09-12T01:51:53.361166+00:00","exception":false,"start_time":"2026-09-12T01:51:53.354196+00:00","status":"completed"},"tags":[]},"source":["# 5.0 — GFNet Multi-Head Model\n","\n","GFNet replaces self-attention with a learned global filter in the frequency domain\n","(2-D real FFT → element-wise complex multiply → inverse FFT). This section defines\n","all supporting layer classes and the multi-head GFNet builder."]},{"cell_type":"markdown","id":"51-gfnet-layers-hdr","metadata":{"papermill":{"duration":0.007024,"end_time":"2026-09-12T01:51:53.375308+00:00","exception":false,"start_time":"2026-09-12T01:51:53.368284+00:00","status":"completed"},"tags":[]},"source":["## 5.1 — GFNet Layer Classes\n","\n","`GF_MLP`, `GF_DropPath`, `GF_Expand_Dims`, `GF_Patches`, `GF_PatchEncoder`,\n","`GF_GlobalFilter`, and `GF_Block` form the GFNet primitive vocabulary.\n","\n","> **Fixes applied:** `GF_Expand_Dims.config` and `GF_GlobalFilter.config` were\n","> incorrectly named (missing `get_` prefix); both renamed to `get_config` so Keras\n","> serialisation works correctly."]},{"cell_type":"code","execution_count":10,"id":"51-gfnet-layers-code","metadata":{"execution":{"iopub.execute_input":"2026-09-12T01:51:53.391538Z","iopub.status.busy":"2026-09-12T01:51:53.391207Z","iopub.status.idle":"2026-09-12T01:51:53.418561Z","shell.execute_reply":"2026-09-12T01:51:53.417645Z"},"papermill":{"duration":0.037883,"end_time":"2026-09-12T01:51:53.420372+00:00","exception":false,"start_time":"2026-09-12T01:51:53.382489+00:00","status":"completed"},"tags":[]},"outputs":[],"source":["class GF_MLP(layers.Layer):\n"," \"\"\"Feed-forward MLP block used by the GFNet architecture.\"\"\"\n","\n"," def __init__(self, in_features, out_features, drop=0.0, **kwargs):\n"," \"\"\"Initialize with in/out feature dims and dropout rate.\"\"\"\n"," super(GF_MLP, self).__init__(**kwargs)\n"," self.in_features = in_features\n"," self.out_features = out_features\n"," self.drop = drop\n"," self.mlp_1 = Dense(in_features, activation=activations.gelu, use_bias=False)\n"," self.drop_1 = Dropout(drop)\n"," self.mlp_2 = Dense(out_features, activation=activations.gelu, use_bias=False)\n"," self.drop_2 = Dropout(drop)\n","\n"," def call(self, x):\n"," \"\"\"Forward pass: two GELU dense layers with dropout.\"\"\"\n"," x = self.drop_1(self.mlp_1(x))\n"," x = self.drop_2(self.mlp_2(x))\n"," return x\n","\n"," def get_config(self):\n"," \"\"\"Return serializable Keras configuration.\"\"\"\n"," config = super(GF_MLP, self).get_config()\n"," config.update({\"in_features\": self.in_features, \"out_features\": self.out_features, \"drop\": self.drop})\n"," return config\n","\n","\n","class GF_DropPath(layers.Layer):\n"," \"\"\"Stochastic depth (DropPath) regularization for residual paths.\"\"\"\n","\n"," def __init__(self, drop_prob=0.0, training=False, **kwargs):\n"," \"\"\"Initialize with drop probability.\"\"\"\n"," super(GF_DropPath, self).__init__(**kwargs)\n"," self.drop_prob = drop_prob\n"," self.training = training\n","\n"," def call(self, x, **kwargs):\n"," \"\"\"Apply stochastic depth; identity when drop_prob==0 or not training.\"\"\"\n"," if self.drop_prob == 0. or not self.training:\n"," return x\n"," keep_prob = 1 - self.drop_prob\n"," shape = (x.shape[0],) + (1,) * (x.ndim - 1)\n"," random_tensor = keep_prob + tf.random.uniform(shape, dtype=x.dtype)\n"," random_tensor.floor_() # binarize\n"," return tf.divide(x, keep_prob) * random_tensor\n","\n"," def get_config(self):\n"," \"\"\"Return serializable Keras configuration.\"\"\"\n"," config = super(GF_DropPath, self).get_config()\n"," config.update({\"drop_prob\": self.drop_prob, \"training\": self.training})\n"," return config\n","\n","\n","class GF_Expand_Dims(layers.Layer):\n"," \"\"\"Expand tensor dimensions along `ndim` for GFNet compatibility.\"\"\"\n","\n"," def __init__(self, ndim, **kwargs):\n"," \"\"\"Initialize with axis index to expand.\"\"\"\n"," super(GF_Expand_Dims, self).__init__(**kwargs)\n"," self.ndim = ndim\n","\n"," def call(self, x):\n"," \"\"\"Insert a dimension at axis `ndim`.\"\"\"\n"," return tf.expand_dims(x, axis=self.ndim)\n","\n"," def get_config(self, **kwargs): # FIX: was incorrectly named `config`; corrected to `get_config`\n"," \"\"\"Return serializable Keras configuration.\"\"\"\n"," config = super(GF_Expand_Dims, self).get_config()\n"," config.update({\"ndim\": self.ndim})\n"," return config\n","\n","\n","class GF_Patches(layers.Layer):\n"," \"\"\"Extract image patches for the GFNet backbone.\n","\n"," Supports two methods:\n"," - 'conv': strided Conv2D projection (shape → [B, N, hidden_dim])\n"," - 'extract': tf.image.extract_patches (shape → [B, N, patch_dim])\n"," \"\"\"\n","\n"," def __init__(self, patch_size=3, hidden_dim=256, patch_method='extract', **kwargs):\n"," \"\"\"Initialize patch extractor with size, projection dim, and method.\"\"\"\n"," super(GF_Patches, self).__init__(**kwargs)\n"," self.patch_size = patch_size\n"," self.hidden_dim = hidden_dim\n"," self.patch_method = patch_method.lower()\n","\n"," def call(self, images):\n"," \"\"\"Extract patches via Conv2D or tf.image.extract_patches.\"\"\"\n"," if self.patch_method == \"conv\":\n"," x = Conv2D(self.hidden_dim, self.patch_size, self.patch_size)(images)\n"," patches = Reshape([-1, x.shape[-1]])(x)\n"," return patches\n"," else:\n"," batch_size = tf.shape(images)[0]\n"," patches = tf.image.extract_patches(\n"," images=images,\n"," sizes=[1, self.patch_size, self.patch_size, 1],\n"," strides=[1, self.patch_size, self.patch_size, 1],\n"," rates=[1, 1, 1, 1],\n"," padding=\"VALID\",\n"," )\n"," patch_dims = patches.shape[-1]\n"," return tf.reshape(patches, [batch_size, -1, patch_dims])\n","\n"," def get_config(self):\n"," \"\"\"Return serializable Keras configuration.\"\"\"\n"," config = super(GF_Patches, self).get_config()\n"," config.update({\"patch_size\": self.patch_size, \"hidden_dim\": self.hidden_dim, \"patch_method\": self.patch_method})\n"," return config\n","\n","\n","class GF_PatchEncoder(layers.Layer):\n"," \"\"\"Project GFNet patches to `projection_dim` and add positional embeddings.\"\"\"\n","\n"," def __init__(self, num_patches, projection_dim, **kwargs):\n"," \"\"\"Initialize with patch count and projection dimensionality.\"\"\"\n"," super(GF_PatchEncoder, self).__init__(**kwargs)\n"," self.num_patches = num_patches\n"," self.projection_dim = projection_dim\n"," self.projection = layers.Dense(units=projection_dim)\n"," self.position_embedding = layers.Embedding(input_dim=num_patches, output_dim=projection_dim)\n","\n"," def call(self, patch, **kwargs):\n"," \"\"\"Project patches and add sinusoidal-free learned positional embeddings.\"\"\"\n"," positions = tf.range(start=0, limit=self.num_patches, delta=1)\n"," return self.projection(patch) + self.position_embedding(positions)\n","\n"," def get_config(self, **kwargs):\n"," \"\"\"Return serializable Keras configuration.\"\"\"\n"," config = super(GF_PatchEncoder, self).get_config()\n"," config.update({\"num_patches\": self.num_patches, \"projection_dim\": self.projection_dim})\n"," return config\n","\n","\n","class GF_GlobalFilter(layers.Layer):\n"," \"\"\"Apply the Global Filter operation: FFT → learned complex multiply → IFFT.\"\"\"\n","\n"," def __init__(self, patch_size, dim, **kwargs):\n"," \"\"\"Initialize with spatial patch size and feature dimensionality.\"\"\"\n"," super(GF_GlobalFilter, self).__init__(**kwargs)\n"," self.patch_size = patch_size\n"," self.dim = dim\n","\n"," def build(self, input_shape):\n"," \"\"\"Create the learnable complex weight tensor.\"\"\"\n"," w_init = tf.random_uniform_initializer()\n"," self.complex_weight = self.add_weight(\n"," name=\"complex_weight\",\n"," shape=(self.patch_size, self.patch_size, input_shape[-1] // 2 + 1, 2),\n"," initializer=w_init,\n"," trainable=True,\n"," )\n"," super().build(input_shape)\n","\n"," def call(self, x, **kwargs):\n"," \"\"\"Apply real-valued 2-D FFT filtering with learned complex weights.\"\"\"\n"," B, N, C = x.shape\n"," a = b = int(math.sqrt(N))\n"," x = tf.reshape(x, [-1, a, b, C])\n"," x = tf.signal.rfft2d(x)\n"," weight = tf.dtypes.complex(self.complex_weight[:, :, :, 0], self.complex_weight[:, :, :, -1])\n"," x = x * weight\n"," x = tf.signal.irfft2d(x)\n"," return tf.reshape(x, [-1, N, C])\n","\n"," def get_config(self, **kwargs): # FIX: was incorrectly named `config`; corrected to `get_config`\n"," \"\"\"Return serializable Keras configuration.\"\"\"\n"," config = super(GF_GlobalFilter, self).get_config()\n"," config.update({\"patch_size\": self.patch_size, \"dim\": self.dim})\n"," return config\n","\n","\n","class GF_Block(tf.keras.layers.Layer):\n"," \"\"\"One GFNet block: LayerNorm → GlobalFilter → DropPath residual → LayerNorm → MLP → DropPath residual.\"\"\"\n","\n"," def __init__(self, patch_size=3, dim=512, mlp_ratio=4.0, drop=0.0, drop_path=0.0, **kwargs):\n"," \"\"\"Initialize GFNet block with filter and MLP hyperparameters.\"\"\"\n"," super(GF_Block, self).__init__(**kwargs)\n"," self.patch_size = patch_size\n"," self.dim = dim\n"," self.mlp_ratio = mlp_ratio\n"," self.drop = drop\n"," self.drop_path_rate = drop_path\n","\n"," self.norm1 = tf.keras.layers.LayerNormalization(axis=-1)\n"," self.filter = GF_GlobalFilter(patch_size, dim)\n"," self.drop_path = GF_DropPath(drop_path)\n"," self.norm2 = tf.keras.layers.LayerNormalization(axis=-1)\n"," mlp_hidden_dim = int(dim * mlp_ratio)\n"," self.mlp = GF_MLP(in_features=mlp_hidden_dim, out_features=dim, drop=drop)\n","\n"," def call(self, x):\n"," \"\"\"Single residual block: filter branch + MLP branch.\"\"\"\n"," x = x + self.drop_path(self.mlp(self.norm2(self.filter(self.norm1(x)))))\n"," return x\n","\n"," def get_config(self):\n"," \"\"\"Return serializable Keras configuration.\"\"\"\n"," config = super(GF_Block, self).get_config()\n"," config.update({\n"," \"patch_size\": self.patch_size,\n"," \"dim\": self.dim,\n"," \"mlp_ratio\": self.mlp_ratio,\n"," \"drop\": self.drop,\n"," \"drop_path\": self.drop_path_rate,\n"," })\n"," return config"]},{"cell_type":"markdown","id":"52-gfnet-builder-hdr","metadata":{"papermill":{"duration":0.007306,"end_time":"2026-09-12T01:51:53.434979+00:00","exception":false,"start_time":"2026-09-12T01:51:53.427673+00:00","status":"completed"},"tags":[]},"source":["## 5.2 — GFNet Model Builder\n","\n","Assembles patch extraction, encoder, `GlobalFilter_layers` GFNet blocks, pooling,\n","and `K_HEADS = 7` softmax output heads."]},{"cell_type":"code","execution_count":11,"id":"52-gfnet-builder-code","metadata":{"execution":{"iopub.execute_input":"2026-09-12T01:51:53.450971Z","iopub.status.busy":"2026-09-12T01:51:53.450157Z","iopub.status.idle":"2026-09-12T01:51:53.458372Z","shell.execute_reply":"2026-09-12T01:51:53.457759Z"},"papermill":{"duration":0.018166,"end_time":"2026-09-12T01:51:53.460007+00:00","exception":false,"start_time":"2026-09-12T01:51:53.441841+00:00","status":"completed"},"tags":[]},"outputs":[],"source":["def GFNet(input_shape=(P_S, P_S, B),\n"," use_pearson_corr=False,\n"," patch_size=3,\n"," num_classes=16,\n"," hidden_dim=512,\n"," GlobalFilter_layers=12,\n"," mlp_ratio=4,\n"," num_patches=9,\n"," dropout_rate=0.0,\n"," drop_path_rate=0.0):\n"," \"\"\"Build the multi-head GFNet classifier.\"\"\"\n"," K_HEADS = 7 # number of independent output heads\n","\n"," x_input = Input(shape=input_shape)\n"," x = Pearson_correlation_masked(P_S)(x_input) if use_pearson_corr else x_input\n","\n"," # ── Patch extraction and encoding ────────────────────────────────────\n"," x = GF_Patches(patch_size)(x)\n"," x = GF_PatchEncoder(num_patches, hidden_dim)(x)\n"," x = Dropout(dropout_rate, name=\"TRAIN_DROPOUT_1\")(x)\n","\n"," # ── Stacked Global Filter blocks ─────────────────────────────────────\n"," for _ in range(GlobalFilter_layers):\n"," x = GF_Block(\n"," patch_size=patch_size,\n"," dim=hidden_dim,\n"," mlp_ratio=mlp_ratio,\n"," drop=dropout_rate,\n"," drop_path=drop_path_rate,\n"," )(x)\n","\n"," # ── Pooling and normalisation ─────────────────────────────────────────\n"," x = Dropout(dropout_rate, name=\"TRAIN_DROPOUT_2\")(x)\n"," x = LayerNormalization()(x)\n"," x = GF_Expand_Dims(ndim=2)(x)\n"," x = GlobalAveragePooling2D()(x)\n"," x = Flatten()(x)\n"," x = Dropout(dropout_rate, name=\"TRAIN_DROPOUT_3\")(x)\n","\n"," # ── Multi-head outputs ─────────────────────────────────────────────────\n"," output_heads = [\n"," Dense(num_classes, activation=\"softmax\", dtype=\"float32\", name=f\"head_{i+1}\")(x)\n"," for i in range(K_HEADS)\n"," ]\n"," return keras.Model(inputs=x_input, outputs=output_heads, name=\"MultiHead_GFNet\")"]},{"cell_type":"markdown","id":"60-vit-hdr","metadata":{"papermill":{"duration":0.007049,"end_time":"2026-09-12T01:51:53.474159+00:00","exception":false,"start_time":"2026-09-12T01:51:53.46711+00:00","status":"completed"},"tags":[]},"source":["# 6.0 — Vision Transformer Multi-Head Model\n","\n","A compact ViT with learnable CLS token, positional embeddings, and a symmetric\n","U-Net-style skip-connection transformer stack (`ViT_TransFormer_Block`).\n","Optional spatial-attention preprocessing modules are also defined here."]},{"cell_type":"markdown","id":"61-vit-layers-hdr","metadata":{"papermill":{"duration":0.007108,"end_time":"2026-09-12T01:51:53.488464+00:00","exception":false,"start_time":"2026-09-12T01:51:53.481356+00:00","status":"completed"},"tags":[]},"source":["## 6.1 — ViT Layer Classes\n","\n","`ViT_SpatialAttention`, `ViT_SpatialAttention1`, `MLP` (functional),\n","`ViT_Patches`, `ViT_PatchEncoder` (with CLS token), `ViT_Weighted_add`,\n","`ViT_TransFormer`, `ViT_TransFormer_Block` (symmetric skips), and\n","`ViT_Class_Token_Norm`."]},{"cell_type":"code","execution_count":12,"id":"61-vit-layers-code","metadata":{"execution":{"iopub.execute_input":"2026-09-12T01:51:53.505065Z","iopub.status.busy":"2026-09-12T01:51:53.504417Z","iopub.status.idle":"2026-09-12T01:51:53.538876Z","shell.execute_reply":"2026-09-12T01:51:53.537986Z"},"papermill":{"duration":0.044898,"end_time":"2026-09-12T01:51:53.540474+00:00","exception":false,"start_time":"2026-09-12T01:51:53.495576+00:00","status":"completed"},"tags":[]},"outputs":[],"source":["class ViT_SpatialAttention(layers.Layer):\n"," \"\"\"Four-conv spatial attention gate producing a sigmoid attention map.\"\"\"\n","\n"," def __init__(self, k_size=3, **kwargs):\n"," \"\"\"Initialize with convolution kernel size.\"\"\"\n"," super().__init__(**kwargs)\n"," self.k_size = k_size\n"," self.norm = layers.BatchNormalization()\n"," self.conv1 = layers.Conv2D(1, kernel_size=(k_size, k_size), padding=\"same\")\n"," self.conv2 = layers.Conv2D(1, kernel_size=(k_size, k_size), padding=\"same\")\n"," self.conv3 = layers.Conv2D(1, kernel_size=(k_size, k_size), padding=\"same\")\n"," self.conv4 = layers.Conv2D(1, kernel_size=(k_size, k_size), padding=\"same\")\n"," self.act_relu = layers.Activation(\"relu\")\n"," self.act_sigmoid = layers.Activation(\"sigmoid\")\n","\n"," def call(self, inputs):\n"," \"\"\"Pass through four convolutions with ReLU; final sigmoid gate.\"\"\"\n"," x = self.conv1(inputs)\n"," x = self.norm(x)\n"," x = self.conv2(x)\n"," x = self.act_relu(x)\n"," x = self.conv3(x)\n"," x = self.act_relu(x)\n"," x = self.conv4(x)\n"," return self.act_sigmoid(x)\n","\n"," def get_config(self):\n"," \"\"\"Return serializable Keras configuration.\"\"\"\n"," config = super().get_config()\n"," config.update({\"k_size\": self.k_size})\n"," return config\n","\n","\n","class ViT_SpatialAttention1(layers.Layer):\n"," \"\"\"Encoder-decoder spatial attention gate with Conv2D + Conv2DTranspose.\"\"\"\n","\n"," def __init__(self, input_shape, **kwargs):\n"," \"\"\"Initialize with expected input spatial shape for shape correction.\"\"\"\n"," super().__init__(**kwargs)\n"," self.input_shape_val = input_shape\n"," self.filters = input_shape[-1]\n"," self.k_size = input_shape[1]\n","\n"," self.norm = layers.BatchNormalization()\n"," self.conv1 = layers.Conv2D(self.filters, kernel_size=3, padding=\"same\", kernel_initializer=\"he_normal\")\n"," self.conv2 = layers.Conv2D(self.filters, kernel_size=3, strides=2, padding=\"same\")\n"," self.conv3 = layers.Conv2D(self.filters, kernel_size=3, strides=2, padding=\"same\")\n"," self.convt1 = layers.Conv2DTranspose(self.filters, kernel_size=3, strides=2, padding=\"same\")\n"," self.convt2 = layers.Conv2DTranspose(self.filters, kernel_size=3, strides=2, padding=\"same\")\n"," self.relu = layers.ReLU()\n"," self.sigmoid = layers.Activation(\"sigmoid\")\n","\n"," def call(self, inputs):\n"," \"\"\"Encoder–decoder path with shape correction and sigmoid gate.\"\"\"\n"," x = self.relu(self.norm(self.conv1(inputs)))\n"," x = self.relu(self.conv2(x))\n"," x = self.relu(self.conv3(x))\n"," x = self.relu(self.convt1(x))\n"," x = self.relu(self.convt2(x))\n"," # Shape correction if stride misalignment occurs\n"," if x.shape[1] != self.input_shape_val[1] or x.shape[2] != self.input_shape_val[2]:\n"," k_size = x.shape[1] - self.k_size + 1\n"," x = layers.Conv2D(self.filters, kernel_size=k_size, strides=1, padding=\"valid\")(x)\n"," return self.sigmoid(x)\n","\n"," def get_config(self):\n"," \"\"\"Return serializable Keras configuration.\"\"\"\n"," config = super().get_config()\n"," config.update({\"input_shape\": self.input_shape_val, \"filters\": self.filters, \"k_size\": self.k_size})\n"," return config\n","\n","\n","def MLP(x, hidden_units, dropout_rate):\n"," \"\"\"Feedforward network used inside Transformer blocks.\n","\n"," hidden_units: list of Dense layer sizes (e.g. [mlp_dim, embed_dim])\n"," dropout_rate: dropout applied after each Dense layer\n"," \"\"\"\n"," for units in hidden_units:\n"," x = layers.Dense(units, activation=tf.keras.activations.gelu)(x)\n"," x = layers.Dropout(dropout_rate)(x)\n"," return x\n","\n","\n","class ViT_Patches(layers.Layer):\n"," \"\"\"Extract non-overlapping patches and project to `embed_dim`.\"\"\"\n","\n"," def __init__(self, patch_size, embed_dim=768, **kwargs):\n"," \"\"\"Initialize with patch size and embedding dimension.\"\"\"\n"," super(ViT_Patches, self).__init__(**kwargs)\n"," self.patch_size = patch_size\n"," self.embed_dim = embed_dim\n","\n"," def build(self, input_shape):\n"," \"\"\"Build the linear projection layer.\"\"\"\n"," self.projection = layers.Dense(self.embed_dim)\n","\n"," def call(self, images):\n"," \"\"\"Extract patches via tf.image.extract_patches and project them.\"\"\"\n"," batch_size = tf.shape(images)[0]\n"," patches = tf.image.extract_patches(\n"," images=images,\n"," sizes=[1, self.patch_size, self.patch_size, 1],\n"," strides=[1, self.patch_size, self.patch_size, 1],\n"," rates=[1, 1, 1, 1],\n"," padding=\"VALID\",\n"," )\n"," patch_dims = patches.shape[-1] # = patch_size * patch_size * C\n"," patches = tf.reshape(patches, [batch_size, -1, patch_dims])\n"," return self.projection(patches) # project raw patches → embed_dim\n","\n"," def get_config(self):\n"," \"\"\"Return serializable Keras configuration.\"\"\"\n"," config = super(ViT_Patches, self).get_config()\n"," config.update({\"patch_size\": self.patch_size, \"embed_dim\": self.embed_dim})\n"," return config\n","\n","\n","class ViT_PatchEncoder(layers.Layer):\n"," \"\"\"Prepend a learnable CLS token and add positional embeddings to ViT patches.\"\"\"\n","\n"," def __init__(self, num_patches, projection_dim, **kwargs):\n"," \"\"\"Initialize with patch count and projection dimensionality.\"\"\"\n"," super(ViT_PatchEncoder, self).__init__(**kwargs)\n"," self.num_patches = num_patches\n"," self.projection_dim = projection_dim\n"," self.projection = layers.Dense(units=projection_dim)\n"," self.position_embedding = layers.Embedding(input_dim=num_patches + 1, output_dim=projection_dim)\n"," # Learnable CLS token initialised to zeros\n"," self.cls_token = self.add_weight(\n"," name=\"cls_token\",\n"," shape=(1, 1, projection_dim),\n"," initializer=tf.zeros_initializer(),\n"," trainable=True,\n"," )\n","\n"," def call(self, patch, **kwargs):\n"," \"\"\"Prepend CLS token, project patches, and add positional embeddings.\"\"\"\n"," batch_size = tf.shape(patch)[0]\n"," patch_proj = self.projection(patch) # [B, N, D]\n"," cls_tokens = tf.repeat(self.cls_token, repeats=batch_size, axis=0) # [B, 1, D]\n"," x = tf.concat([cls_tokens, patch_proj], axis=1) # [B, N+1, D]\n"," positions = tf.range(start=0, limit=self.num_patches + 1, delta=1)\n"," return x + self.position_embedding(positions)\n","\n"," def get_config(self):\n"," \"\"\"Return serializable Keras configuration.\"\"\"\n"," config = super(ViT_PatchEncoder, self).get_config()\n"," config.update({\"num_patches\": self.num_patches, \"projection_dim\": self.projection_dim})\n"," return config\n","\n","\n","class ViT_Weighted_add(layers.Layer):\n"," \"\"\"Combine two tensors with a single learned scalar weight `w`: w*x + (1-w)*y.\"\"\"\n","\n"," def __init__(self, name, **kwargs):\n"," \"\"\"Initialize with a unique identifier for the weight.\"\"\"\n"," super(ViT_Weighted_add, self).__init__(**kwargs)\n"," self.wt_name = name\n","\n"," def build(self, input_shape):\n"," \"\"\"Create the scalar blend weight.\"\"\"\n"," self.w = self.add_weight(\n"," name=\"weighted_add_\" + str(self.wt_name),\n"," shape=(1,),\n"," initializer=tf.random_normal_initializer(),\n"," trainable=True,\n"," )\n","\n"," def call(self, input_1, input_2):\n"," \"\"\"Weighted residual blend.\"\"\"\n"," return input_1 * self.w + input_2 * (1.0 - self.w)\n","\n"," def get_config(self):\n"," \"\"\"Return serializable Keras configuration.\"\"\"\n"," config = super(ViT_Weighted_add, self).get_config()\n"," config.update({\"wt_name\": self.wt_name})\n"," return config\n","\n","\n","class ViT_TransFormer(layers.Layer):\n"," \"\"\"Single transformer block: Multi-Head Attention + FFN, both with learned-weight residuals.\"\"\"\n","\n"," def __init__(self, layer_num, num_heads, projection_dim, dropout=0.1, **kwargs):\n"," \"\"\"Initialize block with layer index, head count, projection dim, and dropout.\"\"\"\n"," super(ViT_TransFormer, self).__init__(**kwargs)\n"," self.layer_num = layer_num\n"," self.num_heads = num_heads\n"," self.projection_dim = projection_dim\n"," self.dropout = dropout\n","\n"," def build(self, input_shape):\n"," \"\"\"Build sub-layers (deferred so `layer_num` is in scope for naming).\"\"\"\n"," self.norm1 = layers.LayerNormalization(epsilon=1e-6, name=f\"ln1_{self.layer_num}\")\n"," self.norm2 = layers.LayerNormalization(epsilon=1e-6, name=f\"ln2_{self.layer_num}\")\n"," self.add1 = ViT_Weighted_add(f\"transformer_1_{self.layer_num}\")\n"," self.add2 = ViT_Weighted_add(f\"transformer_2_{self.layer_num}\")\n"," self.mha = layers.MultiHeadAttention(\n"," num_heads=self.num_heads,\n"," key_dim=self.projection_dim,\n"," dropout=self.dropout,\n"," name=f\"mha_{self.layer_num}\",\n"," )\n"," self.dense1 = layers.Dense(self.projection_dim * 2, activation=tf.keras.activations.gelu)\n"," self.drop1 = layers.Dropout(self.dropout)\n"," self.dense2 = layers.Dense(self.projection_dim, activation=tf.keras.activations.gelu)\n"," self.drop2 = layers.Dropout(self.dropout)\n","\n"," def call(self, inputs, training=None):\n"," \"\"\"MHA residual block then FFN residual block.\"\"\"\n"," # Multi-Head Attention\n"," x1 = self.norm1(inputs)\n"," x1 = self.mha(x1, x1, training=training)\n"," x1 = self.add1(x1, inputs)\n"," # Feed-Forward Network\n"," x2 = self.norm2(x1)\n"," x2 = self.drop1(self.dense1(x2), training=training)\n"," x2 = self.drop2(self.dense2(x2), training=training)\n"," return self.add2(x2, x1)\n","\n"," def get_config(self):\n"," \"\"\"Return serializable Keras configuration.\"\"\"\n"," config = super(ViT_TransFormer, self).get_config()\n"," config.update({\n"," \"layer_num\": self.layer_num, \"num_heads\": self.num_heads,\n"," \"projection_dim\": self.projection_dim, \"dropout\": self.dropout,\n"," })\n"," return config\n","\n","\n","class ViT_TransFormer_Block(layers.Layer):\n"," \"\"\"Stack of `num_layers` transformer blocks with symmetric U-Net-style skip connections.\"\"\"\n","\n"," def __init__(self, num_layers, num_heads, projection_dim, dropout=0.1, **kwargs):\n"," \"\"\"Initialize block stack with depth, head count, projection dim, and dropout.\"\"\"\n"," super(ViT_TransFormer_Block, self).__init__(**kwargs)\n"," self.num_layers = num_layers\n"," self.num_heads = num_heads\n"," self.projection_dim = projection_dim\n"," self.dropout = dropout\n","\n"," def build(self, input_shape):\n"," \"\"\"Build the list of TransFormer sub-layers.\"\"\"\n"," self.Blocks = [\n"," ViT_TransFormer(i, self.num_heads, self.projection_dim, self.dropout)\n"," for i in range(self.num_layers)\n"," ]\n","\n"," def call(self, inputs, training=None):\n"," \"\"\"Forward + symmetric skip connections (mirror like U-Net).\"\"\"\n"," block_list = []\n"," x = inputs\n"," for i in range(self.num_layers):\n"," x = self.Blocks[i](x, training=training)\n"," if i <= self.num_layers // 2:\n"," block_list.append(x)\n"," else:\n"," x = layers.Add()([x, block_list[self.num_layers - i - 1]])\n"," return x\n","\n"," def get_config(self):\n"," \"\"\"Return serializable Keras configuration.\"\"\"\n"," config = super(ViT_TransFormer_Block, self).get_config()\n"," config.update({\n"," \"num_layers\": self.num_layers, \"num_heads\": self.num_heads,\n"," \"projection_dim\": self.projection_dim, \"dropout\": self.dropout,\n"," })\n"," return config\n","\n","\n","class ViT_Class_Token_Norm(layers.Layer):\n"," \"\"\"Apply LayerNormalization and return only the CLS token (index 0).\"\"\"\n","\n"," def __init__(self, eps=1e-6, **kwargs):\n"," \"\"\"Initialize with LayerNorm epsilon.\"\"\"\n"," super(ViT_Class_Token_Norm, self).__init__(**kwargs)\n"," self.eps = eps\n"," self.norm = layers.LayerNormalization(epsilon=self.eps)\n","\n"," def call(self, inputs):\n"," \"\"\"Normalize sequence and extract CLS token → shape (B, D).\"\"\"\n"," x = self.norm(inputs)\n"," return x[:, 0, :] # CLS token\n","\n"," def get_config(self):\n"," \"\"\"Return serializable Keras configuration.\"\"\"\n"," config = super(ViT_Class_Token_Norm, self).get_config()\n"," config.update({\"eps\": self.eps})\n"," return config"]},{"cell_type":"markdown","id":"62-vit-builder-hdr","metadata":{"papermill":{"duration":0.007331,"end_time":"2026-09-12T01:51:53.555405+00:00","exception":false,"start_time":"2026-09-12T01:51:53.548074+00:00","status":"completed"},"tags":[]},"source":["## 6.2 — ViT Model Builder\n","\n","`create_vit_classifier` assembles the full pipeline. Three representation strategies\n","are supported via the `method` argument: `with_cls_tkn` (CLS token), `with_gap`\n","(global average pooling), and `without_gap` (flattened normalization).\n","\n","> **Note:** `projection_dim`, `num_heads`, `transformer_layers`, and `dropout` are\n","> referenced as free variables from the enclosing scope. The authoritative values are\n","> set in Section 7.0 (Training Configuration) below; the defaults here serve as\n","> documentation / fallbacks if Section 7.0 is not run first."]},{"cell_type":"code","execution_count":13,"id":"62-vit-builder-code","metadata":{"execution":{"iopub.execute_input":"2026-09-12T01:51:53.571735Z","iopub.status.busy":"2026-09-12T01:51:53.571351Z","iopub.status.idle":"2026-09-12T01:51:53.581441Z","shell.execute_reply":"2026-09-12T01:51:53.58073Z"},"papermill":{"duration":0.020208,"end_time":"2026-09-12T01:51:53.583001+00:00","exception":false,"start_time":"2026-09-12T01:51:53.562793+00:00","status":"completed"},"tags":[]},"outputs":[],"source":["# Default ViT hyperparameters — overridden by Section 7.0\n","projection_dim = 256\n","num_heads = 4\n","transformer_layers = 12\n","dropout = 0.1 # dropout inside Transformer blocks\n","\n","\n","def create_vit_classifier(input_shape=(P_S, P_S, B),\n"," num_classes=7,\n"," use_pearson_corr=False,\n"," dropout_rate=0.25,\n"," method=\"with_gap\",\n"," k_heads: int = 1):\n"," \"\"\"Build a Vision Transformer classifier with optional multi-head outputs.\n","\n"," k_heads == 1: returns a single logits tensor (shape [B, num_classes]).\n"," k_heads > 1: returns a list of k_heads tensors [ (B, num_classes), ... ].\n"," method: 'with_cls_tkn' | 'with_gap' | 'without_gap'\n"," \"\"\"\n"," inputs = layers.Input(shape=input_shape)\n"," x0 = Pearson_correlation_masked(P_S)(inputs) if use_pearson_corr else inputs\n","\n"," # ── Patch creation & encoding ──────────────────────────────────────\n"," patches = ViT_Patches(patch_size, embed_dim=projection_dim)(x0)\n"," encoded_patches = ViT_PatchEncoder(num_patches, projection_dim)(patches)\n","\n"," # ── Transformer stack ──────────────────────────────────────────────\n"," encoded_patches = ViT_TransFormer_Block(transformer_layers, num_heads, projection_dim, dropout)(encoded_patches)\n"," encoded_patches = layers.Dropout(dropout_rate, name=\"TRAIN_DROPOUT_1\")(encoded_patches)\n","\n"," # ── Representation strategy ────────────────────────────────────────\n"," if method == \"with_cls_tkn\":\n"," representation = ViT_Class_Token_Norm(eps=1e-6)(encoded_patches)\n"," elif method == \"without_gap\":\n"," representation = layers.LayerNormalization(epsilon=1e-6)(encoded_patches)\n"," representation = layers.Flatten()(representation)\n"," elif method == \"with_gap\":\n"," representation = layers.LayerNormalization(epsilon=1e-6)(encoded_patches)\n"," representation = tf.reduce_mean(representation, axis=1) # [B, D]\n"," else:\n"," raise ValueError(f\"Unknown method: {method}\")\n","\n"," # ── Classification MLP ─────────────────────────────────────────────\n"," x = layers.Dense(512, activation=tf.keras.activations.gelu)(representation)\n"," x = layers.Dropout(dropout_rate, name=\"TRAIN_DROPOUT_3\")(x)\n"," x = layers.Dense(256, activation=tf.keras.activations.gelu)(x)\n"," x = layers.Dense(128, activation=tf.keras.activations.gelu)(x)\n"," x = layers.Dropout(dropout_rate, name=\"TRAIN_DROPOUT_5\")(x)\n"," x = layers.Dense(64, activation=tf.keras.activations.gelu)(x)\n"," features = layers.Dropout(dropout_rate, name=\"TRAIN_DROPOUT_6\")(x)\n","\n"," # ── Multi-head outputs ──────────────────────────────────────────────\n"," output_heads = [\n"," layers.Dense(num_classes, activation=\"softmax\", dtype=\"float32\", name=f\"head_{i+1}\")(features)\n"," for i in range(max(1, k_heads))\n"," ]\n","\n"," outputs = output_heads[0] if k_heads == 1 else output_heads\n"," return keras.Model(inputs=inputs, outputs=outputs)"]},{"cell_type":"markdown","id":"70-config-hdr","metadata":{"papermill":{"duration":0.006849,"end_time":"2026-09-12T01:51:53.597257+00:00","exception":false,"start_time":"2026-09-12T01:51:53.590408+00:00","status":"completed"},"tags":[]},"source":["# 7.0 — Training Configuration\n","\n","Authoritative architecture hyperparameters (overriding any defaults set above),\n","the `CUSTOM_OBJECTS` dict required for model deserialisation, and `MODEL_SPECS` —\n","a registry mapping each model name to its builder, optimiser, LR schedule, and\n","output paths."]},{"cell_type":"code","execution_count":14,"id":"70-config-code","metadata":{"execution":{"iopub.execute_input":"2026-09-12T01:51:53.619898Z","iopub.status.busy":"2026-09-12T01:51:53.619466Z","iopub.status.idle":"2026-09-12T01:51:53.629389Z","shell.execute_reply":"2026-09-12T01:51:53.628798Z"},"papermill":{"duration":0.020277,"end_time":"2026-09-12T01:51:53.631144+00:00","exception":false,"start_time":"2026-09-12T01:51:53.610867+00:00","status":"completed"},"tags":[]},"outputs":[],"source":["# ── Architecture hyperparameters (authoritative values) ────────────────────\n","# NOTE: projection_dim, num_heads, transformer_layers, dropout were also set in\n","# Section 6.2 as defaults; the values here take precedence at call time.\n","patch_size = 3\n","hidden_dim = 512\n","GlobalFilter_layers = 12\n","mlp_ratio = 4\n","num_patches = 9\n","projection_dim = 256\n","num_heads = 4\n","transformer_layers = 12\n","dropout = 0.1\n","\n","# ── Custom-object registry for model deserialisation ──────────────────────\n","CUSTOM_OBJECTS = {\n"," 'Pearson_correlation_masked': Pearson_correlation_masked,\n"," 'Dropout_Train': Dropout_Train,\n"," 'GF_Patches': GF_Patches,\n"," 'GF_PatchEncoder': GF_PatchEncoder,\n"," 'GF_GlobalFilter': GF_GlobalFilter,\n"," 'GF_Block': GF_Block,\n"," 'GF_Expand_Dims': GF_Expand_Dims,\n"," 'GF_MLP': GF_MLP,\n"," 'GF_DropPath': GF_DropPath,\n"," 'ViT_Patches': ViT_Patches,\n"," 'ViT_PatchEncoder': ViT_PatchEncoder,\n"," 'ViT_SpatialAttention': ViT_SpatialAttention,\n"," 'ViT_SpatialAttention1': ViT_SpatialAttention1,\n"," 'ViT_Weighted_add': ViT_Weighted_add,\n"," 'ViT_TransFormer': ViT_TransFormer,\n"," 'ViT_TransFormer_Block': ViT_TransFormer_Block,\n"," 'ViT_Class_Token_Norm': ViT_Class_Token_Norm,\n","}\n","\n","# ── Per-model training specifications ─────────────────────────────────────\n","MODEL_SPECS = {\n"," 'AlexNet_CNN_MultiHead': {\n"," 'builder': lambda: AlexNet(input_shape=(P_S, P_S, B), num_classes=num_classes,\n"," use_pearson_corr=use_pearson_corr, dropout_rate=dropout_rate),\n"," 'optimizer': lambda: optimizers.Adagrad(0.01),\n"," 'lr_schedule': 'alexnet',\n"," 'model_file': SAVED_MODEL_DIR / 'AlexNet_CNN_MultiHead_best.keras',\n"," 'final_file': SAVED_MODEL_DIR / 'AlexNet_CNN_MultiHead_final.keras',\n"," },\n"," 'GFNet_MultiHead': {\n"," 'builder': lambda: GFNet((P_S, P_S, B), use_pearson_corr, patch_size, num_classes,\n"," hidden_dim, GlobalFilter_layers, mlp_ratio, num_patches, dropout_rate, 0.0),\n"," 'optimizer': lambda: tf.keras.optimizers.Adam(learning_rate=0.000003),\n"," 'lr_schedule': 'gfnet',\n"," 'model_file': SAVED_MODEL_DIR / 'GFNet_MultiHead_best.keras',\n"," 'final_file': SAVED_MODEL_DIR / 'GFNet_MultiHead_final.keras',\n"," },\n"," 'ViT_UNet_MultiHead': {\n"," 'builder': lambda: create_vit_classifier(input_shape=(P_S, P_S, B), num_classes=num_classes,\n"," use_pearson_corr=use_pearson_corr, dropout_rate=dropout_rate,\n"," method='with_cls_tkn', k_heads=7),\n"," 'optimizer': lambda: optimizers.Adam(3e-6),\n"," 'lr_schedule': 'vit',\n"," 'model_file': SAVED_MODEL_DIR / 'ViT_UNet_MultiHead_best.keras',\n"," 'final_file': SAVED_MODEL_DIR / 'ViT_UNet_MultiHead_final.keras',\n"," },\n","}"]},{"cell_type":"markdown","id":"80-helpers-hdr","metadata":{"papermill":{"duration":0.007051,"end_time":"2026-09-12T01:51:53.645446+00:00","exception":false,"start_time":"2026-09-12T01:51:53.638395+00:00","status":"completed"},"tags":[]},"source":["# 8.0 — Training and Export Helpers\n","\n","Utility functions for workbook management, learning-rate scheduling, multi-head\n","prediction aggregation, performance metrics export, and the reusable training runner."]},{"cell_type":"markdown","id":"81-excel-hdr","metadata":{"papermill":{"duration":0.00728,"end_time":"2026-09-12T01:51:53.660274+00:00","exception":false,"start_time":"2026-09-12T01:51:53.652994+00:00","status":"completed"},"tags":[]},"source":["## 8.1 — Excel Export Utilities\n","\n","`ensure_workbook` creates or opens the results workbook.\n","`autosize_columns` adjusts column widths.\n","`fig_to_buffer` serialises Matplotlib figures to PNG buffers for embedding.\n","`export_training_sheet` writes summary rows, classification reports, confusion matrices,\n","and three figures into a dedicated worksheet."]},{"cell_type":"code","execution_count":15,"id":"81-excel-code","metadata":{"execution":{"iopub.execute_input":"2026-09-12T01:51:53.676723Z","iopub.status.busy":"2026-09-12T01:51:53.676161Z","iopub.status.idle":"2026-09-12T01:51:53.687015Z","shell.execute_reply":"2026-09-12T01:51:53.686239Z"},"papermill":{"duration":0.021224,"end_time":"2026-09-12T01:51:53.688619+00:00","exception":false,"start_time":"2026-09-12T01:51:53.667395+00:00","status":"completed"},"tags":[]},"outputs":[],"source":["def ensure_workbook(path):\n"," \"\"\"Create a new workbook with a Summary sheet, or load an existing one.\"\"\"\n"," if path.exists():\n"," return load_workbook(path)\n"," wb = Workbook()\n"," ws = wb.active\n"," ws.title = 'Summary'\n"," wb.save(path)\n"," return wb\n","\n","\n","def autosize_columns(ws):\n"," \"\"\"Resize each worksheet column to fit its widest cell value (max 40 chars).\"\"\"\n"," for col in ws.columns:\n"," values = [len(str(cell.value)) for cell in col if cell.value is not None]\n"," if values:\n"," ws.column_dimensions[col[0].column_letter].width = min(max(values) + 2, 40)\n","\n","\n","def fig_to_buffer(fig):\n"," \"\"\"Serialise a Matplotlib figure to an in-memory PNG BytesIO buffer.\"\"\"\n"," buf = io.BytesIO()\n"," fig.savefig(buf, format='png', bbox_inches='tight', dpi=200)\n"," buf.seek(0)\n"," return buf\n","\n","\n","def export_training_sheet(wb, model_name, summary_row, report_df, cm, figs):\n"," \"\"\"Write metrics, classification report, confusion matrix, and figures to a new Excel sheet.\"\"\"\n"," ws = wb.create_sheet(title=model_name[:31])\n","\n"," # Summary key-value rows\n"," for idx, (key, value) in enumerate(summary_row.items(), start=1):\n"," ws.cell(row=idx, column=1, value=key)\n"," ws.cell(row=idx, column=2, value=value)\n","\n"," row0 = len(summary_row) + 3\n","\n"," # Classification report table\n"," for r_idx, row in enumerate(dataframe_to_rows(report_df.reset_index(), index=False, header=True), start=row0):\n"," for c_idx, val in enumerate(row, start=1):\n"," ws.cell(row=r_idx, column=c_idx, value=val)\n","\n"," # Confusion matrix (offset to column 10)\n"," for r_idx, row in enumerate(dataframe_to_rows(pd.DataFrame(cm), index=False, header=False), start=row0):\n"," for c_idx, val in enumerate(row, start=10):\n"," ws.cell(row=r_idx, column=c_idx, value=int(val))\n","\n"," # Embedded figures: curve, performance, LR schedule\n"," for anchor, fig in [('L2', figs[0]), ('L30', figs[1]), ('L58', figs[2])]:\n"," img = XLImage(fig_to_buffer(fig))\n"," img.anchor = anchor\n"," ws.add_image(img)\n","\n"," autosize_columns(ws)"]},{"cell_type":"markdown","id":"82-lr-hdr","metadata":{"papermill":{"duration":0.006923,"end_time":"2026-09-12T01:51:53.702818+00:00","exception":false,"start_time":"2026-09-12T01:51:53.695895+00:00","status":"completed"},"tags":[]},"source":["## 8.2 — Learning Rate Scheduling\n","\n","`build_lr_callback` builds a cosine-annealing `LearningRateScheduler` and plots the\n","schedule. The AlexNet and GFNet variants use a **three-stage** multi-step cosine\n","decay; the ViT variant uses a single-stage cosine decay.\n","\n","> **DRY refactor:** the identical multi-step cosine decay logic that was duplicated\n","> in the AlexNet and GFNet branches has been extracted into `_multistep_cosine_lrfn`.\n","> `LR_START` (which was assigned but never read in both branches) has been removed."]},{"cell_type":"code","execution_count":16,"id":"82-lr-code","metadata":{"execution":{"iopub.execute_input":"2026-09-12T01:51:53.719111Z","iopub.status.busy":"2026-09-12T01:51:53.718613Z","iopub.status.idle":"2026-09-12T01:51:53.727962Z","shell.execute_reply":"2026-09-12T01:51:53.727167Z"},"papermill":{"duration":0.019419,"end_time":"2026-09-12T01:51:53.729557+00:00","exception":false,"start_time":"2026-09-12T01:51:53.710138+00:00","status":"completed"},"tags":[]},"outputs":[],"source":["def _multistep_cosine_lrfn(e, steps, lr_max, lr_min):\n"," \"\"\"Cosine-decay LR for a three-stage schedule.\n","\n"," # DRY: consolidated repeated cosine decay logic from alexnet and gfnet branches\n"," stages: [0, steps[0]) → stage 1, [steps[0], steps[0]+steps[1]) → stage 2, else → stage 3\n"," \"\"\"\n"," if e < steps[0]:\n"," epoch2, epochs2 = e, steps[0]\n"," elif e < steps[0] + steps[1]:\n"," epoch2, epochs2 = e - steps[0], steps[1]\n"," else:\n"," epoch2, epochs2 = e - steps[0] - steps[1], steps[2]\n"," phase = math.pi * epoch2 / (epochs2 - 1)\n"," return (lr_max - lr_min) * 0.5 * (1.0 + math.cos(phase)) + lr_min\n","\n","\n","def build_lr_callback(kind):\n"," \"\"\"Build a LearningRateScheduler and return (callback, figure).\"\"\"\n"," if kind == 'alexnet':\n"," LR_MAX, LR_MIN = 0.02, 0.005\n"," STEPS = [epoch, epoch * 2, epoch * 3]\n"," lrfn = lambda e: _multistep_cosine_lrfn(e, STEPS, LR_MAX, LR_MIN) # DRY\n","\n"," elif kind == 'gfnet':\n"," LR_MAX, LR_MIN = 6e-4, 1e-7\n"," STEPS = [epoch, epoch * 2, epoch * 3]\n"," lrfn = lambda e: _multistep_cosine_lrfn(e, STEPS, LR_MAX, LR_MIN) # DRY\n","\n"," else: # 'vit' and any other schedule: single-stage cosine decay\n"," LR_MAX, LR_MIN = 6e-4, 1e-7\n"," def lrfn(e):\n"," \"\"\"Single-stage cosine-decay LR schedule.\"\"\"\n"," phase = math.pi * e / (epoch - 1)\n"," return (LR_MAX - LR_MIN) * 0.5 * (1.0 + math.cos(phase)) + LR_MIN\n","\n"," rng = list(range(epoch))\n"," lr_y = [lrfn(x) for x in rng]\n"," fig = plt.figure(figsize=(10, 4))\n"," plt.plot(rng, lr_y, '-o')\n"," plt.xlabel('Epoch', size=14)\n"," plt.ylabel('Learning Rate', size=14)\n"," plt.title(f'Learning rate schedule: {kind}')\n"," plt.show()\n"," return tf.keras.callbacks.LearningRateScheduler(lrfn, verbose=True), fig"]},{"cell_type":"markdown","id":"83-pred-hdr","metadata":{"papermill":{"duration":0.008766,"end_time":"2026-09-12T01:51:53.745886+00:00","exception":false,"start_time":"2026-09-12T01:51:53.73712+00:00","status":"completed"},"tags":[]},"source":["## 8.3 — Prediction & Metric Helpers\n","\n","`predict_multihead` averages probability distributions across all K heads and returns\n","argmax class predictions. `parameter_summary` counts total / trainable / non-trainable\n","parameters. `history_figure` plots head-1 accuracy and overall loss curves."]},{"cell_type":"code","execution_count":17,"id":"83-pred-code","metadata":{"execution":{"iopub.execute_input":"2026-09-12T01:51:53.762296Z","iopub.status.busy":"2026-09-12T01:51:53.761864Z","iopub.status.idle":"2026-09-12T01:51:53.770997Z","shell.execute_reply":"2026-09-12T01:51:53.770407Z"},"papermill":{"duration":0.019328,"end_time":"2026-09-12T01:51:53.772594+00:00","exception":false,"start_time":"2026-09-12T01:51:53.753266+00:00","status":"completed"},"tags":[]},"outputs":[],"source":["def predict_multihead(model, x_data):\n"," \"\"\"Average multi-head probabilities and return (argmax labels, averaged probs).\"\"\"\n"," y_pred_list = model.predict(x_data, verbose=0)\n"," y_pred_stacked = np.stack(y_pred_list, axis=0)\n"," y_pred_avg_probs = np.mean(y_pred_stacked, axis=0)\n"," y_pred_argmax = np.argmax(y_pred_avg_probs, axis=1)\n"," return y_pred_argmax.reshape(-1, 1), y_pred_avg_probs\n","\n","\n","def parameter_summary(model):\n"," \"\"\"Return (total, trainable, non_trainable) parameter counts for a Keras model.\"\"\"\n"," trainable = int(np.sum([np.prod(v.shape) for v in model.trainable_weights]))\n"," non_trainable = int(np.sum([np.prod(v.shape) for v in model.non_trainable_weights]))\n"," return int(trainable + non_trainable), trainable, non_trainable\n","\n","\n","def history_figure(history, model_name):\n"," \"\"\"Plot head-1 train/val accuracy and overall train/val loss; return the figure.\"\"\"\n"," fig = plt.figure(figsize=(24, 8))\n"," ax = plt.subplot(1, 1, 1)\n"," ax2 = ax.twinx()\n","\n"," ax.plot(history.history.get('head_1_accuracy', []), color='blue', marker='o', linewidth=1.5, markersize=2, label='train_head_1_accuracy')\n"," ax.plot(history.history.get('val_head_1_accuracy', []), color='green', marker='o', linewidth=1.5, markersize=2, label='val_head_1_accuracy')\n"," ax.grid()\n"," ax.legend()\n","\n"," ax2.plot(history.history.get('loss', []), color='black', marker='o', linewidth=1.5, markersize=2, label='train_loss')\n"," ax2.plot(history.history.get('val_loss', []), color='red', marker='o', linewidth=1.5, markersize=2, label='val_loss')\n"," ax2.legend()\n","\n"," plt.title(f'accuracy and loss plot for {model_name}')\n"," plt.show()\n"," return fig"]},{"cell_type":"markdown","id":"84-runner-hdr","metadata":{"papermill":{"duration":0.007438,"end_time":"2026-09-12T01:51:53.787501+00:00","exception":false,"start_time":"2026-09-12T01:51:53.780063+00:00","status":"completed"},"tags":[]},"source":["## 8.4 — Training Runner\n","\n","`run_training` orchestrates the full cycle for a single model: build → compile →\n","fit → reload best checkpoint → evaluate → export to Excel. Returns a summary dict\n","for the master summary sheet."]},{"cell_type":"code","execution_count":18,"id":"84-runner-code","metadata":{"execution":{"iopub.execute_input":"2026-09-12T01:51:53.804189Z","iopub.status.busy":"2026-09-12T01:51:53.803663Z","iopub.status.idle":"2026-09-12T01:51:53.817281Z","shell.execute_reply":"2026-09-12T01:51:53.816612Z"},"papermill":{"duration":0.024132,"end_time":"2026-09-12T01:51:53.818993+00:00","exception":false,"start_time":"2026-09-12T01:51:53.794861+00:00","status":"completed"},"tags":[]},"outputs":[],"source":["def run_training(model_name, spec, workbook):\n"," \"\"\"Train, evaluate, save, and export results for one configured model.\"\"\"\n"," tf.keras.backend.clear_session()\n"," gc.collect()\n","\n"," model = spec['builder']()\n"," lr_callback, lr_fig = build_lr_callback(spec['lr_schedule'])\n"," checkpoint = ModelCheckpoint(\n"," spec['model_file'],\n"," monitor='val_head_1_accuracy',\n"," verbose=1,\n"," save_best_only=True,\n"," save_weights_only=False,\n"," mode='max',\n"," )\n"," model.compile(\n"," optimizer=spec['optimizer'](),\n"," loss=['sparse_categorical_crossentropy'] * 7,\n"," metrics=['accuracy'] * 7,\n"," )\n","\n"," start_time = time.time()\n"," history = model.fit(\n"," x_train, [y_train] * 7,\n"," validation_data=(x_test, [y_test] * 7),\n"," validation_freq=1,\n"," batch_size=BATCH_SIZE,\n"," callbacks=[checkpoint, lr_callback],\n"," shuffle=False,\n"," epochs=epoch,\n"," )\n"," tt = time.time() - start_time\n","\n"," # Reload best checkpoint and copy to final path (.keras)\n"," reloaded = load_model(spec['model_file'], custom_objects=CUSTOM_OBJECTS, safe_mode=False)\n"," shutil.copy2(spec['model_file'], spec['final_file'])\n","\n"," # Save both .keras and .h5 models for all 3 models (best and final)\n"," best_h5_file = Path(str(spec['model_file']).replace('.keras', '.h5'))\n"," final_h5_file = Path(str(spec['final_file']).replace('.keras', '.h5'))\n"," try:\n"," reloaded.save(best_h5_file)\n"," shutil.copy2(best_h5_file, final_h5_file)\n"," print(f\"Saved .h5 models: {best_h5_file.name}, {final_h5_file.name}\")\n"," except Exception as e:\n"," print(f\"Warning: Could not save .h5 model: {e}\")\n","\n"," # Also make flat copies directly in OUTPUT_ROOT (/kaggle/working/results/) for seamless retrieval\n"," if SAVED_MODEL_DIR != OUTPUT_ROOT:\n"," for m_path in [spec['model_file'], spec['final_file'], best_h5_file, final_h5_file]:\n"," if Path(m_path).exists():\n"," try:\n"," shutil.copy2(m_path, OUTPUT_ROOT / Path(m_path).name)\n"," except Exception as e:\n"," print(f\"Warning: Could not make flat copy of {m_path}: {e}\")\n","\n"," # Evaluate\n"," y_pred, _ = predict_multihead(reloaded, x_test)\n"," total_params, trainable_params, non_trainable_params = parameter_summary(reloaded)\n"," accuracy = float(accuracy_score(y_test, y_pred))\n"," kappa = float(cohen_kappa_score(y_test, y_pred))\n"," cm = confusion_matrix(y_test, y_pred)\n"," report_df = pd.DataFrame(classification_report(y_test, y_pred, output_dict=True)).T\n","\n"," # Generate figures (performance_meausures returns fig directly)\n"," curve_fig = history_figure(history, model_name)\n"," perf_fig = performance_meausures(\n"," y_test, y_pred, tt,\n"," total_params, trainable_params, non_trainable_params,\n"," folder_path=folder_path,\n"," )\n","\n"," # Save training plots directly into OUTPUT_ROOT (/kaggle/working/results/)\n"," try:\n"," curve_fig.savefig(OUTPUT_ROOT / f\"{model_name}_accuracy_loss_curve.png\", bbox_inches='tight')\n"," perf_fig.savefig(OUTPUT_ROOT / f\"{model_name}_performance_measures.png\", bbox_inches='tight')\n"," if lr_fig is not None:\n"," lr_fig.savefig(OUTPUT_ROOT / f\"{model_name}_lr_schedule.png\", bbox_inches='tight')\n"," print(f\"Saved training plots directly to {OUTPUT_ROOT}\")\n"," except Exception as e:\n"," print(f\"Warning: Could not save training plot to {OUTPUT_ROOT}: {e}\")\n","\n"," summary_row = {\n"," 'model_name': model_name,\n"," 'best_model_path': str(spec['model_file']),\n"," 'final_model_path': str(spec['final_file']),\n"," 'best_h5_path': str(best_h5_file),\n"," 'final_h5_path': str(final_h5_file),\n"," 'test_accuracy': accuracy,\n"," 'cohen_kappa': kappa,\n"," 'training_time_sec': float(tt),\n"," 'total_params': total_params,\n"," 'trainable_params': trainable_params,\n"," 'non_trainable_params': non_trainable_params,\n"," }\n"," export_training_sheet(workbook, model_name, summary_row, report_df, cm, [curve_fig, perf_fig, lr_fig])\n"," return summary_row\n"]},{"cell_type":"markdown","id":"90-main-hdr","metadata":{"papermill":{"duration":0.007388,"end_time":"2026-09-12T01:51:53.83364+00:00","exception":false,"start_time":"2026-09-12T01:51:53.826252+00:00","status":"completed"},"tags":[]},"source":["# 9.0 — Main Training Execution\n","\n","Iterate over `MODEL_SPECS`, run each model through `run_training`, populate the\n","Summary worksheet, and persist the workbook and model registry JSON."]},{"cell_type":"code","execution_count":19,"id":"90-main-code","metadata":{"execution":{"iopub.execute_input":"2026-09-12T01:51:53.85059Z","iopub.status.busy":"2026-09-12T01:51:53.849583Z","iopub.status.idle":"2026-09-12T03:54:49.315221Z","shell.execute_reply":"2026-09-12T03:54:49.31436Z"},"papermill":{"duration":7376.589165,"end_time":"2026-09-12T03:54:50.430251+00:00","exception":false,"start_time":"2026-09-12T01:51:53.841086+00:00","status":"completed"},"tags":[]},"outputs":[{"name":"stdout","output_type":"stream","text":["\n","============================== Training AlexNet_CNN_MultiHead ==============================\n"]},{"name":"stderr","output_type":"stream","text":["WARNING: All log messages before absl::InitializeLog() is called are written to STDERR\n","I0000 00:00:1789177914.518662 23 gpu_device.cc:2020] Created device /job:localhost/replica:0/task:0/device:GPU:0 with 15511 MB memory: -> device: 0, name: Tesla P100-PCIE-16GB, pci bus id: 0000:00:04.0, compute capability: 6.0\n"]},{"data":{"image/png":"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\n","text/plain":["
"]},"metadata":{},"output_type":"display_data"},{"name":"stdout","output_type":"stream","text":["\n","Epoch 1: LearningRateScheduler setting learning rate to 0.02.\n","Epoch 1/100\n","\u001b[1m 5/316\u001b[0m \u001b[37m━━━━━━━━━━━━━━━━━━━━\u001b[0m \u001b[1m10s\u001b[0m 34ms/step - head_1_accuracy: 0.1653 - head_1_loss: 2.0373 - head_2_accuracy: 0.1747 - head_2_loss: 2.0731 - head_3_accuracy: 0.1559 - head_3_loss: 2.0505 - head_4_accuracy: 0.0951 - head_4_loss: 2.0815 - head_5_accuracy: 0.1954 - head_5_loss: 2.0512 - head_6_accuracy: 0.1484 - head_6_loss: 2.0855 - head_7_accuracy: 0.1468 - head_7_loss: 2.0578 - loss: 14.4368"]},{"name":"stderr","output_type":"stream","text":["I0000 00:00:1789177931.094346 70 device_compiler.h:196] Compiled cluster using XLA! This line is logged at most once for the lifetime of the process.\n"]},{"name":"stdout","output_type":"stream","text":["\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 53ms/step - head_1_accuracy: 0.2315 - head_1_loss: 1.9250 - head_2_accuracy: 0.2375 - head_2_loss: 1.9105 - head_3_accuracy: 0.2401 - head_3_loss: 1.8970 - head_4_accuracy: 0.2284 - head_4_loss: 1.9068 - head_5_accuracy: 0.2454 - head_5_loss: 1.9129 - head_6_accuracy: 0.2343 - head_6_loss: 1.8987 - head_7_accuracy: 0.2364 - head_7_loss: 1.9096 - loss: 13.3606\n","Epoch 1: val_head_1_accuracy improved from None to 0.31861, saving model to /kaggle/working/results/models/AlexNet_CNN_MultiHead_best.keras\n","\n","Epoch 1: finished saving model to /kaggle/working/results/models/AlexNet_CNN_MultiHead_best.keras\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m38s\u001b[0m 80ms/step - head_1_accuracy: 0.2511 - head_1_loss: 1.8824 - head_2_accuracy: 0.2562 - head_2_loss: 1.8698 - head_3_accuracy: 0.2532 - head_3_loss: 1.8532 - head_4_accuracy: 0.2507 - head_4_loss: 1.8564 - head_5_accuracy: 0.2580 - head_5_loss: 1.8736 - head_6_accuracy: 0.2537 - head_6_loss: 1.8517 - head_7_accuracy: 0.2515 - head_7_loss: 1.8855 - loss: 13.0727 - val_head_1_accuracy: 0.3186 - val_head_1_loss: 1.7415 - val_head_2_accuracy: 0.3381 - val_head_2_loss: 1.7299 - val_head_3_accuracy: 0.2360 - val_head_3_loss: 1.7944 - val_head_4_accuracy: 0.3375 - val_head_4_loss: 1.7585 - val_head_5_accuracy: 0.3374 - val_head_5_loss: 1.7461 - val_head_6_accuracy: 0.2360 - val_head_6_loss: 1.8055 - val_head_7_accuracy: 0.2360 - val_head_7_loss: 1.7761 - val_loss: 12.3538 - learning_rate: 0.0200\n","\n","Epoch 2: LearningRateScheduler setting learning rate to 0.019996224067873887.\n","Epoch 2/100\n","\u001b[1m315/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m━\u001b[0m \u001b[1m0s\u001b[0m 32ms/step - head_1_accuracy: 0.3707 - head_1_loss: 1.5383 - head_2_accuracy: 0.3747 - head_2_loss: 1.5412 - head_3_accuracy: 0.3756 - head_3_loss: 1.5420 - head_4_accuracy: 0.3722 - head_4_loss: 1.5436 - head_5_accuracy: 0.3787 - head_5_loss: 1.5351 - head_6_accuracy: 0.3618 - head_6_loss: 1.5634 - head_7_accuracy: 0.3662 - head_7_loss: 1.5483 - loss: 10.8118\n","Epoch 2: val_head_1_accuracy improved from 0.31861 to 0.52278, saving model to /kaggle/working/results/models/AlexNet_CNN_MultiHead_best.keras\n","\n","Epoch 2: finished saving model to /kaggle/working/results/models/AlexNet_CNN_MultiHead_best.keras\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m13s\u001b[0m 41ms/step - head_1_accuracy: 0.4413 - head_1_loss: 1.3504 - head_2_accuracy: 0.4463 - head_2_loss: 1.3523 - head_3_accuracy: 0.4472 - head_3_loss: 1.3502 - head_4_accuracy: 0.4440 - head_4_loss: 1.3555 - head_5_accuracy: 0.4417 - head_5_loss: 1.3443 - head_6_accuracy: 0.4348 - head_6_loss: 1.3717 - head_7_accuracy: 0.4405 - head_7_loss: 1.3560 - loss: 9.4824 - val_head_1_accuracy: 0.5228 - val_head_1_loss: 0.9922 - val_head_2_accuracy: 0.5241 - val_head_2_loss: 1.0047 - val_head_3_accuracy: 0.5096 - val_head_3_loss: 1.0122 - val_head_4_accuracy: 0.5361 - val_head_4_loss: 1.0102 - val_head_5_accuracy: 0.5210 - val_head_5_loss: 1.0181 - val_head_6_accuracy: 0.5197 - val_head_6_loss: 1.0459 - val_head_7_accuracy: 0.5210 - val_head_7_loss: 1.0044 - val_loss: 7.0937 - learning_rate: 0.0200\n","\n","Epoch 3: LearningRateScheduler setting learning rate to 0.019984900073539133.\n","Epoch 3/100\n","\u001b[1m315/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m━\u001b[0m \u001b[1m0s\u001b[0m 32ms/step - head_1_accuracy: 0.5195 - head_1_loss: 3.1371 - head_2_accuracy: 0.5132 - head_2_loss: 3.6414 - head_3_accuracy: 0.5144 - head_3_loss: 3.0355 - head_4_accuracy: 0.5242 - head_4_loss: 2.9795 - head_5_accuracy: 0.5128 - head_5_loss: 3.5488 - head_6_accuracy: 0.5064 - head_6_loss: 2.8600 - head_7_accuracy: 0.5122 - head_7_loss: 4.1767 - loss: 23.3790\n","Epoch 3: val_head_1_accuracy did not improve from 0.52278\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m12s\u001b[0m 37ms/step - head_1_accuracy: 0.4344 - head_1_loss: 5.8482 - head_2_accuracy: 0.4298 - head_2_loss: 6.9574 - head_3_accuracy: 0.4287 - head_3_loss: 5.6100 - head_4_accuracy: 0.4363 - head_4_loss: 5.4880 - head_5_accuracy: 0.4450 - head_5_loss: 6.7438 - head_6_accuracy: 0.4253 - head_6_loss: 5.1975 - head_7_accuracy: 0.4331 - head_7_loss: 8.1476 - loss: 44.0161 - val_head_1_accuracy: 0.3334 - val_head_1_loss: 1.5532 - val_head_2_accuracy: 0.2559 - val_head_2_loss: 1.5710 - val_head_3_accuracy: 0.3238 - val_head_3_loss: 1.5491 - val_head_4_accuracy: 0.3373 - val_head_4_loss: 1.5392 - val_head_5_accuracy: 0.5178 - val_head_5_loss: 1.5510 - val_head_6_accuracy: 0.3266 - val_head_6_loss: 1.5689 - val_head_7_accuracy: 0.5107 - val_head_7_loss: 1.5542 - val_loss: 10.8865 - learning_rate: 0.0200\n","\n","Epoch 4: LearningRateScheduler setting learning rate to 0.019966039419298132.\n","Epoch 4/100\n","\u001b[1m315/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m━\u001b[0m \u001b[1m0s\u001b[0m 32ms/step - head_1_accuracy: 0.4021 - head_1_loss: 1.4211 - head_2_accuracy: 0.3504 - head_2_loss: 1.4738 - head_3_accuracy: 0.4283 - head_3_loss: 1.3669 - head_4_accuracy: 0.3959 - head_4_loss: 1.3979 - head_5_accuracy: 0.4201 - head_5_loss: 1.4143 - head_6_accuracy: 0.3756 - head_6_loss: 1.4214 - head_7_accuracy: 0.3906 - head_7_loss: 1.3999 - loss: 9.8953\n","Epoch 4: val_head_1_accuracy improved from 0.52278 to 0.61667, saving model to /kaggle/working/results/models/AlexNet_CNN_MultiHead_best.keras\n","\n","Epoch 4: finished saving model to /kaggle/working/results/models/AlexNet_CNN_MultiHead_best.keras\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m13s\u001b[0m 41ms/step - head_1_accuracy: 0.4435 - head_1_loss: 1.3010 - head_2_accuracy: 0.4281 - head_2_loss: 1.3203 - head_3_accuracy: 0.4721 - head_3_loss: 1.2689 - head_4_accuracy: 0.4415 - head_4_loss: 1.2848 - head_5_accuracy: 0.4388 - head_5_loss: 1.2942 - head_6_accuracy: 0.4485 - head_6_loss: 1.2911 - head_7_accuracy: 0.4663 - head_7_loss: 1.2828 - loss: 9.0448 - val_head_1_accuracy: 0.6167 - val_head_1_loss: 1.1435 - val_head_2_accuracy: 0.5945 - val_head_2_loss: 1.1294 - val_head_3_accuracy: 0.6240 - val_head_3_loss: 1.1159 - val_head_4_accuracy: 0.6221 - val_head_4_loss: 1.1339 - val_head_5_accuracy: 0.5182 - val_head_5_loss: 1.1213 - val_head_6_accuracy: 0.5963 - val_head_6_loss: 1.1279 - val_head_7_accuracy: 0.6239 - val_head_7_loss: 1.1165 - val_loss: 7.8881 - learning_rate: 0.0200\n","\n","Epoch 5: LearningRateScheduler setting learning rate to 0.019939661096230965.\n","Epoch 5/100\n","\u001b[1m315/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m━\u001b[0m \u001b[1m0s\u001b[0m 32ms/step - head_1_accuracy: 0.6015 - head_1_loss: 1.1292 - head_2_accuracy: 0.6065 - head_2_loss: 1.1173 - head_3_accuracy: 0.6052 - head_3_loss: 1.1152 - head_4_accuracy: 0.6045 - head_4_loss: 1.1167 - head_5_accuracy: 0.5782 - head_5_loss: 1.1182 - head_6_accuracy: 0.6021 - head_6_loss: 1.1209 - head_7_accuracy: 0.6068 - head_7_loss: 1.1129 - loss: 7.8305\n","Epoch 5: val_head_1_accuracy improved from 0.61667 to 0.65148, saving model to /kaggle/working/results/models/AlexNet_CNN_MultiHead_best.keras\n","\n","Epoch 5: finished saving model to /kaggle/working/results/models/AlexNet_CNN_MultiHead_best.keras\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m13s\u001b[0m 41ms/step - head_1_accuracy: 0.6255 - head_1_loss: 1.0935 - head_2_accuracy: 0.6270 - head_2_loss: 1.0831 - head_3_accuracy: 0.6253 - head_3_loss: 1.0824 - head_4_accuracy: 0.6291 - head_4_loss: 1.0803 - head_5_accuracy: 0.6193 - head_5_loss: 1.0852 - head_6_accuracy: 0.6246 - head_6_loss: 1.0873 - head_7_accuracy: 0.6291 - head_7_loss: 1.0781 - loss: 7.5910 - val_head_1_accuracy: 0.6515 - val_head_1_loss: 1.0059 - val_head_2_accuracy: 0.6436 - val_head_2_loss: 1.0061 - val_head_3_accuracy: 0.6445 - val_head_3_loss: 1.0019 - val_head_4_accuracy: 0.6545 - val_head_4_loss: 1.0014 - val_head_5_accuracy: 0.6498 - val_head_5_loss: 1.0062 - val_head_6_accuracy: 0.6438 - val_head_6_loss: 1.0082 - val_head_7_accuracy: 0.6450 - val_head_7_loss: 0.9999 - val_loss: 7.0309 - learning_rate: 0.0199\n","\n","Epoch 6: LearningRateScheduler setting learning rate to 0.019905791665072958.\n","Epoch 6/100\n","\u001b[1m315/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m━\u001b[0m \u001b[1m0s\u001b[0m 32ms/step - head_1_accuracy: 0.6716 - head_1_loss: 0.9913 - head_2_accuracy: 0.6670 - head_2_loss: 0.9914 - head_3_accuracy: 0.6684 - head_3_loss: 0.9872 - head_4_accuracy: 0.6743 - head_4_loss: 0.9869 - head_5_accuracy: 0.6702 - head_5_loss: 0.9882 - head_6_accuracy: 0.6710 - head_6_loss: 0.9878 - head_7_accuracy: 0.6715 - head_7_loss: 0.9849 - loss: 6.9176\n","Epoch 6: val_head_1_accuracy improved from 0.65148 to 0.74596, saving model to /kaggle/working/results/models/AlexNet_CNN_MultiHead_best.keras\n","\n","Epoch 6: finished saving model to /kaggle/working/results/models/AlexNet_CNN_MultiHead_best.keras\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m13s\u001b[0m 41ms/step - head_1_accuracy: 0.6824 - head_1_loss: 0.9726 - head_2_accuracy: 0.6772 - head_2_loss: 0.9732 - head_3_accuracy: 0.6804 - head_3_loss: 0.9690 - head_4_accuracy: 0.6837 - head_4_loss: 0.9685 - head_5_accuracy: 0.6814 - head_5_loss: 0.9730 - head_6_accuracy: 0.6819 - head_6_loss: 0.9691 - head_7_accuracy: 0.6840 - head_7_loss: 0.9670 - loss: 6.7937 - val_head_1_accuracy: 0.7460 - val_head_1_loss: 0.8284 - val_head_2_accuracy: 0.7413 - val_head_2_loss: 0.8299 - val_head_3_accuracy: 0.7464 - val_head_3_loss: 0.8267 - val_head_4_accuracy: 0.7431 - val_head_4_loss: 0.8180 - val_head_5_accuracy: 0.7425 - val_head_5_loss: 0.8244 - val_head_6_accuracy: 0.7444 - val_head_6_loss: 0.8253 - val_head_7_accuracy: 0.7470 - val_head_7_loss: 0.8126 - val_loss: 5.7685 - learning_rate: 0.0199\n","\n","Epoch 7: LearningRateScheduler setting learning rate to 0.019864465229470298.\n","Epoch 7/100\n","\u001b[1m315/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m━\u001b[0m \u001b[1m0s\u001b[0m 32ms/step - head_1_accuracy: 0.7157 - head_1_loss: 0.8822 - head_2_accuracy: 0.7135 - head_2_loss: 0.8785 - head_3_accuracy: 0.7158 - head_3_loss: 0.8738 - head_4_accuracy: 0.7162 - head_4_loss: 0.8784 - head_5_accuracy: 0.7160 - head_5_loss: 0.8778 - head_6_accuracy: 0.7163 - head_6_loss: 0.8740 - head_7_accuracy: 0.7181 - head_7_loss: 0.8756 - loss: 6.1404\n","Epoch 7: val_head_1_accuracy improved from 0.74596 to 0.79546, saving model to /kaggle/working/results/models/AlexNet_CNN_MultiHead_best.keras\n","\n","Epoch 7: finished saving model to /kaggle/working/results/models/AlexNet_CNN_MultiHead_best.keras\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m13s\u001b[0m 41ms/step - head_1_accuracy: 0.7271 - head_1_loss: 0.8530 - head_2_accuracy: 0.7253 - head_2_loss: 0.8480 - head_3_accuracy: 0.7277 - head_3_loss: 0.8442 - head_4_accuracy: 0.7284 - head_4_loss: 0.8478 - head_5_accuracy: 0.7293 - head_5_loss: 0.8492 - head_6_accuracy: 0.7284 - head_6_loss: 0.8439 - head_7_accuracy: 0.7292 - head_7_loss: 0.8462 - loss: 5.9338 - val_head_1_accuracy: 0.7955 - val_head_1_loss: 0.6944 - val_head_2_accuracy: 0.7937 - val_head_2_loss: 0.6949 - val_head_3_accuracy: 0.7984 - val_head_3_loss: 0.6959 - val_head_4_accuracy: 0.7943 - val_head_4_loss: 0.6964 - val_head_5_accuracy: 0.7929 - val_head_5_loss: 0.7002 - val_head_6_accuracy: 0.7925 - val_head_6_loss: 0.6933 - val_head_7_accuracy: 0.7889 - val_head_7_loss: 0.6924 - val_loss: 4.8712 - learning_rate: 0.0199\n","\n","Epoch 8: LearningRateScheduler setting learning rate to 0.019815723401640555.\n","Epoch 8/100\n","\u001b[1m315/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m━\u001b[0m \u001b[1m0s\u001b[0m 32ms/step - head_1_accuracy: 0.7627 - head_1_loss: 0.7610 - head_2_accuracy: 0.7601 - head_2_loss: 0.7591 - head_3_accuracy: 0.7627 - head_3_loss: 0.7567 - head_4_accuracy: 0.7644 - head_4_loss: 0.7550 - head_5_accuracy: 0.7632 - head_5_loss: 0.7611 - head_6_accuracy: 0.7642 - head_6_loss: 0.7539 - head_7_accuracy: 0.7638 - head_7_loss: 0.7576 - loss: 5.3044\n","Epoch 8: val_head_1_accuracy improved from 0.79546 to 0.83175, saving model to /kaggle/working/results/models/AlexNet_CNN_MultiHead_best.keras\n","\n","Epoch 8: finished saving model to /kaggle/working/results/models/AlexNet_CNN_MultiHead_best.keras\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m13s\u001b[0m 41ms/step - head_1_accuracy: 0.7691 - head_1_loss: 0.7389 - head_2_accuracy: 0.7680 - head_2_loss: 0.7372 - head_3_accuracy: 0.7699 - head_3_loss: 0.7346 - head_4_accuracy: 0.7715 - head_4_loss: 0.7344 - head_5_accuracy: 0.7708 - head_5_loss: 0.7390 - head_6_accuracy: 0.7713 - head_6_loss: 0.7322 - head_7_accuracy: 0.7722 - head_7_loss: 0.7350 - loss: 5.1526 - val_head_1_accuracy: 0.8318 - val_head_1_loss: 0.5910 - val_head_2_accuracy: 0.8277 - val_head_2_loss: 0.5959 - val_head_3_accuracy: 0.8287 - val_head_3_loss: 0.5947 - val_head_4_accuracy: 0.8220 - val_head_4_loss: 0.5893 - val_head_5_accuracy: 0.8226 - val_head_5_loss: 0.5951 - val_head_6_accuracy: 0.8197 - val_head_6_loss: 0.5909 - val_head_7_accuracy: 0.8155 - val_head_7_loss: 0.5929 - val_loss: 4.1535 - learning_rate: 0.0198\n","\n","Epoch 9: LearningRateScheduler setting learning rate to 0.019759615260472672.\n","Epoch 9/100\n","\u001b[1m315/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m━\u001b[0m \u001b[1m0s\u001b[0m 32ms/step - head_1_accuracy: 0.7870 - head_1_loss: 0.6807 - head_2_accuracy: 0.7863 - head_2_loss: 0.6785 - head_3_accuracy: 0.7870 - head_3_loss: 0.6728 - head_4_accuracy: 0.7888 - head_4_loss: 0.6755 - head_5_accuracy: 0.7872 - head_5_loss: 0.6798 - head_6_accuracy: 0.7894 - head_6_loss: 0.6720 - head_7_accuracy: 0.7900 - head_7_loss: 0.6780 - loss: 4.7372\n","Epoch 9: val_head_1_accuracy improved from 0.83175 to 0.83242, saving model to /kaggle/working/results/models/AlexNet_CNN_MultiHead_best.keras\n","\n","Epoch 9: finished saving model to /kaggle/working/results/models/AlexNet_CNN_MultiHead_best.keras\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m13s\u001b[0m 41ms/step - head_1_accuracy: 0.7923 - head_1_loss: 0.6652 - head_2_accuracy: 0.7930 - head_2_loss: 0.6621 - head_3_accuracy: 0.7923 - head_3_loss: 0.6580 - head_4_accuracy: 0.7939 - head_4_loss: 0.6613 - head_5_accuracy: 0.7928 - head_5_loss: 0.6644 - head_6_accuracy: 0.7950 - head_6_loss: 0.6574 - head_7_accuracy: 0.7946 - head_7_loss: 0.6613 - loss: 4.6310 - val_head_1_accuracy: 0.8324 - val_head_1_loss: 0.5648 - val_head_2_accuracy: 0.8263 - val_head_2_loss: 0.5648 - val_head_3_accuracy: 0.8303 - val_head_3_loss: 0.5581 - val_head_4_accuracy: 0.8289 - val_head_4_loss: 0.5668 - val_head_5_accuracy: 0.8297 - val_head_5_loss: 0.5677 - val_head_6_accuracy: 0.8247 - val_head_6_loss: 0.5668 - val_head_7_accuracy: 0.8300 - val_head_7_loss: 0.5675 - val_loss: 3.9608 - learning_rate: 0.0198\n","\n","Epoch 10: LearningRateScheduler setting learning rate to 0.01969619730210873.\n","Epoch 10/100\n","\u001b[1m315/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m━\u001b[0m \u001b[1m0s\u001b[0m 32ms/step - head_1_accuracy: 0.8059 - head_1_loss: 0.6182 - head_2_accuracy: 0.8047 - head_2_loss: 0.6154 - head_3_accuracy: 0.8044 - head_3_loss: 0.6105 - head_4_accuracy: 0.8098 - head_4_loss: 0.6140 - head_5_accuracy: 0.8060 - head_5_loss: 0.6183 - head_6_accuracy: 0.8087 - head_6_loss: 0.6104 - head_7_accuracy: 0.8065 - head_7_loss: 0.6145 - loss: 4.3012\n","Epoch 10: val_head_1_accuracy did not improve from 0.83242\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m12s\u001b[0m 37ms/step - head_1_accuracy: 0.8092 - head_1_loss: 0.6087 - head_2_accuracy: 0.8079 - head_2_loss: 0.6062 - head_3_accuracy: 0.8074 - head_3_loss: 0.6014 - head_4_accuracy: 0.8118 - head_4_loss: 0.6047 - head_5_accuracy: 0.8090 - head_5_loss: 0.6091 - head_6_accuracy: 0.8116 - head_6_loss: 0.6018 - head_7_accuracy: 0.8097 - head_7_loss: 0.6052 - loss: 4.2374 - val_head_1_accuracy: 0.8152 - val_head_1_loss: 0.5845 - val_head_2_accuracy: 0.8108 - val_head_2_loss: 0.5794 - val_head_3_accuracy: 0.8142 - val_head_3_loss: 0.5704 - val_head_4_accuracy: 0.8153 - val_head_4_loss: 0.5868 - val_head_5_accuracy: 0.8123 - val_head_5_loss: 0.5848 - val_head_6_accuracy: 0.8024 - val_head_6_loss: 0.5850 - val_head_7_accuracy: 0.8136 - val_head_7_loss: 0.5846 - val_loss: 4.0795 - learning_rate: 0.0197\n","\n","Epoch 11: LearningRateScheduler setting learning rate to 0.01962553338305709.\n","Epoch 11/100\n","\u001b[1m315/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m━\u001b[0m \u001b[1m0s\u001b[0m 32ms/step - head_1_accuracy: 0.8144 - head_1_loss: 0.5798 - head_2_accuracy: 0.8127 - head_2_loss: 0.5797 - head_3_accuracy: 0.8142 - head_3_loss: 0.5715 - head_4_accuracy: 0.8178 - head_4_loss: 0.5766 - head_5_accuracy: 0.8172 - head_5_loss: 0.5783 - head_6_accuracy: 0.8154 - head_6_loss: 0.5750 - head_7_accuracy: 0.8164 - head_7_loss: 0.5787 - loss: 4.0395\n","Epoch 11: val_head_1_accuracy improved from 0.83242 to 0.84741, saving model to /kaggle/working/results/models/AlexNet_CNN_MultiHead_best.keras\n","\n","Epoch 11: finished saving model to /kaggle/working/results/models/AlexNet_CNN_MultiHead_best.keras\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m13s\u001b[0m 42ms/step - head_1_accuracy: 0.7974 - head_1_loss: 0.6180 - head_2_accuracy: 0.7958 - head_2_loss: 0.6185 - head_3_accuracy: 0.7953 - head_3_loss: 0.6133 - head_4_accuracy: 0.8001 - head_4_loss: 0.6161 - head_5_accuracy: 0.7992 - head_5_loss: 0.6175 - head_6_accuracy: 0.7982 - head_6_loss: 0.6161 - head_7_accuracy: 0.7991 - head_7_loss: 0.6183 - loss: 4.3192 - val_head_1_accuracy: 0.8474 - val_head_1_loss: 0.4814 - val_head_2_accuracy: 0.8488 - val_head_2_loss: 0.4832 - val_head_3_accuracy: 0.8469 - val_head_3_loss: 0.4758 - val_head_4_accuracy: 0.8541 - val_head_4_loss: 0.4779 - val_head_5_accuracy: 0.8517 - val_head_5_loss: 0.4785 - val_head_6_accuracy: 0.8549 - val_head_6_loss: 0.4755 - val_head_7_accuracy: 0.8490 - val_head_7_loss: 0.4800 - val_loss: 3.3558 - learning_rate: 0.0196\n","\n","Epoch 12: LearningRateScheduler setting learning rate to 0.019547694655894312.\n","Epoch 12/100\n","\u001b[1m315/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m━\u001b[0m \u001b[1m0s\u001b[0m 32ms/step - head_1_accuracy: 0.8319 - head_1_loss: 0.5328 - head_2_accuracy: 0.8295 - head_2_loss: 0.5361 - head_3_accuracy: 0.8293 - head_3_loss: 0.5274 - head_4_accuracy: 0.8349 - head_4_loss: 0.5314 - head_5_accuracy: 0.8327 - head_5_loss: 0.5323 - head_6_accuracy: 0.8334 - head_6_loss: 0.5287 - head_7_accuracy: 0.8325 - head_7_loss: 0.5343 - loss: 3.7231\n","Epoch 12: val_head_1_accuracy improved from 0.84741 to 0.84748, saving model to /kaggle/working/results/models/AlexNet_CNN_MultiHead_best.keras\n","\n","Epoch 12: finished saving model to /kaggle/working/results/models/AlexNet_CNN_MultiHead_best.keras\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m13s\u001b[0m 41ms/step - head_1_accuracy: 0.8354 - head_1_loss: 0.5250 - head_2_accuracy: 0.8339 - head_2_loss: 0.5266 - head_3_accuracy: 0.8336 - head_3_loss: 0.5193 - head_4_accuracy: 0.8373 - head_4_loss: 0.5243 - head_5_accuracy: 0.8363 - head_5_loss: 0.5239 - head_6_accuracy: 0.8357 - head_6_loss: 0.5211 - head_7_accuracy: 0.8356 - head_7_loss: 0.5247 - loss: 3.6657 - val_head_1_accuracy: 0.8475 - val_head_1_loss: 0.4488 - val_head_2_accuracy: 0.8533 - val_head_2_loss: 0.4498 - val_head_3_accuracy: 0.8469 - val_head_3_loss: 0.4451 - val_head_4_accuracy: 0.8522 - val_head_4_loss: 0.4490 - val_head_5_accuracy: 0.8510 - val_head_5_loss: 0.4476 - val_head_6_accuracy: 0.8559 - val_head_6_loss: 0.4427 - val_head_7_accuracy: 0.8445 - val_head_7_loss: 0.4542 - val_loss: 3.1394 - learning_rate: 0.0195\n","\n","Epoch 13: LearningRateScheduler setting learning rate to 0.019462759497620545.\n","Epoch 13/100\n","\u001b[1m315/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m━\u001b[0m \u001b[1m0s\u001b[0m 32ms/step - head_1_accuracy: 0.8451 - head_1_loss: 0.4933 - head_2_accuracy: 0.8444 - head_2_loss: 0.4957 - head_3_accuracy: 0.8430 - head_3_loss: 0.4864 - head_4_accuracy: 0.8427 - head_4_loss: 0.4922 - head_5_accuracy: 0.8436 - head_5_loss: 0.4907 - head_6_accuracy: 0.8444 - head_6_loss: 0.4893 - head_7_accuracy: 0.8433 - head_7_loss: 0.4935 - loss: 3.4410\n","Epoch 13: val_head_1_accuracy did not improve from 0.84748\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m12s\u001b[0m 37ms/step - head_1_accuracy: 0.8472 - head_1_loss: 0.4895 - head_2_accuracy: 0.8447 - head_2_loss: 0.4924 - head_3_accuracy: 0.8455 - head_3_loss: 0.4825 - head_4_accuracy: 0.8471 - head_4_loss: 0.4885 - head_5_accuracy: 0.8470 - head_5_loss: 0.4869 - head_6_accuracy: 0.8481 - head_6_loss: 0.4850 - head_7_accuracy: 0.8464 - head_7_loss: 0.4888 - loss: 3.4140 - val_head_1_accuracy: 0.8407 - val_head_1_loss: 0.5033 - val_head_2_accuracy: 0.8375 - val_head_2_loss: 0.5056 - val_head_3_accuracy: 0.8441 - val_head_3_loss: 0.4873 - val_head_4_accuracy: 0.8381 - val_head_4_loss: 0.5100 - val_head_5_accuracy: 0.8397 - val_head_5_loss: 0.5046 - val_head_6_accuracy: 0.8278 - val_head_6_loss: 0.5133 - val_head_7_accuracy: 0.8378 - val_head_7_loss: 0.5122 - val_loss: 3.5383 - learning_rate: 0.0195\n","\n","Epoch 14: LearningRateScheduler setting learning rate to 0.01937081343074052.\n","Epoch 14/100\n","\u001b[1m315/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m━\u001b[0m \u001b[1m0s\u001b[0m 32ms/step - head_1_accuracy: 0.8572 - head_1_loss: 0.4541 - head_2_accuracy: 0.8576 - head_2_loss: 0.4591 - head_3_accuracy: 0.8575 - head_3_loss: 0.4490 - head_4_accuracy: 0.8591 - head_4_loss: 0.4541 - head_5_accuracy: 0.8582 - head_5_loss: 0.4520 - head_6_accuracy: 0.8585 - head_6_loss: 0.4537 - head_7_accuracy: 0.8572 - head_7_loss: 0.4555 - loss: 3.1774\n","Epoch 14: val_head_1_accuracy did not improve from 0.84748\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m12s\u001b[0m 37ms/step - head_1_accuracy: 0.8594 - head_1_loss: 0.4498 - head_2_accuracy: 0.8588 - head_2_loss: 0.4537 - head_3_accuracy: 0.8592 - head_3_loss: 0.4443 - head_4_accuracy: 0.8598 - head_4_loss: 0.4496 - head_5_accuracy: 0.8605 - head_5_loss: 0.4471 - head_6_accuracy: 0.8608 - head_6_loss: 0.4473 - head_7_accuracy: 0.8591 - head_7_loss: 0.4489 - loss: 3.1405 - val_head_1_accuracy: 0.8355 - val_head_1_loss: 0.5260 - val_head_2_accuracy: 0.8352 - val_head_2_loss: 0.5243 - val_head_3_accuracy: 0.8424 - val_head_3_loss: 0.5077 - val_head_4_accuracy: 0.8335 - val_head_4_loss: 0.5307 - val_head_5_accuracy: 0.8345 - val_head_5_loss: 0.5255 - val_head_6_accuracy: 0.8283 - val_head_6_loss: 0.5248 - val_head_7_accuracy: 0.8390 - val_head_7_loss: 0.5309 - val_loss: 3.6716 - learning_rate: 0.0194\n","\n","Epoch 15: LearningRateScheduler setting learning rate to 0.01927194903714966.\n","Epoch 15/100\n","\u001b[1m315/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m━\u001b[0m \u001b[1m0s\u001b[0m 32ms/step - head_1_accuracy: 0.8746 - head_1_loss: 0.4120 - head_2_accuracy: 0.8723 - head_2_loss: 0.4176 - head_3_accuracy: 0.8731 - head_3_loss: 0.4087 - head_4_accuracy: 0.8736 - head_4_loss: 0.4139 - head_5_accuracy: 0.8746 - head_5_loss: 0.4128 - head_6_accuracy: 0.8736 - head_6_loss: 0.4117 - head_7_accuracy: 0.8739 - head_7_loss: 0.4131 - loss: 2.8898\n","Epoch 15: val_head_1_accuracy improved from 0.84748 to 0.85661, saving model to /kaggle/working/results/models/AlexNet_CNN_MultiHead_best.keras\n","\n","Epoch 15: finished saving model to /kaggle/working/results/models/AlexNet_CNN_MultiHead_best.keras\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m13s\u001b[0m 41ms/step - head_1_accuracy: 0.8755 - head_1_loss: 0.4102 - head_2_accuracy: 0.8724 - head_2_loss: 0.4151 - head_3_accuracy: 0.8743 - head_3_loss: 0.4053 - head_4_accuracy: 0.8749 - head_4_loss: 0.4116 - head_5_accuracy: 0.8759 - head_5_loss: 0.4093 - head_6_accuracy: 0.8753 - head_6_loss: 0.4090 - head_7_accuracy: 0.8749 - head_7_loss: 0.4100 - loss: 2.8701 - val_head_1_accuracy: 0.8566 - val_head_1_loss: 0.4572 - val_head_2_accuracy: 0.8557 - val_head_2_loss: 0.4505 - val_head_3_accuracy: 0.8572 - val_head_3_loss: 0.4479 - val_head_4_accuracy: 0.8500 - val_head_4_loss: 0.4671 - val_head_5_accuracy: 0.8539 - val_head_5_loss: 0.4631 - val_head_6_accuracy: 0.8482 - val_head_6_loss: 0.4660 - val_head_7_accuracy: 0.8574 - val_head_7_loss: 0.4675 - val_loss: 3.2206 - learning_rate: 0.0193\n","\n","Epoch 16: LearningRateScheduler setting learning rate to 0.019166265864911925.\n","Epoch 16/100\n","\u001b[1m315/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m━\u001b[0m \u001b[1m0s\u001b[0m 32ms/step - head_1_accuracy: 0.8902 - head_1_loss: 0.3746 - head_2_accuracy: 0.8891 - head_2_loss: 0.3802 - head_3_accuracy: 0.8884 - head_3_loss: 0.3731 - head_4_accuracy: 0.8898 - head_4_loss: 0.3768 - head_5_accuracy: 0.8890 - head_5_loss: 0.3744 - head_6_accuracy: 0.8892 - head_6_loss: 0.3734 - head_7_accuracy: 0.8889 - head_7_loss: 0.3766 - loss: 2.6291\n","Epoch 16: val_head_1_accuracy improved from 0.85661 to 0.87242, saving model to /kaggle/working/results/models/AlexNet_CNN_MultiHead_best.keras\n","\n","Epoch 16: finished saving model to /kaggle/working/results/models/AlexNet_CNN_MultiHead_best.keras\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m13s\u001b[0m 41ms/step - head_1_accuracy: 0.8877 - head_1_loss: 0.3751 - head_2_accuracy: 0.8875 - head_2_loss: 0.3798 - head_3_accuracy: 0.8876 - head_3_loss: 0.3721 - head_4_accuracy: 0.8889 - head_4_loss: 0.3766 - head_5_accuracy: 0.8885 - head_5_loss: 0.3739 - head_6_accuracy: 0.8890 - head_6_loss: 0.3730 - head_7_accuracy: 0.8882 - head_7_loss: 0.3753 - loss: 2.6258 - val_head_1_accuracy: 0.8724 - val_head_1_loss: 0.3937 - val_head_2_accuracy: 0.8681 - val_head_2_loss: 0.3929 - val_head_3_accuracy: 0.8677 - val_head_3_loss: 0.3957 - val_head_4_accuracy: 0.8679 - val_head_4_loss: 0.4022 - val_head_5_accuracy: 0.8651 - val_head_5_loss: 0.4054 - val_head_6_accuracy: 0.8633 - val_head_6_loss: 0.4041 - val_head_7_accuracy: 0.8699 - val_head_7_loss: 0.4017 - val_loss: 2.7958 - learning_rate: 0.0192\n","\n","Epoch 17: LearningRateScheduler setting learning rate to 0.019053870328023387.\n","Epoch 17/100\n","\u001b[1m315/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m━\u001b[0m \u001b[1m0s\u001b[0m 32ms/step - head_1_accuracy: 0.9025 - head_1_loss: 0.3291 - head_2_accuracy: 0.9013 - head_2_loss: 0.3332 - head_3_accuracy: 0.9024 - head_3_loss: 0.3261 - head_4_accuracy: 0.9030 - head_4_loss: 0.3302 - head_5_accuracy: 0.9036 - head_5_loss: 0.3301 - head_6_accuracy: 0.9035 - head_6_loss: 0.3276 - head_7_accuracy: 0.9036 - head_7_loss: 0.3283 - loss: 2.3047\n","Epoch 17: val_head_1_accuracy improved from 0.87242 to 0.90122, saving model to /kaggle/working/results/models/AlexNet_CNN_MultiHead_best.keras\n","\n","Epoch 17: finished saving model to /kaggle/working/results/models/AlexNet_CNN_MultiHead_best.keras\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m13s\u001b[0m 42ms/step - head_1_accuracy: 0.8994 - head_1_loss: 0.3360 - head_2_accuracy: 0.8994 - head_2_loss: 0.3405 - head_3_accuracy: 0.9000 - head_3_loss: 0.3331 - head_4_accuracy: 0.9010 - head_4_loss: 0.3368 - head_5_accuracy: 0.9013 - head_5_loss: 0.3366 - head_6_accuracy: 0.9012 - head_6_loss: 0.3339 - head_7_accuracy: 0.9008 - head_7_loss: 0.3355 - loss: 2.3529 - val_head_1_accuracy: 0.9012 - val_head_1_loss: 0.3196 - val_head_2_accuracy: 0.9031 - val_head_2_loss: 0.3217 - val_head_3_accuracy: 0.9023 - val_head_3_loss: 0.3118 - val_head_4_accuracy: 0.9033 - val_head_4_loss: 0.3226 - val_head_5_accuracy: 0.9031 - val_head_5_loss: 0.3164 - val_head_6_accuracy: 0.9081 - val_head_6_loss: 0.3149 - val_head_7_accuracy: 0.8994 - val_head_7_loss: 0.3198 - val_loss: 2.2288 - learning_rate: 0.0191\n","\n","Epoch 18: LearningRateScheduler setting learning rate to 0.018934875599262328.\n","Epoch 18/100\n","\u001b[1m315/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m━\u001b[0m \u001b[1m0s\u001b[0m 32ms/step - head_1_accuracy: 0.9140 - head_1_loss: 0.3000 - head_2_accuracy: 0.9144 - head_2_loss: 0.3035 - head_3_accuracy: 0.9156 - head_3_loss: 0.2973 - head_4_accuracy: 0.9155 - head_4_loss: 0.3012 - head_5_accuracy: 0.9156 - head_5_loss: 0.3005 - head_6_accuracy: 0.9138 - head_6_loss: 0.2990 - head_7_accuracy: 0.9136 - head_7_loss: 0.3008 - loss: 2.1022\n","Epoch 18: val_head_1_accuracy improved from 0.90122 to 0.90693, saving model to /kaggle/working/results/models/AlexNet_CNN_MultiHead_best.keras\n","\n","Epoch 18: finished saving model to /kaggle/working/results/models/AlexNet_CNN_MultiHead_best.keras\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m13s\u001b[0m 42ms/step - head_1_accuracy: 0.9149 - head_1_loss: 0.2984 - head_2_accuracy: 0.9141 - head_2_loss: 0.3026 - head_3_accuracy: 0.9156 - head_3_loss: 0.2956 - head_4_accuracy: 0.9158 - head_4_loss: 0.2996 - head_5_accuracy: 0.9146 - head_5_loss: 0.2993 - head_6_accuracy: 0.9140 - head_6_loss: 0.2978 - head_7_accuracy: 0.9140 - head_7_loss: 0.2993 - loss: 2.0920 - val_head_1_accuracy: 0.9069 - val_head_1_loss: 0.3055 - val_head_2_accuracy: 0.9035 - val_head_2_loss: 0.3109 - val_head_3_accuracy: 0.9060 - val_head_3_loss: 0.3065 - val_head_4_accuracy: 0.9085 - val_head_4_loss: 0.3006 - val_head_5_accuracy: 0.9083 - val_head_5_loss: 0.3027 - val_head_6_accuracy: 0.9014 - val_head_6_loss: 0.3042 - val_head_7_accuracy: 0.9102 - val_head_7_loss: 0.2995 - val_loss: 2.1305 - learning_rate: 0.0189\n","\n","Epoch 19: LearningRateScheduler setting learning rate to 0.01880940149623386.\n","Epoch 19/100\n","\u001b[1m315/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m━\u001b[0m \u001b[1m0s\u001b[0m 32ms/step - head_1_accuracy: 0.9245 - head_1_loss: 0.2720 - head_2_accuracy: 0.9227 - head_2_loss: 0.2766 - head_3_accuracy: 0.9241 - head_3_loss: 0.2692 - head_4_accuracy: 0.9243 - head_4_loss: 0.2724 - head_5_accuracy: 0.9246 - head_5_loss: 0.2723 - head_6_accuracy: 0.9242 - head_6_loss: 0.2721 - head_7_accuracy: 0.9230 - head_7_loss: 0.2727 - loss: 1.9074\n","Epoch 19: val_head_1_accuracy improved from 0.90693 to 0.92794, saving model to /kaggle/working/results/models/AlexNet_CNN_MultiHead_best.keras\n","\n","Epoch 19: finished saving model to /kaggle/working/results/models/AlexNet_CNN_MultiHead_best.keras\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m13s\u001b[0m 42ms/step - head_1_accuracy: 0.9219 - head_1_loss: 0.2792 - head_2_accuracy: 0.9207 - head_2_loss: 0.2835 - head_3_accuracy: 0.9218 - head_3_loss: 0.2759 - head_4_accuracy: 0.9219 - head_4_loss: 0.2789 - head_5_accuracy: 0.9222 - head_5_loss: 0.2794 - head_6_accuracy: 0.9220 - head_6_loss: 0.2786 - head_7_accuracy: 0.9212 - head_7_loss: 0.2791 - loss: 1.9549 - val_head_1_accuracy: 0.9279 - val_head_1_loss: 0.2630 - val_head_2_accuracy: 0.9271 - val_head_2_loss: 0.2648 - val_head_3_accuracy: 0.9264 - val_head_3_loss: 0.2624 - val_head_4_accuracy: 0.9277 - val_head_4_loss: 0.2638 - val_head_5_accuracy: 0.9285 - val_head_5_loss: 0.2633 - val_head_6_accuracy: 0.9272 - val_head_6_loss: 0.2625 - val_head_7_accuracy: 0.9274 - val_head_7_loss: 0.2631 - val_loss: 1.8438 - learning_rate: 0.0188\n","\n","Epoch 20: LearningRateScheduler setting learning rate to 0.018677574360723745.\n","Epoch 20/100\n","\u001b[1m315/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m━\u001b[0m \u001b[1m0s\u001b[0m 32ms/step - head_1_accuracy: 0.9367 - head_1_loss: 0.2348 - head_2_accuracy: 0.9369 - head_2_loss: 0.2384 - head_3_accuracy: 0.9375 - head_3_loss: 0.2331 - head_4_accuracy: 0.9380 - head_4_loss: 0.2360 - head_5_accuracy: 0.9372 - head_5_loss: 0.2359 - head_6_accuracy: 0.9379 - head_6_loss: 0.2343 - head_7_accuracy: 0.9375 - head_7_loss: 0.2345 - loss: 1.6471\n","Epoch 20: val_head_1_accuracy improved from 0.92794 to 0.95629, saving model to /kaggle/working/results/models/AlexNet_CNN_MultiHead_best.keras\n","\n","Epoch 20: finished saving model to /kaggle/working/results/models/AlexNet_CNN_MultiHead_best.keras\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m13s\u001b[0m 42ms/step - head_1_accuracy: 0.9341 - head_1_loss: 0.2419 - head_2_accuracy: 0.9340 - head_2_loss: 0.2458 - head_3_accuracy: 0.9348 - head_3_loss: 0.2393 - head_4_accuracy: 0.9354 - head_4_loss: 0.2430 - head_5_accuracy: 0.9346 - head_5_loss: 0.2418 - head_6_accuracy: 0.9351 - head_6_loss: 0.2409 - head_7_accuracy: 0.9351 - head_7_loss: 0.2407 - loss: 1.6938 - val_head_1_accuracy: 0.9563 - val_head_1_loss: 0.1941 - val_head_2_accuracy: 0.9563 - val_head_2_loss: 0.1935 - val_head_3_accuracy: 0.9553 - val_head_3_loss: 0.1917 - val_head_4_accuracy: 0.9564 - val_head_4_loss: 0.1936 - val_head_5_accuracy: 0.9577 - val_head_5_loss: 0.1914 - val_head_6_accuracy: 0.9565 - val_head_6_loss: 0.1886 - val_head_7_accuracy: 0.9561 - val_head_7_loss: 0.1947 - val_loss: 1.3482 - learning_rate: 0.0187\n","\n","Epoch 21: LearningRateScheduler setting learning rate to 0.01853952693148294.\n","Epoch 21/100\n","\u001b[1m315/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m━\u001b[0m \u001b[1m0s\u001b[0m 32ms/step - head_1_accuracy: 0.9485 - head_1_loss: 0.2097 - head_2_accuracy: 0.9479 - head_2_loss: 0.2124 - head_3_accuracy: 0.9484 - head_3_loss: 0.2076 - head_4_accuracy: 0.9488 - head_4_loss: 0.2100 - head_5_accuracy: 0.9485 - head_5_loss: 0.2098 - head_6_accuracy: 0.9475 - head_6_loss: 0.2075 - head_7_accuracy: 0.9481 - head_7_loss: 0.2101 - loss: 1.4671\n","Epoch 21: val_head_1_accuracy did not improve from 0.95629\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m12s\u001b[0m 37ms/step - head_1_accuracy: 0.9473 - head_1_loss: 0.2104 - head_2_accuracy: 0.9468 - head_2_loss: 0.2134 - head_3_accuracy: 0.9473 - head_3_loss: 0.2086 - head_4_accuracy: 0.9477 - head_4_loss: 0.2109 - head_5_accuracy: 0.9474 - head_5_loss: 0.2108 - head_6_accuracy: 0.9465 - head_6_loss: 0.2089 - head_7_accuracy: 0.9473 - head_7_loss: 0.2100 - loss: 1.4735 - val_head_1_accuracy: 0.9514 - val_head_1_loss: 0.1889 - val_head_2_accuracy: 0.9512 - val_head_2_loss: 0.1903 - val_head_3_accuracy: 0.9505 - val_head_3_loss: 0.1870 - val_head_4_accuracy: 0.9529 - val_head_4_loss: 0.1853 - val_head_5_accuracy: 0.9500 - val_head_5_loss: 0.1910 - val_head_6_accuracy: 0.9515 - val_head_6_loss: 0.1845 - val_head_7_accuracy: 0.9518 - val_head_7_loss: 0.1874 - val_loss: 1.3157 - learning_rate: 0.0185\n","\n","Epoch 22: LearningRateScheduler setting learning rate to 0.018395398210570905.\n","Epoch 22/100\n","\u001b[1m315/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m━\u001b[0m \u001b[1m0s\u001b[0m 32ms/step - head_1_accuracy: 0.9476 - head_1_loss: 0.2133 - head_2_accuracy: 0.9467 - head_2_loss: 0.2152 - head_3_accuracy: 0.9473 - head_3_loss: 0.2092 - head_4_accuracy: 0.9481 - head_4_loss: 0.2120 - head_5_accuracy: 0.9478 - head_5_loss: 0.2116 - head_6_accuracy: 0.9473 - head_6_loss: 0.2107 - head_7_accuracy: 0.9465 - head_7_loss: 0.2122 - loss: 1.4842\n","Epoch 22: val_head_1_accuracy did not improve from 0.95629\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m12s\u001b[0m 37ms/step - head_1_accuracy: 0.9495 - head_1_loss: 0.2028 - head_2_accuracy: 0.9487 - head_2_loss: 0.2047 - head_3_accuracy: 0.9493 - head_3_loss: 0.1995 - head_4_accuracy: 0.9498 - head_4_loss: 0.2019 - head_5_accuracy: 0.9500 - head_5_loss: 0.2019 - head_6_accuracy: 0.9492 - head_6_loss: 0.2008 - head_7_accuracy: 0.9494 - head_7_loss: 0.2013 - loss: 1.4115 - val_head_1_accuracy: 0.8684 - val_head_1_loss: 0.4610 - val_head_2_accuracy: 0.8695 - val_head_2_loss: 0.4510 - val_head_3_accuracy: 0.8703 - val_head_3_loss: 0.4419 - val_head_4_accuracy: 0.8713 - val_head_4_loss: 0.4589 - val_head_5_accuracy: 0.8706 - val_head_5_loss: 0.4614 - val_head_6_accuracy: 0.8687 - val_head_6_loss: 0.4698 - val_head_7_accuracy: 0.8710 - val_head_7_loss: 0.4632 - val_loss: 3.2109 - learning_rate: 0.0184\n","\n","Epoch 23: LearningRateScheduler setting learning rate to 0.018245333323392336.\n","Epoch 23/100\n","\u001b[1m315/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m━\u001b[0m \u001b[1m0s\u001b[0m 32ms/step - head_1_accuracy: 0.9560 - head_1_loss: 0.1883 - head_2_accuracy: 0.9553 - head_2_loss: 0.1906 - head_3_accuracy: 0.9551 - head_3_loss: 0.1856 - head_4_accuracy: 0.9561 - head_4_loss: 0.1880 - head_5_accuracy: 0.9553 - head_5_loss: 0.1883 - head_6_accuracy: 0.9551 - head_6_loss: 0.1869 - head_7_accuracy: 0.9558 - head_7_loss: 0.1873 - loss: 1.3150\n","Epoch 23: val_head_1_accuracy improved from 0.95629 to 0.96096, saving model to /kaggle/working/results/models/AlexNet_CNN_MultiHead_best.keras\n","\n","Epoch 23: finished saving model to /kaggle/working/results/models/AlexNet_CNN_MultiHead_best.keras\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m13s\u001b[0m 42ms/step - head_1_accuracy: 0.9568 - head_1_loss: 0.1836 - head_2_accuracy: 0.9561 - head_2_loss: 0.1865 - head_3_accuracy: 0.9559 - head_3_loss: 0.1810 - head_4_accuracy: 0.9570 - head_4_loss: 0.1835 - head_5_accuracy: 0.9567 - head_5_loss: 0.1832 - head_6_accuracy: 0.9563 - head_6_loss: 0.1821 - head_7_accuracy: 0.9566 - head_7_loss: 0.1824 - loss: 1.2823 - val_head_1_accuracy: 0.9610 - val_head_1_loss: 0.1659 - val_head_2_accuracy: 0.9582 - val_head_2_loss: 0.1678 - val_head_3_accuracy: 0.9598 - val_head_3_loss: 0.1663 - val_head_4_accuracy: 0.9597 - val_head_4_loss: 0.1666 - val_head_5_accuracy: 0.9589 - val_head_5_loss: 0.1684 - val_head_6_accuracy: 0.9566 - val_head_6_loss: 0.1686 - val_head_7_accuracy: 0.9606 - val_head_7_loss: 0.1668 - val_loss: 1.1707 - learning_rate: 0.0182\n","\n","Epoch 24: LearningRateScheduler setting learning rate to 0.01808948337256816.\n","Epoch 24/100\n","\u001b[1m315/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m━\u001b[0m \u001b[1m0s\u001b[0m 33ms/step - head_1_accuracy: 0.9628 - head_1_loss: 0.1625 - head_2_accuracy: 0.9618 - head_2_loss: 0.1657 - head_3_accuracy: 0.9630 - head_3_loss: 0.1598 - head_4_accuracy: 0.9622 - head_4_loss: 0.1625 - head_5_accuracy: 0.9628 - head_5_loss: 0.1615 - head_6_accuracy: 0.9621 - head_6_loss: 0.1618 - head_7_accuracy: 0.9637 - head_7_loss: 0.1618 - loss: 1.1357\n","Epoch 24: val_head_1_accuracy improved from 0.96096 to 0.97536, saving model to /kaggle/working/results/models/AlexNet_CNN_MultiHead_best.keras\n","\n","Epoch 24: finished saving model to /kaggle/working/results/models/AlexNet_CNN_MultiHead_best.keras\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m13s\u001b[0m 42ms/step - head_1_accuracy: 0.9624 - head_1_loss: 0.1642 - head_2_accuracy: 0.9616 - head_2_loss: 0.1672 - head_3_accuracy: 0.9621 - head_3_loss: 0.1606 - head_4_accuracy: 0.9622 - head_4_loss: 0.1643 - head_5_accuracy: 0.9621 - head_5_loss: 0.1632 - head_6_accuracy: 0.9621 - head_6_loss: 0.1630 - head_7_accuracy: 0.9625 - head_7_loss: 0.1630 - loss: 1.1458 - val_head_1_accuracy: 0.9754 - val_head_1_loss: 0.1246 - val_head_2_accuracy: 0.9751 - val_head_2_loss: 0.1238 - val_head_3_accuracy: 0.9751 - val_head_3_loss: 0.1242 - val_head_4_accuracy: 0.9753 - val_head_4_loss: 0.1247 - val_head_5_accuracy: 0.9754 - val_head_5_loss: 0.1264 - val_head_6_accuracy: 0.9755 - val_head_6_loss: 0.1239 - val_head_7_accuracy: 0.9751 - val_head_7_loss: 0.1259 - val_loss: 0.8734 - learning_rate: 0.0181\n","\n","Epoch 25: LearningRateScheduler setting learning rate to 0.017928005285788026.\n","Epoch 25/100\n","\u001b[1m315/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m━\u001b[0m \u001b[1m0s\u001b[0m 32ms/step - head_1_accuracy: 0.9738 - head_1_loss: 0.1267 - head_2_accuracy: 0.9738 - head_2_loss: 0.1270 - head_3_accuracy: 0.9740 - head_3_loss: 0.1248 - head_4_accuracy: 0.9732 - head_4_loss: 0.1272 - head_5_accuracy: 0.9741 - head_5_loss: 0.1274 - head_6_accuracy: 0.9742 - head_6_loss: 0.1258 - head_7_accuracy: 0.9729 - head_7_loss: 0.1265 - loss: 0.8854\n","Epoch 25: val_head_1_accuracy did not improve from 0.97536\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m12s\u001b[0m 38ms/step - head_1_accuracy: 0.9703 - head_1_loss: 0.1369 - head_2_accuracy: 0.9705 - head_2_loss: 0.1385 - head_3_accuracy: 0.9708 - head_3_loss: 0.1351 - head_4_accuracy: 0.9705 - head_4_loss: 0.1371 - head_5_accuracy: 0.9706 - head_5_loss: 0.1378 - head_6_accuracy: 0.9703 - head_6_loss: 0.1359 - head_7_accuracy: 0.9701 - head_7_loss: 0.1361 - loss: 0.9573 - val_head_1_accuracy: 0.9498 - val_head_1_loss: 0.1908 - val_head_2_accuracy: 0.9478 - val_head_2_loss: 0.1912 - val_head_3_accuracy: 0.9479 - val_head_3_loss: 0.1894 - val_head_4_accuracy: 0.9496 - val_head_4_loss: 0.1890 - val_head_5_accuracy: 0.9483 - val_head_5_loss: 0.1908 - val_head_6_accuracy: 0.9492 - val_head_6_loss: 0.1890 - val_head_7_accuracy: 0.9489 - val_head_7_loss: 0.1893 - val_loss: 1.3311 - learning_rate: 0.0179\n","\n","Epoch 26: LearningRateScheduler setting learning rate to 0.01776106165779741.\n","Epoch 26/100\n","\u001b[1m315/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m━\u001b[0m \u001b[1m0s\u001b[0m 32ms/step - head_1_accuracy: 0.9739 - head_1_loss: 0.1315 - head_2_accuracy: 0.9731 - head_2_loss: 0.1333 - head_3_accuracy: 0.9734 - head_3_loss: 0.1303 - head_4_accuracy: 0.9734 - head_4_loss: 0.1317 - head_5_accuracy: 0.9735 - head_5_loss: 0.1316 - head_6_accuracy: 0.9730 - head_6_loss: 0.1316 - head_7_accuracy: 0.9736 - head_7_loss: 0.1321 - loss: 0.9221\n","Epoch 26: val_head_1_accuracy improved from 0.97536 to 0.98048, saving model to /kaggle/working/results/models/AlexNet_CNN_MultiHead_best.keras\n","\n","Epoch 26: finished saving model to /kaggle/working/results/models/AlexNet_CNN_MultiHead_best.keras\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m13s\u001b[0m 42ms/step - head_1_accuracy: 0.9719 - head_1_loss: 0.1381 - head_2_accuracy: 0.9712 - head_2_loss: 0.1404 - head_3_accuracy: 0.9714 - head_3_loss: 0.1363 - head_4_accuracy: 0.9716 - head_4_loss: 0.1385 - head_5_accuracy: 0.9714 - head_5_loss: 0.1379 - head_6_accuracy: 0.9715 - head_6_loss: 0.1383 - head_7_accuracy: 0.9719 - head_7_loss: 0.1381 - loss: 0.9679 - val_head_1_accuracy: 0.9805 - val_head_1_loss: 0.1114 - val_head_2_accuracy: 0.9806 - val_head_2_loss: 0.1108 - val_head_3_accuracy: 0.9797 - val_head_3_loss: 0.1110 - val_head_4_accuracy: 0.9798 - val_head_4_loss: 0.1113 - val_head_5_accuracy: 0.9797 - val_head_5_loss: 0.1135 - val_head_6_accuracy: 0.9797 - val_head_6_loss: 0.1108 - val_head_7_accuracy: 0.9808 - val_head_7_loss: 0.1119 - val_loss: 0.7799 - learning_rate: 0.0178\n","\n","Epoch 27: LearningRateScheduler setting learning rate to 0.017588820586678492.\n","Epoch 27/100\n","\u001b[1m315/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m━\u001b[0m \u001b[1m0s\u001b[0m 32ms/step - head_1_accuracy: 0.9787 - head_1_loss: 0.1109 - head_2_accuracy: 0.9784 - head_2_loss: 0.1112 - head_3_accuracy: 0.9783 - head_3_loss: 0.1087 - head_4_accuracy: 0.9781 - head_4_loss: 0.1106 - head_5_accuracy: 0.9784 - head_5_loss: 0.1111 - head_6_accuracy: 0.9785 - head_6_loss: 0.1098 - head_7_accuracy: 0.9783 - head_7_loss: 0.1101 - loss: 0.7723\n","Epoch 27: val_head_1_accuracy did not improve from 0.98048\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m12s\u001b[0m 37ms/step - head_1_accuracy: 0.9743 - head_1_loss: 0.1269 - head_2_accuracy: 0.9744 - head_2_loss: 0.1283 - head_3_accuracy: 0.9746 - head_3_loss: 0.1241 - head_4_accuracy: 0.9742 - head_4_loss: 0.1266 - head_5_accuracy: 0.9745 - head_5_loss: 0.1264 - head_6_accuracy: 0.9742 - head_6_loss: 0.1258 - head_7_accuracy: 0.9743 - head_7_loss: 0.1260 - loss: 0.8842 - val_head_1_accuracy: 0.9724 - val_head_1_loss: 0.1219 - val_head_2_accuracy: 0.9722 - val_head_2_loss: 0.1228 - val_head_3_accuracy: 0.9725 - val_head_3_loss: 0.1195 - val_head_4_accuracy: 0.9726 - val_head_4_loss: 0.1211 - val_head_5_accuracy: 0.9725 - val_head_5_loss: 0.1206 - val_head_6_accuracy: 0.9733 - val_head_6_loss: 0.1187 - val_head_7_accuracy: 0.9725 - val_head_7_loss: 0.1215 - val_loss: 0.8470 - learning_rate: 0.0176\n","\n","Epoch 28: LearningRateScheduler setting learning rate to 0.01741145550458964.\n","Epoch 28/100\n","\u001b[1m315/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m━\u001b[0m \u001b[1m0s\u001b[0m 32ms/step - head_1_accuracy: 0.9653 - head_1_loss: 0.1528 - head_2_accuracy: 0.9658 - head_2_loss: 0.1546 - head_3_accuracy: 0.9667 - head_3_loss: 0.1496 - head_4_accuracy: 0.9660 - head_4_loss: 0.1521 - head_5_accuracy: 0.9651 - head_5_loss: 0.1521 - head_6_accuracy: 0.9659 - head_6_loss: 0.1516 - head_7_accuracy: 0.9664 - head_7_loss: 0.1516 - loss: 1.0644\n","Epoch 28: val_head_1_accuracy improved from 0.98048 to 0.98508, saving model to /kaggle/working/results/models/AlexNet_CNN_MultiHead_best.keras\n","\n","Epoch 28: finished saving model to /kaggle/working/results/models/AlexNet_CNN_MultiHead_best.keras\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m13s\u001b[0m 42ms/step - head_1_accuracy: 0.9766 - head_1_loss: 0.1145 - head_2_accuracy: 0.9769 - head_2_loss: 0.1153 - head_3_accuracy: 0.9774 - head_3_loss: 0.1124 - head_4_accuracy: 0.9770 - head_4_loss: 0.1144 - head_5_accuracy: 0.9767 - head_5_loss: 0.1151 - head_6_accuracy: 0.9768 - head_6_loss: 0.1140 - head_7_accuracy: 0.9772 - head_7_loss: 0.1141 - loss: 0.8000 - val_head_1_accuracy: 0.9851 - val_head_1_loss: 0.0840 - val_head_2_accuracy: 0.9852 - val_head_2_loss: 0.0838 - val_head_3_accuracy: 0.9857 - val_head_3_loss: 0.0826 - val_head_4_accuracy: 0.9851 - val_head_4_loss: 0.0842 - val_head_5_accuracy: 0.9852 - val_head_5_loss: 0.0847 - val_head_6_accuracy: 0.9852 - val_head_6_loss: 0.0837 - val_head_7_accuracy: 0.9862 - val_head_7_loss: 0.0835 - val_loss: 0.5858 - learning_rate: 0.0174\n","\n","Epoch 29: LearningRateScheduler setting learning rate to 0.017229145003133918.\n","Epoch 29/100\n","\u001b[1m315/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m━\u001b[0m \u001b[1m0s\u001b[0m 32ms/step - head_1_accuracy: 0.9782 - head_1_loss: 0.1179 - head_2_accuracy: 0.9782 - head_2_loss: 0.1196 - head_3_accuracy: 0.9777 - head_3_loss: 0.1178 - head_4_accuracy: 0.9777 - head_4_loss: 0.1188 - head_5_accuracy: 0.9782 - head_5_loss: 0.1201 - head_6_accuracy: 0.9775 - head_6_loss: 0.1197 - head_7_accuracy: 0.9778 - head_7_loss: 0.1182 - loss: 0.8322\n","Epoch 29: val_head_1_accuracy improved from 0.98508 to 0.98516, saving model to /kaggle/working/results/models/AlexNet_CNN_MultiHead_best.keras\n","\n","Epoch 29: finished saving model to /kaggle/working/results/models/AlexNet_CNN_MultiHead_best.keras\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m13s\u001b[0m 42ms/step - head_1_accuracy: 0.9794 - head_1_loss: 0.1135 - head_2_accuracy: 0.9790 - head_2_loss: 0.1145 - head_3_accuracy: 0.9786 - head_3_loss: 0.1124 - head_4_accuracy: 0.9788 - head_4_loss: 0.1136 - head_5_accuracy: 0.9788 - head_5_loss: 0.1162 - head_6_accuracy: 0.9787 - head_6_loss: 0.1138 - head_7_accuracy: 0.9789 - head_7_loss: 0.1141 - loss: 0.7982 - val_head_1_accuracy: 0.9852 - val_head_1_loss: 0.0839 - val_head_2_accuracy: 0.9849 - val_head_2_loss: 0.0794 - val_head_3_accuracy: 0.9849 - val_head_3_loss: 0.0773 - val_head_4_accuracy: 0.9852 - val_head_4_loss: 0.0784 - val_head_5_accuracy: 0.9849 - val_head_5_loss: 0.0881 - val_head_6_accuracy: 0.9852 - val_head_6_loss: 0.0744 - val_head_7_accuracy: 0.9852 - val_head_7_loss: 0.0839 - val_loss: 0.5652 - learning_rate: 0.0172\n","\n","Epoch 30: LearningRateScheduler setting learning rate to 0.0170420726535325.\n","Epoch 30/100\n","\u001b[1m315/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m━\u001b[0m \u001b[1m0s\u001b[0m 32ms/step - head_1_accuracy: 0.9844 - head_1_loss: 0.0903 - head_2_accuracy: 0.9839 - head_2_loss: 0.0808 - head_3_accuracy: 0.9842 - head_3_loss: 0.0774 - head_4_accuracy: 0.9846 - head_4_loss: 0.0786 - head_5_accuracy: 0.9847 - head_5_loss: 0.0969 - head_6_accuracy: 0.9843 - head_6_loss: 0.0755 - head_7_accuracy: 0.9844 - head_7_loss: 0.0949 - loss: 0.5944\n","Epoch 30: val_head_1_accuracy improved from 0.98516 to 0.99065, saving model to /kaggle/working/results/models/AlexNet_CNN_MultiHead_best.keras\n","\n","Epoch 30: finished saving model to /kaggle/working/results/models/AlexNet_CNN_MultiHead_best.keras\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m13s\u001b[0m 42ms/step - head_1_accuracy: 0.9867 - head_1_loss: 0.0827 - head_2_accuracy: 0.9863 - head_2_loss: 0.0692 - head_3_accuracy: 0.9869 - head_3_loss: 0.0669 - head_4_accuracy: 0.9868 - head_4_loss: 0.0669 - head_5_accuracy: 0.9871 - head_5_loss: 0.0903 - head_6_accuracy: 0.9867 - head_6_loss: 0.0643 - head_7_accuracy: 0.9868 - head_7_loss: 0.0873 - loss: 0.5278 - val_head_1_accuracy: 0.9906 - val_head_1_loss: 0.0536 - val_head_2_accuracy: 0.9909 - val_head_2_loss: 0.0343 - val_head_3_accuracy: 0.9904 - val_head_3_loss: 0.0349 - val_head_4_accuracy: 0.9898 - val_head_4_loss: 0.0352 - val_head_5_accuracy: 0.9895 - val_head_5_loss: 0.0579 - val_head_6_accuracy: 0.9898 - val_head_6_loss: 0.0336 - val_head_7_accuracy: 0.9899 - val_head_7_loss: 0.0530 - val_loss: 0.3024 - learning_rate: 0.0170\n","\n","Epoch 31: LearningRateScheduler setting learning rate to 0.016850426821783986.\n","Epoch 31/100\n","\u001b[1m315/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m━\u001b[0m \u001b[1m0s\u001b[0m 32ms/step - head_1_accuracy: 0.9878 - head_1_loss: 0.0494 - head_2_accuracy: 0.9875 - head_2_loss: 0.0440 - head_3_accuracy: 0.9878 - head_3_loss: 0.0422 - head_4_accuracy: 0.9883 - head_4_loss: 0.0420 - head_5_accuracy: 0.9881 - head_5_loss: 0.0503 - head_6_accuracy: 0.9873 - head_6_loss: 0.0426 - head_7_accuracy: 0.9877 - head_7_loss: 0.0497 - loss: 0.3202\n","Epoch 31: val_head_1_accuracy did not improve from 0.99065\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m12s\u001b[0m 37ms/step - head_1_accuracy: 0.9849 - head_1_loss: 0.0572 - head_2_accuracy: 0.9845 - head_2_loss: 0.0564 - head_3_accuracy: 0.9850 - head_3_loss: 0.0541 - head_4_accuracy: 0.9846 - head_4_loss: 0.0540 - head_5_accuracy: 0.9847 - head_5_loss: 0.0571 - head_6_accuracy: 0.9839 - head_6_loss: 0.0545 - head_7_accuracy: 0.9848 - head_7_loss: 0.0570 - loss: 0.3898 - val_head_1_accuracy: 0.9681 - val_head_1_loss: 0.0982 - val_head_2_accuracy: 0.9688 - val_head_2_loss: 0.0950 - val_head_3_accuracy: 0.9683 - val_head_3_loss: 0.0966 - val_head_4_accuracy: 0.9682 - val_head_4_loss: 0.0979 - val_head_5_accuracy: 0.9679 - val_head_5_loss: 0.0984 - val_head_6_accuracy: 0.9676 - val_head_6_loss: 0.0965 - val_head_7_accuracy: 0.9677 - val_head_7_loss: 0.0995 - val_loss: 0.6828 - learning_rate: 0.0169\n","\n","Epoch 32: LearningRateScheduler setting learning rate to 0.01665440047899583.\n","Epoch 32/100\n","\u001b[1m315/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m━\u001b[0m \u001b[1m0s\u001b[0m 32ms/step - head_1_accuracy: 0.9900 - head_1_loss: 0.0389 - head_2_accuracy: 0.9898 - head_2_loss: 0.0388 - head_3_accuracy: 0.9897 - head_3_loss: 0.0388 - head_4_accuracy: 0.9904 - head_4_loss: 0.0380 - head_5_accuracy: 0.9891 - head_5_loss: 0.0384 - head_6_accuracy: 0.9895 - head_6_loss: 0.0387 - head_7_accuracy: 0.9900 - head_7_loss: 0.0394 - loss: 0.2710\n","Epoch 32: val_head_1_accuracy did not improve from 0.99065\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m12s\u001b[0m 37ms/step - head_1_accuracy: 0.9901 - head_1_loss: 0.0383 - head_2_accuracy: 0.9899 - head_2_loss: 0.0383 - head_3_accuracy: 0.9898 - head_3_loss: 0.0378 - head_4_accuracy: 0.9902 - head_4_loss: 0.0374 - head_5_accuracy: 0.9898 - head_5_loss: 0.0376 - head_6_accuracy: 0.9898 - head_6_loss: 0.0384 - head_7_accuracy: 0.9899 - head_7_loss: 0.0387 - loss: 0.2665 - val_head_1_accuracy: 0.9863 - val_head_1_loss: 0.0374 - val_head_2_accuracy: 0.9867 - val_head_2_loss: 0.0364 - val_head_3_accuracy: 0.9867 - val_head_3_loss: 0.0362 - val_head_4_accuracy: 0.9862 - val_head_4_loss: 0.0364 - val_head_5_accuracy: 0.9862 - val_head_5_loss: 0.0375 - val_head_6_accuracy: 0.9860 - val_head_6_loss: 0.0368 - val_head_7_accuracy: 0.9862 - val_head_7_loss: 0.0356 - val_loss: 0.2569 - learning_rate: 0.0167\n","\n","Epoch 33: LearningRateScheduler setting learning rate to 0.016454191007078768.\n","Epoch 33/100\n","\u001b[1m315/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m━\u001b[0m \u001b[1m0s\u001b[0m 32ms/step - head_1_accuracy: 0.9901 - head_1_loss: 0.0357 - head_2_accuracy: 0.9899 - head_2_loss: 0.0364 - head_3_accuracy: 0.9894 - head_3_loss: 0.0356 - head_4_accuracy: 0.9900 - head_4_loss: 0.0347 - head_5_accuracy: 0.9897 - head_5_loss: 0.0356 - head_6_accuracy: 0.9896 - head_6_loss: 0.0362 - head_7_accuracy: 0.9900 - head_7_loss: 0.0357 - loss: 0.2499\n","Epoch 33: val_head_1_accuracy improved from 0.99065 to 0.99495, saving model to /kaggle/working/results/models/AlexNet_CNN_MultiHead_best.keras\n","\n","Epoch 33: finished saving model to /kaggle/working/results/models/AlexNet_CNN_MultiHead_best.keras\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m22s\u001b[0m 42ms/step - head_1_accuracy: 0.9913 - head_1_loss: 0.0352 - head_2_accuracy: 0.9913 - head_2_loss: 0.0360 - head_3_accuracy: 0.9909 - head_3_loss: 0.0351 - head_4_accuracy: 0.9913 - head_4_loss: 0.0345 - head_5_accuracy: 0.9908 - head_5_loss: 0.0354 - head_6_accuracy: 0.9909 - head_6_loss: 0.0356 - head_7_accuracy: 0.9913 - head_7_loss: 0.0352 - loss: 0.2471 - val_head_1_accuracy: 0.9950 - val_head_1_loss: 0.0160 - val_head_2_accuracy: 0.9947 - val_head_2_loss: 0.0157 - val_head_3_accuracy: 0.9954 - val_head_3_loss: 0.0158 - val_head_4_accuracy: 0.9954 - val_head_4_loss: 0.0155 - val_head_5_accuracy: 0.9958 - val_head_5_loss: 0.0156 - val_head_6_accuracy: 0.9955 - val_head_6_loss: 0.0156 - val_head_7_accuracy: 0.9956 - val_head_7_loss: 0.0154 - val_loss: 0.1093 - learning_rate: 0.0165\n","\n","Epoch 34: LearningRateScheduler setting learning rate to 0.01625.\n","Epoch 34/100\n","\u001b[1m315/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m━\u001b[0m \u001b[1m0s\u001b[0m 32ms/step - head_1_accuracy: 0.9937 - head_1_loss: 0.0218 - head_2_accuracy: 0.9932 - head_2_loss: 0.0222 - head_3_accuracy: 0.9933 - head_3_loss: 0.0216 - head_4_accuracy: 0.9940 - head_4_loss: 0.0214 - head_5_accuracy: 0.9939 - head_5_loss: 0.0215 - head_6_accuracy: 0.9939 - head_6_loss: 0.0218 - head_7_accuracy: 0.9938 - head_7_loss: 0.0216 - loss: 0.1519\n","Epoch 34: val_head_1_accuracy did not improve from 0.99495\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m12s\u001b[0m 37ms/step - head_1_accuracy: 0.9934 - head_1_loss: 0.0235 - head_2_accuracy: 0.9928 - head_2_loss: 0.0237 - head_3_accuracy: 0.9929 - head_3_loss: 0.0232 - head_4_accuracy: 0.9933 - head_4_loss: 0.0232 - head_5_accuracy: 0.9931 - head_5_loss: 0.0235 - head_6_accuracy: 0.9933 - head_6_loss: 0.0238 - head_7_accuracy: 0.9931 - head_7_loss: 0.0235 - loss: 0.1645 - val_head_1_accuracy: 0.9941 - val_head_1_loss: 0.0209 - val_head_2_accuracy: 0.9943 - val_head_2_loss: 0.0199 - val_head_3_accuracy: 0.9939 - val_head_3_loss: 0.0203 - val_head_4_accuracy: 0.9941 - val_head_4_loss: 0.0203 - val_head_5_accuracy: 0.9938 - val_head_5_loss: 0.0221 - val_head_6_accuracy: 0.9939 - val_head_6_loss: 0.0202 - val_head_7_accuracy: 0.9944 - val_head_7_loss: 0.0194 - val_loss: 0.1371 - learning_rate: 0.0162\n","\n","Epoch 35: LearningRateScheduler setting learning rate to 0.01604203306079512.\n","Epoch 35/100\n","\u001b[1m315/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m━\u001b[0m \u001b[1m0s\u001b[0m 32ms/step - head_1_accuracy: 0.9948 - head_1_loss: 0.0170 - head_2_accuracy: 0.9951 - head_2_loss: 0.0175 - head_3_accuracy: 0.9947 - head_3_loss: 0.0167 - head_4_accuracy: 0.9950 - head_4_loss: 0.0167 - head_5_accuracy: 0.9950 - head_5_loss: 0.0167 - head_6_accuracy: 0.9952 - head_6_loss: 0.0171 - head_7_accuracy: 0.9950 - head_7_loss: 0.0173 - loss: 0.1190\n","Epoch 35: val_head_1_accuracy improved from 0.99495 to 0.99584, saving model to /kaggle/working/results/models/AlexNet_CNN_MultiHead_best.keras\n","\n","Epoch 35: finished saving model to /kaggle/working/results/models/AlexNet_CNN_MultiHead_best.keras\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m13s\u001b[0m 42ms/step - head_1_accuracy: 0.9942 - head_1_loss: 0.0196 - head_2_accuracy: 0.9944 - head_2_loss: 0.0200 - head_3_accuracy: 0.9943 - head_3_loss: 0.0195 - head_4_accuracy: 0.9943 - head_4_loss: 0.0194 - head_5_accuracy: 0.9944 - head_5_loss: 0.0195 - head_6_accuracy: 0.9942 - head_6_loss: 0.0202 - head_7_accuracy: 0.9942 - head_7_loss: 0.0200 - loss: 0.1383 - val_head_1_accuracy: 0.9958 - val_head_1_loss: 0.0146 - val_head_2_accuracy: 0.9957 - val_head_2_loss: 0.0141 - val_head_3_accuracy: 0.9958 - val_head_3_loss: 0.0147 - val_head_4_accuracy: 0.9960 - val_head_4_loss: 0.0147 - val_head_5_accuracy: 0.9960 - val_head_5_loss: 0.0145 - val_head_6_accuracy: 0.9959 - val_head_6_loss: 0.0143 - val_head_7_accuracy: 0.9960 - val_head_7_loss: 0.0144 - val_loss: 0.1020 - learning_rate: 0.0160\n","\n","Epoch 36: LearningRateScheduler setting learning rate to 0.015830499594543305.\n","Epoch 36/100\n","\u001b[1m315/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m━\u001b[0m \u001b[1m0s\u001b[0m 32ms/step - head_1_accuracy: 0.9946 - head_1_loss: 0.0227 - head_2_accuracy: 0.9943 - head_2_loss: 0.0228 - head_3_accuracy: 0.9945 - head_3_loss: 0.0224 - head_4_accuracy: 0.9944 - head_4_loss: 0.0223 - head_5_accuracy: 0.9946 - head_5_loss: 0.0227 - head_6_accuracy: 0.9943 - head_6_loss: 0.0238 - head_7_accuracy: 0.9945 - head_7_loss: 0.0227 - loss: 0.1594\n","Epoch 36: val_head_1_accuracy improved from 0.99584 to 0.99673, saving model to /kaggle/working/results/models/AlexNet_CNN_MultiHead_best.keras\n","\n","Epoch 36: finished saving model to /kaggle/working/results/models/AlexNet_CNN_MultiHead_best.keras\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m20s\u001b[0m 42ms/step - head_1_accuracy: 0.9937 - head_1_loss: 0.0273 - head_2_accuracy: 0.9935 - head_2_loss: 0.0276 - head_3_accuracy: 0.9936 - head_3_loss: 0.0272 - head_4_accuracy: 0.9937 - head_4_loss: 0.0268 - head_5_accuracy: 0.9937 - head_5_loss: 0.0275 - head_6_accuracy: 0.9933 - head_6_loss: 0.0288 - head_7_accuracy: 0.9936 - head_7_loss: 0.0273 - loss: 0.1926 - val_head_1_accuracy: 0.9967 - val_head_1_loss: 0.0114 - val_head_2_accuracy: 0.9967 - val_head_2_loss: 0.0112 - val_head_3_accuracy: 0.9966 - val_head_3_loss: 0.0112 - val_head_4_accuracy: 0.9967 - val_head_4_loss: 0.0113 - val_head_5_accuracy: 0.9967 - val_head_5_loss: 0.0113 - val_head_6_accuracy: 0.9970 - val_head_6_loss: 0.0112 - val_head_7_accuracy: 0.9968 - val_head_7_loss: 0.0112 - val_loss: 0.0793 - learning_rate: 0.0158\n","\n","Epoch 37: LearningRateScheduler setting learning rate to 0.01561561259751415.\n","Epoch 37/100\n","\u001b[1m315/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m━\u001b[0m \u001b[1m0s\u001b[0m 32ms/step - head_1_accuracy: 0.9983 - head_1_loss: 0.0075 - head_2_accuracy: 0.9980 - head_2_loss: 0.0079 - head_3_accuracy: 0.9981 - head_3_loss: 0.0077 - head_4_accuracy: 0.9980 - head_4_loss: 0.0075 - head_5_accuracy: 0.9978 - head_5_loss: 0.0076 - head_6_accuracy: 0.9978 - head_6_loss: 0.0078 - head_7_accuracy: 0.9979 - head_7_loss: 0.0081 - loss: 0.0542\n","Epoch 37: val_head_1_accuracy did not improve from 0.99673\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m12s\u001b[0m 37ms/step - head_1_accuracy: 0.9983 - head_1_loss: 0.0069 - head_2_accuracy: 0.9980 - head_2_loss: 0.0071 - head_3_accuracy: 0.9981 - head_3_loss: 0.0070 - head_4_accuracy: 0.9981 - head_4_loss: 0.0069 - head_5_accuracy: 0.9980 - head_5_loss: 0.0070 - head_6_accuracy: 0.9980 - head_6_loss: 0.0072 - head_7_accuracy: 0.9980 - head_7_loss: 0.0073 - loss: 0.0492 - val_head_1_accuracy: 0.9950 - val_head_1_loss: 0.0140 - val_head_2_accuracy: 0.9950 - val_head_2_loss: 0.0139 - val_head_3_accuracy: 0.9950 - val_head_3_loss: 0.0139 - val_head_4_accuracy: 0.9942 - val_head_4_loss: 0.0149 - val_head_5_accuracy: 0.9941 - val_head_5_loss: 0.0154 - val_head_6_accuracy: 0.9944 - val_head_6_loss: 0.0146 - val_head_7_accuracy: 0.9950 - val_head_7_loss: 0.0139 - val_loss: 0.1013 - learning_rate: 0.0156\n","\n","Epoch 38: LearningRateScheduler setting learning rate to 0.015397588442698465.\n","Epoch 38/100\n","\u001b[1m315/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m━\u001b[0m \u001b[1m0s\u001b[0m 32ms/step - head_1_accuracy: 0.9929 - head_1_loss: 0.0253 - head_2_accuracy: 0.9932 - head_2_loss: 0.0259 - head_3_accuracy: 0.9929 - head_3_loss: 0.0251 - head_4_accuracy: 0.9931 - head_4_loss: 0.0249 - head_5_accuracy: 0.9931 - head_5_loss: 0.0251 - head_6_accuracy: 0.9931 - head_6_loss: 0.0254 - head_7_accuracy: 0.9930 - head_7_loss: 0.0246 - loss: 0.1763\n","Epoch 38: val_head_1_accuracy improved from 0.99673 to 0.99785, saving model to /kaggle/working/results/models/AlexNet_CNN_MultiHead_best.keras\n","\n","Epoch 38: finished saving model to /kaggle/working/results/models/AlexNet_CNN_MultiHead_best.keras\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m13s\u001b[0m 42ms/step - head_1_accuracy: 0.9914 - head_1_loss: 0.0328 - head_2_accuracy: 0.9917 - head_2_loss: 0.0335 - head_3_accuracy: 0.9913 - head_3_loss: 0.0323 - head_4_accuracy: 0.9916 - head_4_loss: 0.0320 - head_5_accuracy: 0.9916 - head_5_loss: 0.0322 - head_6_accuracy: 0.9917 - head_6_loss: 0.0324 - head_7_accuracy: 0.9915 - head_7_loss: 0.0320 - loss: 0.2273 - val_head_1_accuracy: 0.9978 - val_head_1_loss: 0.0085 - val_head_2_accuracy: 0.9981 - val_head_2_loss: 0.0079 - val_head_3_accuracy: 0.9980 - val_head_3_loss: 0.0082 - val_head_4_accuracy: 0.9982 - val_head_4_loss: 0.0084 - val_head_5_accuracy: 0.9981 - val_head_5_loss: 0.0084 - val_head_6_accuracy: 0.9981 - val_head_6_loss: 0.0085 - val_head_7_accuracy: 0.9980 - val_head_7_loss: 0.0086 - val_loss: 0.0587 - learning_rate: 0.0154\n","\n","Epoch 39: LearningRateScheduler setting learning rate to 0.015176646661939037.\n","Epoch 39/100\n","\u001b[1m315/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m━\u001b[0m \u001b[1m0s\u001b[0m 32ms/step - head_1_accuracy: 0.9984 - head_1_loss: 0.0073 - head_2_accuracy: 0.9985 - head_2_loss: 0.0068 - head_3_accuracy: 0.9988 - head_3_loss: 0.0068 - head_4_accuracy: 0.9985 - head_4_loss: 0.0070 - head_5_accuracy: 0.9985 - head_5_loss: 0.0072 - head_6_accuracy: 0.9986 - head_6_loss: 0.0073 - head_7_accuracy: 0.9985 - head_7_loss: 0.0076 - loss: 0.0500\n","Epoch 39: val_head_1_accuracy did not improve from 0.99785\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m12s\u001b[0m 37ms/step - head_1_accuracy: 0.9961 - head_1_loss: 0.0172 - head_2_accuracy: 0.9962 - head_2_loss: 0.0173 - head_3_accuracy: 0.9965 - head_3_loss: 0.0167 - head_4_accuracy: 0.9963 - head_4_loss: 0.0170 - head_5_accuracy: 0.9964 - head_5_loss: 0.0168 - head_6_accuracy: 0.9961 - head_6_loss: 0.0174 - head_7_accuracy: 0.9962 - head_7_loss: 0.0168 - loss: 0.1193 - val_head_1_accuracy: 0.9967 - val_head_1_loss: 0.0133 - val_head_2_accuracy: 0.9969 - val_head_2_loss: 0.0121 - val_head_3_accuracy: 0.9968 - val_head_3_loss: 0.0131 - val_head_4_accuracy: 0.9967 - val_head_4_loss: 0.0128 - val_head_5_accuracy: 0.9967 - val_head_5_loss: 0.0133 - val_head_6_accuracy: 0.9970 - val_head_6_loss: 0.0128 - val_head_7_accuracy: 0.9969 - val_head_7_loss: 0.0132 - val_loss: 0.0905 - learning_rate: 0.0152\n","\n","Epoch 40: LearningRateScheduler setting learning rate to 0.014953009724880662.\n","Epoch 40/100\n","\u001b[1m315/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m━\u001b[0m \u001b[1m0s\u001b[0m 32ms/step - head_1_accuracy: 0.9970 - head_1_loss: 0.0141 - head_2_accuracy: 0.9971 - head_2_loss: 0.0136 - head_3_accuracy: 0.9968 - head_3_loss: 0.0141 - head_4_accuracy: 0.9969 - head_4_loss: 0.0140 - head_5_accuracy: 0.9967 - head_5_loss: 0.0143 - head_6_accuracy: 0.9969 - head_6_loss: 0.0139 - head_7_accuracy: 0.9966 - head_7_loss: 0.0145 - loss: 0.0985\n","Epoch 40: val_head_1_accuracy did not improve from 0.99785\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m12s\u001b[0m 38ms/step - head_1_accuracy: 0.9958 - head_1_loss: 0.0179 - head_2_accuracy: 0.9958 - head_2_loss: 0.0180 - head_3_accuracy: 0.9956 - head_3_loss: 0.0179 - head_4_accuracy: 0.9956 - head_4_loss: 0.0178 - head_5_accuracy: 0.9956 - head_5_loss: 0.0178 - head_6_accuracy: 0.9957 - head_6_loss: 0.0179 - head_7_accuracy: 0.9955 - head_7_loss: 0.0183 - loss: 0.1255 - val_head_1_accuracy: 0.9967 - val_head_1_loss: 0.0108 - val_head_2_accuracy: 0.9968 - val_head_2_loss: 0.0108 - val_head_3_accuracy: 0.9968 - val_head_3_loss: 0.0107 - val_head_4_accuracy: 0.9968 - val_head_4_loss: 0.0109 - val_head_5_accuracy: 0.9966 - val_head_5_loss: 0.0109 - val_head_6_accuracy: 0.9969 - val_head_6_loss: 0.0108 - val_head_7_accuracy: 0.9970 - val_head_7_loss: 0.0111 - val_loss: 0.0758 - learning_rate: 0.0150\n","\n","Epoch 41: LearningRateScheduler setting learning rate to 0.014726902814962063.\n","Epoch 41/100\n","\u001b[1m315/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m━\u001b[0m \u001b[1m0s\u001b[0m 32ms/step - head_1_accuracy: 0.9985 - head_1_loss: 0.0062 - head_2_accuracy: 0.9986 - head_2_loss: 0.0060 - head_3_accuracy: 0.9985 - head_3_loss: 0.0064 - head_4_accuracy: 0.9983 - head_4_loss: 0.0061 - head_5_accuracy: 0.9983 - head_5_loss: 0.0066 - head_6_accuracy: 0.9981 - head_6_loss: 0.0064 - head_7_accuracy: 0.9983 - head_7_loss: 0.0067 - loss: 0.0444\n","Epoch 41: val_head_1_accuracy did not improve from 0.99785\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m12s\u001b[0m 37ms/step - head_1_accuracy: 0.9986 - head_1_loss: 0.0054 - head_2_accuracy: 0.9987 - head_2_loss: 0.0051 - head_3_accuracy: 0.9987 - head_3_loss: 0.0055 - head_4_accuracy: 0.9985 - head_4_loss: 0.0052 - head_5_accuracy: 0.9986 - head_5_loss: 0.0055 - head_6_accuracy: 0.9984 - head_6_loss: 0.0055 - head_7_accuracy: 0.9985 - head_7_loss: 0.0057 - loss: 0.0380 - val_head_1_accuracy: 0.9978 - val_head_1_loss: 0.0069 - val_head_2_accuracy: 0.9978 - val_head_2_loss: 0.0062 - val_head_3_accuracy: 0.9978 - val_head_3_loss: 0.0069 - val_head_4_accuracy: 0.9977 - val_head_4_loss: 0.0065 - val_head_5_accuracy: 0.9978 - val_head_5_loss: 0.0073 - val_head_6_accuracy: 0.9979 - val_head_6_loss: 0.0062 - val_head_7_accuracy: 0.9978 - val_head_7_loss: 0.0067 - val_loss: 0.0468 - learning_rate: 0.0147\n","\n","Epoch 42: LearningRateScheduler setting learning rate to 0.014498553602675261.\n","Epoch 42/100\n","\u001b[1m315/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m━\u001b[0m \u001b[1m0s\u001b[0m 32ms/step - head_1_accuracy: 0.9980 - head_1_loss: 0.0091 - head_2_accuracy: 0.9982 - head_2_loss: 0.0091 - head_3_accuracy: 0.9982 - head_3_loss: 0.0089 - head_4_accuracy: 0.9981 - head_4_loss: 0.0090 - head_5_accuracy: 0.9981 - head_5_loss: 0.0089 - head_6_accuracy: 0.9981 - head_6_loss: 0.0090 - head_7_accuracy: 0.9980 - head_7_loss: 0.0090 - loss: 0.0630\n","Epoch 42: val_head_1_accuracy did not improve from 0.99785\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m12s\u001b[0m 37ms/step - head_1_accuracy: 0.9938 - head_1_loss: 0.0303 - head_2_accuracy: 0.9937 - head_2_loss: 0.0311 - head_3_accuracy: 0.9937 - head_3_loss: 0.0300 - head_4_accuracy: 0.9936 - head_4_loss: 0.0302 - head_5_accuracy: 0.9938 - head_5_loss: 0.0299 - head_6_accuracy: 0.9937 - head_6_loss: 0.0303 - head_7_accuracy: 0.9937 - head_7_loss: 0.0297 - loss: 0.2116 - val_head_1_accuracy: 0.9967 - val_head_1_loss: 0.0135 - val_head_2_accuracy: 0.9970 - val_head_2_loss: 0.0129 - val_head_3_accuracy: 0.9969 - val_head_3_loss: 0.0137 - val_head_4_accuracy: 0.9969 - val_head_4_loss: 0.0135 - val_head_5_accuracy: 0.9967 - val_head_5_loss: 0.0138 - val_head_6_accuracy: 0.9967 - val_head_6_loss: 0.0137 - val_head_7_accuracy: 0.9969 - val_head_7_loss: 0.0140 - val_loss: 0.0955 - learning_rate: 0.0145\n","\n","Epoch 43: LearningRateScheduler setting learning rate to 0.014268192016320704.\n","Epoch 43/100\n","\u001b[1m315/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m━\u001b[0m \u001b[1m0s\u001b[0m 32ms/step - head_1_accuracy: 0.9983 - head_1_loss: 0.0079 - head_2_accuracy: 0.9986 - head_2_loss: 0.0076 - head_3_accuracy: 0.9984 - head_3_loss: 0.0078 - head_4_accuracy: 0.9984 - head_4_loss: 0.0079 - head_5_accuracy: 0.9985 - head_5_loss: 0.0080 - head_6_accuracy: 0.9982 - head_6_loss: 0.0080 - head_7_accuracy: 0.9985 - head_7_loss: 0.0083 - loss: 0.0554\n","Epoch 43: val_head_1_accuracy did not improve from 0.99785\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m12s\u001b[0m 38ms/step - head_1_accuracy: 0.9986 - head_1_loss: 0.0064 - head_2_accuracy: 0.9987 - head_2_loss: 0.0060 - head_3_accuracy: 0.9986 - head_3_loss: 0.0063 - head_4_accuracy: 0.9986 - head_4_loss: 0.0063 - head_5_accuracy: 0.9986 - head_5_loss: 0.0065 - head_6_accuracy: 0.9985 - head_6_loss: 0.0066 - head_7_accuracy: 0.9985 - head_7_loss: 0.0068 - loss: 0.0449 - val_head_1_accuracy: 0.9978 - val_head_1_loss: 0.0072 - val_head_2_accuracy: 0.9980 - val_head_2_loss: 0.0071 - val_head_3_accuracy: 0.9979 - val_head_3_loss: 0.0073 - val_head_4_accuracy: 0.9979 - val_head_4_loss: 0.0072 - val_head_5_accuracy: 0.9981 - val_head_5_loss: 0.0072 - val_head_6_accuracy: 0.9979 - val_head_6_loss: 0.0071 - val_head_7_accuracy: 0.9980 - val_head_7_loss: 0.0069 - val_loss: 0.0504 - learning_rate: 0.0143\n","\n","Epoch 44: LearningRateScheduler setting learning rate to 0.01403605001048893.\n","Epoch 44/100\n","\u001b[1m315/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m━\u001b[0m \u001b[1m0s\u001b[0m 32ms/step - head_1_accuracy: 0.9995 - head_1_loss: 0.0028 - head_2_accuracy: 0.9995 - head_2_loss: 0.0028 - head_3_accuracy: 0.9995 - head_3_loss: 0.0028 - head_4_accuracy: 0.9995 - head_4_loss: 0.0027 - head_5_accuracy: 0.9994 - head_5_loss: 0.0029 - head_6_accuracy: 0.9992 - head_6_loss: 0.0029 - head_7_accuracy: 0.9994 - head_7_loss: 0.0030 - loss: 0.0200\n","Epoch 44: val_head_1_accuracy did not improve from 0.99785\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m12s\u001b[0m 37ms/step - head_1_accuracy: 0.9994 - head_1_loss: 0.0027 - head_2_accuracy: 0.9994 - head_2_loss: 0.0026 - head_3_accuracy: 0.9994 - head_3_loss: 0.0028 - head_4_accuracy: 0.9994 - head_4_loss: 0.0027 - head_5_accuracy: 0.9993 - head_5_loss: 0.0027 - head_6_accuracy: 0.9992 - head_6_loss: 0.0029 - head_7_accuracy: 0.9993 - head_7_loss: 0.0029 - loss: 0.0193 - val_head_1_accuracy: 0.9948 - val_head_1_loss: 0.0155 - val_head_2_accuracy: 0.9951 - val_head_2_loss: 0.0154 - val_head_3_accuracy: 0.9947 - val_head_3_loss: 0.0148 - val_head_4_accuracy: 0.9950 - val_head_4_loss: 0.0149 - val_head_5_accuracy: 0.9947 - val_head_5_loss: 0.0154 - val_head_6_accuracy: 0.9947 - val_head_6_loss: 0.0146 - val_head_7_accuracy: 0.9953 - val_head_7_loss: 0.0138 - val_loss: 0.1017 - learning_rate: 0.0140\n","\n","Epoch 45: LearningRateScheduler setting learning rate to 0.01380236133250198.\n","Epoch 45/100\n","\u001b[1m315/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m━\u001b[0m \u001b[1m0s\u001b[0m 32ms/step - head_1_accuracy: 0.9946 - head_1_loss: 0.0212 - head_2_accuracy: 0.9945 - head_2_loss: 0.0215 - head_3_accuracy: 0.9946 - head_3_loss: 0.0206 - head_4_accuracy: 0.9943 - head_4_loss: 0.0209 - head_5_accuracy: 0.9943 - head_5_loss: 0.0208 - head_6_accuracy: 0.9945 - head_6_loss: 0.0208 - head_7_accuracy: 0.9945 - head_7_loss: 0.0210 - loss: 0.1468\n","Epoch 45: val_head_1_accuracy improved from 0.99785 to 0.99829, saving model to /kaggle/working/results/models/AlexNet_CNN_MultiHead_best.keras\n","\n","Epoch 45: finished saving model to /kaggle/working/results/models/AlexNet_CNN_MultiHead_best.keras\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m13s\u001b[0m 42ms/step - head_1_accuracy: 0.9970 - head_1_loss: 0.0123 - head_2_accuracy: 0.9969 - head_2_loss: 0.0124 - head_3_accuracy: 0.9969 - head_3_loss: 0.0121 - head_4_accuracy: 0.9968 - head_4_loss: 0.0122 - head_5_accuracy: 0.9969 - head_5_loss: 0.0121 - head_6_accuracy: 0.9969 - head_6_loss: 0.0122 - head_7_accuracy: 0.9968 - head_7_loss: 0.0123 - loss: 0.0857 - val_head_1_accuracy: 0.9983 - val_head_1_loss: 0.0060 - val_head_2_accuracy: 0.9982 - val_head_2_loss: 0.0059 - val_head_3_accuracy: 0.9983 - val_head_3_loss: 0.0058 - val_head_4_accuracy: 0.9982 - val_head_4_loss: 0.0061 - val_head_5_accuracy: 0.9981 - val_head_5_loss: 0.0059 - val_head_6_accuracy: 0.9981 - val_head_6_loss: 0.0059 - val_head_7_accuracy: 0.9979 - val_head_7_loss: 0.0060 - val_loss: 0.0418 - learning_rate: 0.0138\n","\n","Epoch 46: LearningRateScheduler setting learning rate to 0.013567361287049638.\n","Epoch 46/100\n","\u001b[1m315/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m━\u001b[0m \u001b[1m0s\u001b[0m 32ms/step - head_1_accuracy: 0.9993 - head_1_loss: 0.0033 - head_2_accuracy: 0.9993 - head_2_loss: 0.0036 - head_3_accuracy: 0.9993 - head_3_loss: 0.0032 - head_4_accuracy: 0.9992 - head_4_loss: 0.0032 - head_5_accuracy: 0.9994 - head_5_loss: 0.0031 - head_6_accuracy: 0.9994 - head_6_loss: 0.0032 - head_7_accuracy: 0.9993 - head_7_loss: 0.0035 - loss: 0.0231\n","Epoch 46: val_head_1_accuracy improved from 0.99829 to 0.99881, saving model to /kaggle/working/results/models/AlexNet_CNN_MultiHead_best.keras\n","\n","Epoch 46: finished saving model to /kaggle/working/results/models/AlexNet_CNN_MultiHead_best.keras\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m13s\u001b[0m 42ms/step - head_1_accuracy: 0.9992 - head_1_loss: 0.0033 - head_2_accuracy: 0.9992 - head_2_loss: 0.0033 - head_3_accuracy: 0.9993 - head_3_loss: 0.0031 - head_4_accuracy: 0.9992 - head_4_loss: 0.0032 - head_5_accuracy: 0.9993 - head_5_loss: 0.0031 - head_6_accuracy: 0.9993 - head_6_loss: 0.0032 - head_7_accuracy: 0.9993 - head_7_loss: 0.0034 - loss: 0.0226 - val_head_1_accuracy: 0.9988 - val_head_1_loss: 0.0049 - val_head_2_accuracy: 0.9988 - val_head_2_loss: 0.0047 - val_head_3_accuracy: 0.9990 - val_head_3_loss: 0.0047 - val_head_4_accuracy: 0.9988 - val_head_4_loss: 0.0049 - val_head_5_accuracy: 0.9988 - val_head_5_loss: 0.0050 - val_head_6_accuracy: 0.9990 - val_head_6_loss: 0.0046 - val_head_7_accuracy: 0.9990 - val_head_7_loss: 0.0047 - val_loss: 0.0336 - learning_rate: 0.0136\n","\n","Epoch 47: LearningRateScheduler setting learning rate to 0.013331286499257584.\n","Epoch 47/100\n","\u001b[1m315/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m━\u001b[0m \u001b[1m0s\u001b[0m 32ms/step - head_1_accuracy: 0.9991 - head_1_loss: 0.0028 - head_2_accuracy: 0.9991 - head_2_loss: 0.0026 - head_3_accuracy: 0.9993 - head_3_loss: 0.0027 - head_4_accuracy: 0.9991 - head_4_loss: 0.0029 - head_5_accuracy: 0.9991 - head_5_loss: 0.0030 - head_6_accuracy: 0.9991 - head_6_loss: 0.0031 - head_7_accuracy: 0.9991 - head_7_loss: 0.0029 - loss: 0.0201\n","Epoch 47: val_head_1_accuracy did not improve from 0.99881\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m12s\u001b[0m 37ms/step - head_1_accuracy: 0.9995 - head_1_loss: 0.0022 - head_2_accuracy: 0.9994 - head_2_loss: 0.0021 - head_3_accuracy: 0.9995 - head_3_loss: 0.0022 - head_4_accuracy: 0.9994 - head_4_loss: 0.0022 - head_5_accuracy: 0.9993 - head_5_loss: 0.0023 - head_6_accuracy: 0.9994 - head_6_loss: 0.0024 - head_7_accuracy: 0.9993 - head_7_loss: 0.0024 - loss: 0.0158 - val_head_1_accuracy: 0.9970 - val_head_1_loss: 0.0082 - val_head_2_accuracy: 0.9973 - val_head_2_loss: 0.0082 - val_head_3_accuracy: 0.9973 - val_head_3_loss: 0.0076 - val_head_4_accuracy: 0.9974 - val_head_4_loss: 0.0078 - val_head_5_accuracy: 0.9972 - val_head_5_loss: 0.0081 - val_head_6_accuracy: 0.9974 - val_head_6_loss: 0.0075 - val_head_7_accuracy: 0.9974 - val_head_7_loss: 0.0071 - val_loss: 0.0548 - learning_rate: 0.0133\n","\n","Epoch 48: LearningRateScheduler setting learning rate to 0.013094374676425914.\n","Epoch 48/100\n","\u001b[1m315/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m━\u001b[0m \u001b[1m0s\u001b[0m 32ms/step - head_1_accuracy: 0.9995 - head_1_loss: 0.0018 - head_2_accuracy: 0.9995 - head_2_loss: 0.0018 - head_3_accuracy: 0.9996 - head_3_loss: 0.0016 - head_4_accuracy: 0.9994 - head_4_loss: 0.0019 - head_5_accuracy: 0.9994 - head_5_loss: 0.0017 - head_6_accuracy: 0.9993 - head_6_loss: 0.0018 - head_7_accuracy: 0.9997 - head_7_loss: 0.0018 - loss: 0.0123\n","Epoch 48: val_head_1_accuracy improved from 0.99881 to 0.99918, saving model to /kaggle/working/results/models/AlexNet_CNN_MultiHead_best.keras\n","\n","Epoch 48: finished saving model to /kaggle/working/results/models/AlexNet_CNN_MultiHead_best.keras\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m13s\u001b[0m 42ms/step - head_1_accuracy: 0.9996 - head_1_loss: 0.0017 - head_2_accuracy: 0.9996 - head_2_loss: 0.0017 - head_3_accuracy: 0.9997 - head_3_loss: 0.0016 - head_4_accuracy: 0.9995 - head_4_loss: 0.0017 - head_5_accuracy: 0.9996 - head_5_loss: 0.0017 - head_6_accuracy: 0.9995 - head_6_loss: 0.0018 - head_7_accuracy: 0.9997 - head_7_loss: 0.0017 - loss: 0.0118 - val_head_1_accuracy: 0.9992 - val_head_1_loss: 0.0035 - val_head_2_accuracy: 0.9992 - val_head_2_loss: 0.0034 - val_head_3_accuracy: 0.9992 - val_head_3_loss: 0.0034 - val_head_4_accuracy: 0.9991 - val_head_4_loss: 0.0036 - val_head_5_accuracy: 0.9991 - val_head_5_loss: 0.0035 - val_head_6_accuracy: 0.9992 - val_head_6_loss: 0.0034 - val_head_7_accuracy: 0.9990 - val_head_7_loss: 0.0036 - val_loss: 0.0246 - learning_rate: 0.0131\n","\n","Epoch 49: LearningRateScheduler setting learning rate to 0.01285686436867807.\n","Epoch 49/100\n","\u001b[1m315/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m━\u001b[0m \u001b[1m0s\u001b[0m 32ms/step - head_1_accuracy: 0.9998 - head_1_loss: 9.6323e-04 - head_2_accuracy: 0.9998 - head_2_loss: 8.6003e-04 - head_3_accuracy: 0.9999 - head_3_loss: 8.9105e-04 - head_4_accuracy: 0.9999 - head_4_loss: 9.9056e-04 - head_5_accuracy: 0.9999 - head_5_loss: 0.0010 - head_6_accuracy: 0.9997 - head_6_loss: 0.0011 - head_7_accuracy: 0.9999 - head_7_loss: 0.0011 - loss: 0.0069\n","Epoch 49: val_head_1_accuracy did not improve from 0.99918\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m12s\u001b[0m 37ms/step - head_1_accuracy: 0.9999 - head_1_loss: 9.3309e-04 - head_2_accuracy: 0.9999 - head_2_loss: 8.3821e-04 - head_3_accuracy: 0.9999 - head_3_loss: 8.8813e-04 - head_4_accuracy: 0.9999 - head_4_loss: 9.4773e-04 - head_5_accuracy: 0.9998 - head_5_loss: 9.9475e-04 - head_6_accuracy: 0.9998 - head_6_loss: 0.0011 - head_7_accuracy: 0.9999 - head_7_loss: 0.0011 - loss: 0.0068 - val_head_1_accuracy: 0.9988 - val_head_1_loss: 0.0043 - val_head_2_accuracy: 0.9989 - val_head_2_loss: 0.0041 - val_head_3_accuracy: 0.9989 - val_head_3_loss: 0.0041 - val_head_4_accuracy: 0.9989 - val_head_4_loss: 0.0041 - val_head_5_accuracy: 0.9990 - val_head_5_loss: 0.0043 - val_head_6_accuracy: 0.9990 - val_head_6_loss: 0.0040 - val_head_7_accuracy: 0.9990 - val_head_7_loss: 0.0040 - val_loss: 0.0291 - learning_rate: 0.0129\n","\n","Epoch 50: LearningRateScheduler setting learning rate to 0.012618994728761062.\n","Epoch 50/100\n","\u001b[1m315/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m━\u001b[0m \u001b[1m0s\u001b[0m 32ms/step - head_1_accuracy: 0.9998 - head_1_loss: 0.0013 - head_2_accuracy: 0.9997 - head_2_loss: 0.0013 - head_3_accuracy: 0.9997 - head_3_loss: 0.0012 - head_4_accuracy: 0.9998 - head_4_loss: 0.0013 - head_5_accuracy: 0.9998 - head_5_loss: 0.0013 - head_6_accuracy: 0.9996 - head_6_loss: 0.0015 - head_7_accuracy: 0.9995 - head_7_loss: 0.0015 - loss: 0.0092\n","Epoch 50: val_head_1_accuracy improved from 0.99918 to 0.99933, saving model to /kaggle/working/results/models/AlexNet_CNN_MultiHead_best.keras\n","\n","Epoch 50: finished saving model to /kaggle/working/results/models/AlexNet_CNN_MultiHead_best.keras\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m13s\u001b[0m 42ms/step - head_1_accuracy: 0.9999 - head_1_loss: 8.4072e-04 - head_2_accuracy: 0.9999 - head_2_loss: 7.7518e-04 - head_3_accuracy: 0.9999 - head_3_loss: 7.8820e-04 - head_4_accuracy: 0.9999 - head_4_loss: 8.5741e-04 - head_5_accuracy: 0.9999 - head_5_loss: 8.2994e-04 - head_6_accuracy: 0.9998 - head_6_loss: 9.7770e-04 - head_7_accuracy: 0.9998 - head_7_loss: 9.7175e-04 - loss: 0.0060 - val_head_1_accuracy: 0.9993 - val_head_1_loss: 0.0025 - val_head_2_accuracy: 0.9993 - val_head_2_loss: 0.0024 - val_head_3_accuracy: 0.9992 - val_head_3_loss: 0.0025 - val_head_4_accuracy: 0.9994 - val_head_4_loss: 0.0024 - val_head_5_accuracy: 0.9995 - val_head_5_loss: 0.0026 - val_head_6_accuracy: 0.9994 - val_head_6_loss: 0.0023 - val_head_7_accuracy: 0.9993 - val_head_7_loss: 0.0024 - val_loss: 0.0172 - learning_rate: 0.0126\n","\n","Epoch 51: LearningRateScheduler setting learning rate to 0.01238100527123894.\n","Epoch 51/100\n","\u001b[1m315/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m━\u001b[0m \u001b[1m0s\u001b[0m 32ms/step - head_1_accuracy: 0.9998 - head_1_loss: 8.0918e-04 - head_2_accuracy: 1.0000 - head_2_loss: 6.1519e-04 - head_3_accuracy: 1.0000 - head_3_loss: 6.4947e-04 - head_4_accuracy: 1.0000 - head_4_loss: 6.7019e-04 - head_5_accuracy: 0.9998 - head_5_loss: 8.0432e-04 - head_6_accuracy: 0.9999 - head_6_loss: 8.1044e-04 - head_7_accuracy: 1.0000 - head_7_loss: 7.0175e-04 - loss: 0.0051\n","Epoch 51: val_head_1_accuracy did not improve from 0.99933\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m12s\u001b[0m 37ms/step - head_1_accuracy: 0.9999 - head_1_loss: 7.5044e-04 - head_2_accuracy: 1.0000 - head_2_loss: 6.1778e-04 - head_3_accuracy: 1.0000 - head_3_loss: 6.6972e-04 - head_4_accuracy: 1.0000 - head_4_loss: 6.7541e-04 - head_5_accuracy: 0.9999 - head_5_loss: 7.6989e-04 - head_6_accuracy: 0.9999 - head_6_loss: 7.7179e-04 - head_7_accuracy: 1.0000 - head_7_loss: 7.4814e-04 - loss: 0.0050 - val_head_1_accuracy: 0.9989 - val_head_1_loss: 0.0046 - val_head_2_accuracy: 0.9989 - val_head_2_loss: 0.0045 - val_head_3_accuracy: 0.9990 - val_head_3_loss: 0.0043 - val_head_4_accuracy: 0.9990 - val_head_4_loss: 0.0045 - val_head_5_accuracy: 0.9990 - val_head_5_loss: 0.0047 - val_head_6_accuracy: 0.9990 - val_head_6_loss: 0.0045 - val_head_7_accuracy: 0.9990 - val_head_7_loss: 0.0043 - val_loss: 0.0317 - learning_rate: 0.0124\n","\n","Epoch 52: LearningRateScheduler setting learning rate to 0.012143135631321932.\n","Epoch 52/100\n","\u001b[1m315/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m━\u001b[0m \u001b[1m0s\u001b[0m 32ms/step - head_1_accuracy: 1.0000 - head_1_loss: 5.8751e-04 - head_2_accuracy: 1.0000 - head_2_loss: 4.8631e-04 - head_3_accuracy: 1.0000 - head_3_loss: 5.0604e-04 - head_4_accuracy: 1.0000 - head_4_loss: 5.6895e-04 - head_5_accuracy: 1.0000 - head_5_loss: 5.4005e-04 - head_6_accuracy: 0.9999 - head_6_loss: 6.0201e-04 - head_7_accuracy: 0.9999 - head_7_loss: 6.7523e-04 - loss: 0.0040\n","Epoch 52: val_head_1_accuracy did not improve from 0.99933\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m12s\u001b[0m 37ms/step - head_1_accuracy: 1.0000 - head_1_loss: 5.5133e-04 - head_2_accuracy: 1.0000 - head_2_loss: 5.0987e-04 - head_3_accuracy: 1.0000 - head_3_loss: 5.3542e-04 - head_4_accuracy: 0.9999 - head_4_loss: 5.5331e-04 - head_5_accuracy: 1.0000 - head_5_loss: 5.2102e-04 - head_6_accuracy: 0.9999 - head_6_loss: 5.8370e-04 - head_7_accuracy: 0.9999 - head_7_loss: 6.4945e-04 - loss: 0.0039 - val_head_1_accuracy: 0.9993 - val_head_1_loss: 0.0032 - val_head_2_accuracy: 0.9992 - val_head_2_loss: 0.0032 - val_head_3_accuracy: 0.9991 - val_head_3_loss: 0.0032 - val_head_4_accuracy: 0.9991 - val_head_4_loss: 0.0031 - val_head_5_accuracy: 0.9992 - val_head_5_loss: 0.0032 - val_head_6_accuracy: 0.9992 - val_head_6_loss: 0.0030 - val_head_7_accuracy: 0.9992 - val_head_7_loss: 0.0031 - val_loss: 0.0222 - learning_rate: 0.0121\n","\n","Epoch 53: LearningRateScheduler setting learning rate to 0.011905625323574085.\n","Epoch 53/100\n","\u001b[1m315/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m━\u001b[0m \u001b[1m0s\u001b[0m 32ms/step - head_1_accuracy: 0.9998 - head_1_loss: 7.1693e-04 - head_2_accuracy: 0.9998 - head_2_loss: 5.6689e-04 - head_3_accuracy: 0.9998 - head_3_loss: 6.9346e-04 - head_4_accuracy: 0.9997 - head_4_loss: 8.6478e-04 - head_5_accuracy: 0.9998 - head_5_loss: 7.8663e-04 - head_6_accuracy: 0.9997 - head_6_loss: 8.0238e-04 - head_7_accuracy: 0.9998 - head_7_loss: 8.7916e-04 - loss: 0.0053\n","Epoch 53: val_head_1_accuracy did not improve from 0.99933\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m12s\u001b[0m 37ms/step - head_1_accuracy: 0.9999 - head_1_loss: 5.6269e-04 - head_2_accuracy: 0.9999 - head_2_loss: 4.9383e-04 - head_3_accuracy: 0.9999 - head_3_loss: 5.5421e-04 - head_4_accuracy: 0.9999 - head_4_loss: 6.1751e-04 - head_5_accuracy: 0.9998 - head_5_loss: 7.0662e-04 - head_6_accuracy: 0.9999 - head_6_loss: 6.4336e-04 - head_7_accuracy: 0.9999 - head_7_loss: 6.8251e-04 - loss: 0.0043 - val_head_1_accuracy: 0.9992 - val_head_1_loss: 0.0023 - val_head_2_accuracy: 0.9993 - val_head_2_loss: 0.0022 - val_head_3_accuracy: 0.9992 - val_head_3_loss: 0.0022 - val_head_4_accuracy: 0.9993 - val_head_4_loss: 0.0022 - val_head_5_accuracy: 0.9993 - val_head_5_loss: 0.0023 - val_head_6_accuracy: 0.9994 - val_head_6_loss: 0.0022 - val_head_7_accuracy: 0.9993 - val_head_7_loss: 0.0022 - val_loss: 0.0157 - learning_rate: 0.0119\n","\n","Epoch 54: LearningRateScheduler setting learning rate to 0.011668713500742417.\n","Epoch 54/100\n","\u001b[1m315/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m━\u001b[0m \u001b[1m0s\u001b[0m 32ms/step - head_1_accuracy: 0.9999 - head_1_loss: 6.3555e-04 - head_2_accuracy: 0.9999 - head_2_loss: 6.1072e-04 - head_3_accuracy: 0.9999 - head_3_loss: 6.3238e-04 - head_4_accuracy: 0.9999 - head_4_loss: 6.2637e-04 - head_5_accuracy: 0.9999 - head_5_loss: 6.2502e-04 - head_6_accuracy: 0.9999 - head_6_loss: 6.6072e-04 - head_7_accuracy: 0.9998 - head_7_loss: 7.1861e-04 - loss: 0.0045\n","Epoch 54: val_head_1_accuracy did not improve from 0.99933\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m12s\u001b[0m 37ms/step - head_1_accuracy: 0.9999 - head_1_loss: 8.1234e-04 - head_2_accuracy: 0.9999 - head_2_loss: 8.2765e-04 - head_3_accuracy: 0.9999 - head_3_loss: 8.1072e-04 - head_4_accuracy: 0.9999 - head_4_loss: 8.1413e-04 - head_5_accuracy: 0.9999 - head_5_loss: 8.2050e-04 - head_6_accuracy: 0.9998 - head_6_loss: 8.6282e-04 - head_7_accuracy: 0.9998 - head_7_loss: 8.5536e-04 - loss: 0.0058 - val_head_1_accuracy: 0.9992 - val_head_1_loss: 0.0031 - val_head_2_accuracy: 0.9990 - val_head_2_loss: 0.0031 - val_head_3_accuracy: 0.9991 - val_head_3_loss: 0.0031 - val_head_4_accuracy: 0.9991 - val_head_4_loss: 0.0030 - val_head_5_accuracy: 0.9992 - val_head_5_loss: 0.0032 - val_head_6_accuracy: 0.9991 - val_head_6_loss: 0.0028 - val_head_7_accuracy: 0.9992 - val_head_7_loss: 0.0030 - val_loss: 0.0215 - learning_rate: 0.0117\n","\n","Epoch 55: LearningRateScheduler setting learning rate to 0.011432638712950362.\n","Epoch 55/100\n","\u001b[1m315/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m━\u001b[0m \u001b[1m0s\u001b[0m 32ms/step - head_1_accuracy: 0.9994 - head_1_loss: 0.0022 - head_2_accuracy: 0.9994 - head_2_loss: 0.0020 - head_3_accuracy: 0.9994 - head_3_loss: 0.0020 - head_4_accuracy: 0.9995 - head_4_loss: 0.0019 - head_5_accuracy: 0.9995 - head_5_loss: 0.0020 - head_6_accuracy: 0.9993 - head_6_loss: 0.0021 - head_7_accuracy: 0.9993 - head_7_loss: 0.0021 - loss: 0.0143\n","Epoch 55: val_head_1_accuracy did not improve from 0.99933\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m12s\u001b[0m 37ms/step - head_1_accuracy: 0.9995 - head_1_loss: 0.0019 - head_2_accuracy: 0.9994 - head_2_loss: 0.0018 - head_3_accuracy: 0.9995 - head_3_loss: 0.0018 - head_4_accuracy: 0.9995 - head_4_loss: 0.0017 - head_5_accuracy: 0.9995 - head_5_loss: 0.0018 - head_6_accuracy: 0.9994 - head_6_loss: 0.0019 - head_7_accuracy: 0.9994 - head_7_loss: 0.0019 - loss: 0.0130 - val_head_1_accuracy: 0.9990 - val_head_1_loss: 0.0034 - val_head_2_accuracy: 0.9991 - val_head_2_loss: 0.0033 - val_head_3_accuracy: 0.9992 - val_head_3_loss: 0.0033 - val_head_4_accuracy: 0.9991 - val_head_4_loss: 0.0033 - val_head_5_accuracy: 0.9990 - val_head_5_loss: 0.0033 - val_head_6_accuracy: 0.9991 - val_head_6_loss: 0.0033 - val_head_7_accuracy: 0.9991 - val_head_7_loss: 0.0033 - val_loss: 0.0233 - learning_rate: 0.0114\n","\n","Epoch 56: LearningRateScheduler setting learning rate to 0.011197638667498021.\n","Epoch 56/100\n","\u001b[1m315/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m━\u001b[0m \u001b[1m0s\u001b[0m 32ms/step - head_1_accuracy: 0.9998 - head_1_loss: 9.5946e-04 - head_2_accuracy: 0.9997 - head_2_loss: 0.0011 - head_3_accuracy: 0.9998 - head_3_loss: 9.9742e-04 - head_4_accuracy: 0.9998 - head_4_loss: 0.0010 - head_5_accuracy: 0.9998 - head_5_loss: 0.0010 - head_6_accuracy: 0.9997 - head_6_loss: 0.0011 - head_7_accuracy: 0.9998 - head_7_loss: 0.0010 - loss: 0.0072\n","Epoch 56: val_head_1_accuracy improved from 0.99933 to 0.99941, saving model to /kaggle/working/results/models/AlexNet_CNN_MultiHead_best.keras\n","\n","Epoch 56: finished saving model to /kaggle/working/results/models/AlexNet_CNN_MultiHead_best.keras\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m13s\u001b[0m 42ms/step - head_1_accuracy: 0.9999 - head_1_loss: 7.5266e-04 - head_2_accuracy: 0.9998 - head_2_loss: 8.0983e-04 - head_3_accuracy: 0.9999 - head_3_loss: 7.4436e-04 - head_4_accuracy: 0.9999 - head_4_loss: 7.6874e-04 - head_5_accuracy: 0.9999 - head_5_loss: 7.6982e-04 - head_6_accuracy: 0.9998 - head_6_loss: 8.6375e-04 - head_7_accuracy: 0.9999 - head_7_loss: 7.9232e-04 - loss: 0.0055 - val_head_1_accuracy: 0.9994 - val_head_1_loss: 0.0021 - val_head_2_accuracy: 0.9995 - val_head_2_loss: 0.0020 - val_head_3_accuracy: 0.9993 - val_head_3_loss: 0.0021 - val_head_4_accuracy: 0.9993 - val_head_4_loss: 0.0020 - val_head_5_accuracy: 0.9996 - val_head_5_loss: 0.0021 - val_head_6_accuracy: 0.9994 - val_head_6_loss: 0.0020 - val_head_7_accuracy: 0.9995 - val_head_7_loss: 0.0021 - val_loss: 0.0145 - learning_rate: 0.0112\n","\n","Epoch 57: LearningRateScheduler setting learning rate to 0.010963949989511071.\n","Epoch 57/100\n","\u001b[1m315/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m━\u001b[0m \u001b[1m0s\u001b[0m 32ms/step - head_1_accuracy: 0.9997 - head_1_loss: 0.0014 - head_2_accuracy: 0.9996 - head_2_loss: 0.0014 - head_3_accuracy: 0.9998 - head_3_loss: 0.0014 - head_4_accuracy: 0.9997 - head_4_loss: 0.0014 - head_5_accuracy: 0.9997 - head_5_loss: 0.0014 - head_6_accuracy: 0.9997 - head_6_loss: 0.0015 - head_7_accuracy: 0.9998 - head_7_loss: 0.0014 - loss: 0.0099\n","Epoch 57: val_head_1_accuracy did not improve from 0.99941\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m12s\u001b[0m 37ms/step - head_1_accuracy: 0.9997 - head_1_loss: 0.0015 - head_2_accuracy: 0.9997 - head_2_loss: 0.0016 - head_3_accuracy: 0.9997 - head_3_loss: 0.0015 - head_4_accuracy: 0.9997 - head_4_loss: 0.0015 - head_5_accuracy: 0.9997 - head_5_loss: 0.0015 - head_6_accuracy: 0.9997 - head_6_loss: 0.0015 - head_7_accuracy: 0.9997 - head_7_loss: 0.0015 - loss: 0.0107 - val_head_1_accuracy: 0.9990 - val_head_1_loss: 0.0030 - val_head_2_accuracy: 0.9991 - val_head_2_loss: 0.0029 - val_head_3_accuracy: 0.9990 - val_head_3_loss: 0.0030 - val_head_4_accuracy: 0.9992 - val_head_4_loss: 0.0028 - val_head_5_accuracy: 0.9992 - val_head_5_loss: 0.0029 - val_head_6_accuracy: 0.9991 - val_head_6_loss: 0.0028 - val_head_7_accuracy: 0.9991 - val_head_7_loss: 0.0030 - val_loss: 0.0205 - learning_rate: 0.0110\n","\n","Epoch 58: LearningRateScheduler setting learning rate to 0.010731807983679297.\n","Epoch 58/100\n","\u001b[1m315/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m━\u001b[0m \u001b[1m0s\u001b[0m 32ms/step - head_1_accuracy: 0.9995 - head_1_loss: 0.0014 - head_2_accuracy: 0.9996 - head_2_loss: 0.0014 - head_3_accuracy: 0.9996 - head_3_loss: 0.0014 - head_4_accuracy: 0.9996 - head_4_loss: 0.0013 - head_5_accuracy: 0.9994 - head_5_loss: 0.0014 - head_6_accuracy: 0.9996 - head_6_loss: 0.0015 - head_7_accuracy: 0.9997 - head_7_loss: 0.0015 - loss: 0.0099\n","Epoch 58: val_head_1_accuracy did not improve from 0.99941\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m12s\u001b[0m 37ms/step - head_1_accuracy: 0.9998 - head_1_loss: 8.7016e-04 - head_2_accuracy: 0.9998 - head_2_loss: 8.8333e-04 - head_3_accuracy: 0.9998 - head_3_loss: 8.4274e-04 - head_4_accuracy: 0.9998 - head_4_loss: 8.4919e-04 - head_5_accuracy: 0.9997 - head_5_loss: 9.2463e-04 - head_6_accuracy: 0.9998 - head_6_loss: 9.8416e-04 - head_7_accuracy: 0.9999 - head_7_loss: 9.9682e-04 - loss: 0.0064 - val_head_1_accuracy: 0.9992 - val_head_1_loss: 0.0019 - val_head_2_accuracy: 0.9992 - val_head_2_loss: 0.0018 - val_head_3_accuracy: 0.9992 - val_head_3_loss: 0.0019 - val_head_4_accuracy: 0.9993 - val_head_4_loss: 0.0018 - val_head_5_accuracy: 0.9993 - val_head_5_loss: 0.0019 - val_head_6_accuracy: 0.9993 - val_head_6_loss: 0.0017 - val_head_7_accuracy: 0.9992 - val_head_7_loss: 0.0018 - val_loss: 0.0129 - learning_rate: 0.0107\n","\n","Epoch 59: LearningRateScheduler setting learning rate to 0.010501446397324737.\n","Epoch 59/100\n","\u001b[1m315/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m━\u001b[0m \u001b[1m0s\u001b[0m 32ms/step - head_1_accuracy: 0.9998 - head_1_loss: 5.4532e-04 - head_2_accuracy: 0.9999 - head_2_loss: 5.1439e-04 - head_3_accuracy: 0.9999 - head_3_loss: 4.9318e-04 - head_4_accuracy: 0.9999 - head_4_loss: 5.8812e-04 - head_5_accuracy: 0.9998 - head_5_loss: 5.3503e-04 - head_6_accuracy: 0.9998 - head_6_loss: 6.8425e-04 - head_7_accuracy: 0.9999 - head_7_loss: 6.9090e-04 - loss: 0.0041\n","Epoch 59: val_head_1_accuracy did not improve from 0.99941\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m12s\u001b[0m 37ms/step - head_1_accuracy: 0.9999 - head_1_loss: 4.8648e-04 - head_2_accuracy: 0.9999 - head_2_loss: 4.2629e-04 - head_3_accuracy: 0.9999 - head_3_loss: 4.6046e-04 - head_4_accuracy: 1.0000 - head_4_loss: 4.8750e-04 - head_5_accuracy: 0.9999 - head_5_loss: 4.9026e-04 - head_6_accuracy: 0.9999 - head_6_loss: 5.6086e-04 - head_7_accuracy: 0.9999 - head_7_loss: 5.2517e-04 - loss: 0.0034 - val_head_1_accuracy: 0.9993 - val_head_1_loss: 0.0023 - val_head_2_accuracy: 0.9993 - val_head_2_loss: 0.0023 - val_head_3_accuracy: 0.9993 - val_head_3_loss: 0.0023 - val_head_4_accuracy: 0.9993 - val_head_4_loss: 0.0023 - val_head_5_accuracy: 0.9994 - val_head_5_loss: 0.0023 - val_head_6_accuracy: 0.9993 - val_head_6_loss: 0.0022 - val_head_7_accuracy: 0.9994 - val_head_7_loss: 0.0023 - val_loss: 0.0160 - learning_rate: 0.0105\n","\n","Epoch 60: LearningRateScheduler setting learning rate to 0.010273097185037938.\n","Epoch 60/100\n","\u001b[1m315/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m━\u001b[0m \u001b[1m0s\u001b[0m 32ms/step - head_1_accuracy: 0.9999 - head_1_loss: 4.6880e-04 - head_2_accuracy: 1.0000 - head_2_loss: 4.0420e-04 - head_3_accuracy: 1.0000 - head_3_loss: 3.9911e-04 - head_4_accuracy: 0.9999 - head_4_loss: 4.2216e-04 - head_5_accuracy: 1.0000 - head_5_loss: 4.3730e-04 - head_6_accuracy: 0.9999 - head_6_loss: 5.1562e-04 - head_7_accuracy: 1.0000 - head_7_loss: 4.5872e-04 - loss: 0.0031\n","Epoch 60: val_head_1_accuracy did not improve from 0.99941\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m12s\u001b[0m 37ms/step - head_1_accuracy: 0.9999 - head_1_loss: 6.3023e-04 - head_2_accuracy: 0.9999 - head_2_loss: 5.8812e-04 - head_3_accuracy: 0.9999 - head_3_loss: 5.9812e-04 - head_4_accuracy: 0.9999 - head_4_loss: 5.9497e-04 - head_5_accuracy: 0.9999 - head_5_loss: 5.8078e-04 - head_6_accuracy: 0.9999 - head_6_loss: 6.9097e-04 - head_7_accuracy: 0.9999 - head_7_loss: 6.1426e-04 - loss: 0.0043 - val_head_1_accuracy: 0.9992 - val_head_1_loss: 0.0026 - val_head_2_accuracy: 0.9993 - val_head_2_loss: 0.0026 - val_head_3_accuracy: 0.9993 - val_head_3_loss: 0.0025 - val_head_4_accuracy: 0.9992 - val_head_4_loss: 0.0026 - val_head_5_accuracy: 0.9992 - val_head_5_loss: 0.0026 - val_head_6_accuracy: 0.9991 - val_head_6_loss: 0.0025 - val_head_7_accuracy: 0.9992 - val_head_7_loss: 0.0026 - val_loss: 0.0182 - learning_rate: 0.0103\n","\n","Epoch 61: LearningRateScheduler setting learning rate to 0.01004699027511934.\n","Epoch 61/100\n","\u001b[1m315/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m━\u001b[0m \u001b[1m0s\u001b[0m 32ms/step - head_1_accuracy: 1.0000 - head_1_loss: 4.2878e-04 - head_2_accuracy: 1.0000 - head_2_loss: 3.6611e-04 - head_3_accuracy: 1.0000 - head_3_loss: 3.9830e-04 - head_4_accuracy: 1.0000 - head_4_loss: 4.2554e-04 - head_5_accuracy: 1.0000 - head_5_loss: 4.3965e-04 - head_6_accuracy: 0.9999 - head_6_loss: 5.3246e-04 - head_7_accuracy: 1.0000 - head_7_loss: 4.4243e-04 - loss: 0.0030\n","Epoch 61: val_head_1_accuracy did not improve from 0.99941\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m12s\u001b[0m 37ms/step - head_1_accuracy: 1.0000 - head_1_loss: 4.0817e-04 - head_2_accuracy: 1.0000 - head_2_loss: 3.6087e-04 - head_3_accuracy: 0.9999 - head_3_loss: 3.9788e-04 - head_4_accuracy: 1.0000 - head_4_loss: 4.1619e-04 - head_5_accuracy: 1.0000 - head_5_loss: 4.3826e-04 - head_6_accuracy: 0.9999 - head_6_loss: 4.7535e-04 - head_7_accuracy: 1.0000 - head_7_loss: 4.6327e-04 - loss: 0.0030 - val_head_1_accuracy: 0.9993 - val_head_1_loss: 0.0019 - val_head_2_accuracy: 0.9993 - val_head_2_loss: 0.0019 - val_head_3_accuracy: 0.9993 - val_head_3_loss: 0.0018 - val_head_4_accuracy: 0.9994 - val_head_4_loss: 0.0019 - val_head_5_accuracy: 0.9994 - val_head_5_loss: 0.0019 - val_head_6_accuracy: 0.9994 - val_head_6_loss: 0.0019 - val_head_7_accuracy: 0.9994 - val_head_7_loss: 0.0019 - val_loss: 0.0134 - learning_rate: 0.0100\n","\n","Epoch 62: LearningRateScheduler setting learning rate to 0.009823353338060961.\n","Epoch 62/100\n","\u001b[1m315/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m━\u001b[0m \u001b[1m0s\u001b[0m 32ms/step - head_1_accuracy: 0.9999 - head_1_loss: 4.0086e-04 - head_2_accuracy: 0.9998 - head_2_loss: 4.1911e-04 - head_3_accuracy: 0.9999 - head_3_loss: 3.4261e-04 - head_4_accuracy: 0.9999 - head_4_loss: 3.4151e-04 - head_5_accuracy: 0.9999 - head_5_loss: 3.9242e-04 - head_6_accuracy: 0.9999 - head_6_loss: 4.7480e-04 - head_7_accuracy: 0.9999 - head_7_loss: 4.4382e-04 - loss: 0.0028\n","Epoch 62: val_head_1_accuracy did not improve from 0.99941\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m12s\u001b[0m 37ms/step - head_1_accuracy: 0.9997 - head_1_loss: 9.4958e-04 - head_2_accuracy: 0.9997 - head_2_loss: 9.7287e-04 - head_3_accuracy: 0.9997 - head_3_loss: 9.4168e-04 - head_4_accuracy: 0.9998 - head_4_loss: 9.1824e-04 - head_5_accuracy: 0.9998 - head_5_loss: 8.7501e-04 - head_6_accuracy: 0.9997 - head_6_loss: 9.3600e-04 - head_7_accuracy: 0.9997 - head_7_loss: 9.7865e-04 - loss: 0.0066 - val_head_1_accuracy: 0.9989 - val_head_1_loss: 0.0033 - val_head_2_accuracy: 0.9989 - val_head_2_loss: 0.0037 - val_head_3_accuracy: 0.9990 - val_head_3_loss: 0.0029 - val_head_4_accuracy: 0.9990 - val_head_4_loss: 0.0030 - val_head_5_accuracy: 0.9991 - val_head_5_loss: 0.0030 - val_head_6_accuracy: 0.9991 - val_head_6_loss: 0.0030 - val_head_7_accuracy: 0.9991 - val_head_7_loss: 0.0029 - val_loss: 0.0221 - learning_rate: 0.0098\n","\n","Epoch 63: LearningRateScheduler setting learning rate to 0.009602411557301536.\n","Epoch 63/100\n","\u001b[1m315/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m━\u001b[0m \u001b[1m0s\u001b[0m 32ms/step - head_1_accuracy: 0.9998 - head_1_loss: 6.7412e-04 - head_2_accuracy: 0.9999 - head_2_loss: 6.6007e-04 - head_3_accuracy: 0.9998 - head_3_loss: 5.9746e-04 - head_4_accuracy: 0.9999 - head_4_loss: 5.6370e-04 - head_5_accuracy: 0.9998 - head_5_loss: 7.1480e-04 - head_6_accuracy: 0.9999 - head_6_loss: 7.6796e-04 - head_7_accuracy: 0.9999 - head_7_loss: 6.4142e-04 - loss: 0.0046\n","Epoch 63: val_head_1_accuracy did not improve from 0.99941\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m12s\u001b[0m 37ms/step - head_1_accuracy: 0.9999 - head_1_loss: 5.8453e-04 - head_2_accuracy: 0.9999 - head_2_loss: 5.6711e-04 - head_3_accuracy: 0.9999 - head_3_loss: 5.6795e-04 - head_4_accuracy: 0.9999 - head_4_loss: 5.4602e-04 - head_5_accuracy: 0.9999 - head_5_loss: 7.1253e-04 - head_6_accuracy: 0.9999 - head_6_loss: 6.9848e-04 - head_7_accuracy: 0.9999 - head_7_loss: 7.0270e-04 - loss: 0.0044 - val_head_1_accuracy: 0.9990 - val_head_1_loss: 0.0031 - val_head_2_accuracy: 0.9990 - val_head_2_loss: 0.0032 - val_head_3_accuracy: 0.9991 - val_head_3_loss: 0.0030 - val_head_4_accuracy: 0.9990 - val_head_4_loss: 0.0029 - val_head_5_accuracy: 0.9992 - val_head_5_loss: 0.0030 - val_head_6_accuracy: 0.9990 - val_head_6_loss: 0.0028 - val_head_7_accuracy: 0.9991 - val_head_7_loss: 0.0030 - val_loss: 0.0211 - learning_rate: 0.0096\n","\n","Epoch 64: LearningRateScheduler setting learning rate to 0.009384387402485851.\n","Epoch 64/100\n","\u001b[1m315/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m━\u001b[0m \u001b[1m0s\u001b[0m 33ms/step - head_1_accuracy: 1.0000 - head_1_loss: 4.3292e-04 - head_2_accuracy: 0.9999 - head_2_loss: 4.0506e-04 - head_3_accuracy: 0.9999 - head_3_loss: 4.7760e-04 - head_4_accuracy: 0.9999 - head_4_loss: 4.4677e-04 - head_5_accuracy: 0.9998 - head_5_loss: 5.8644e-04 - head_6_accuracy: 0.9999 - head_6_loss: 4.9987e-04 - head_7_accuracy: 0.9998 - head_7_loss: 5.4055e-04 - loss: 0.0034\n","Epoch 64: val_head_1_accuracy did not improve from 0.99941\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m12s\u001b[0m 38ms/step - head_1_accuracy: 0.9999 - head_1_loss: 4.7678e-04 - head_2_accuracy: 0.9999 - head_2_loss: 4.6956e-04 - head_3_accuracy: 0.9998 - head_3_loss: 5.0697e-04 - head_4_accuracy: 0.9999 - head_4_loss: 4.5258e-04 - head_5_accuracy: 0.9999 - head_5_loss: 5.4443e-04 - head_6_accuracy: 0.9999 - head_6_loss: 5.5740e-04 - head_7_accuracy: 0.9999 - head_7_loss: 5.4329e-04 - loss: 0.0036 - val_head_1_accuracy: 0.9993 - val_head_1_loss: 0.0023 - val_head_2_accuracy: 0.9993 - val_head_2_loss: 0.0023 - val_head_3_accuracy: 0.9994 - val_head_3_loss: 0.0022 - val_head_4_accuracy: 0.9993 - val_head_4_loss: 0.0023 - val_head_5_accuracy: 0.9996 - val_head_5_loss: 0.0023 - val_head_6_accuracy: 0.9994 - val_head_6_loss: 0.0022 - val_head_7_accuracy: 0.9993 - val_head_7_loss: 0.0022 - val_loss: 0.0158 - learning_rate: 0.0094\n","\n","Epoch 65: LearningRateScheduler setting learning rate to 0.009169500405456693.\n","Epoch 65/100\n","\u001b[1m315/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m━\u001b[0m \u001b[1m0s\u001b[0m 32ms/step - head_1_accuracy: 1.0000 - head_1_loss: 4.7953e-04 - head_2_accuracy: 0.9998 - head_2_loss: 5.7174e-04 - head_3_accuracy: 0.9999 - head_3_loss: 4.5258e-04 - head_4_accuracy: 0.9999 - head_4_loss: 4.9780e-04 - head_5_accuracy: 0.9999 - head_5_loss: 5.1948e-04 - head_6_accuracy: 0.9999 - head_6_loss: 6.1022e-04 - head_7_accuracy: 0.9999 - head_7_loss: 5.0842e-04 - loss: 0.0036\n","Epoch 65: val_head_1_accuracy improved from 0.99941 to 0.99948, saving model to /kaggle/working/results/models/AlexNet_CNN_MultiHead_best.keras\n","\n","Epoch 65: finished saving model to /kaggle/working/results/models/AlexNet_CNN_MultiHead_best.keras\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m13s\u001b[0m 42ms/step - head_1_accuracy: 1.0000 - head_1_loss: 3.3751e-04 - head_2_accuracy: 0.9999 - head_2_loss: 3.2780e-04 - head_3_accuracy: 1.0000 - head_3_loss: 2.9951e-04 - head_4_accuracy: 1.0000 - head_4_loss: 3.4781e-04 - head_5_accuracy: 1.0000 - head_5_loss: 3.4316e-04 - head_6_accuracy: 1.0000 - head_6_loss: 3.8968e-04 - head_7_accuracy: 1.0000 - head_7_loss: 3.5260e-04 - loss: 0.0024 - val_head_1_accuracy: 0.9995 - val_head_1_loss: 0.0019 - val_head_2_accuracy: 0.9995 - val_head_2_loss: 0.0018 - val_head_3_accuracy: 0.9996 - val_head_3_loss: 0.0018 - val_head_4_accuracy: 0.9995 - val_head_4_loss: 0.0018 - val_head_5_accuracy: 0.9995 - val_head_5_loss: 0.0019 - val_head_6_accuracy: 0.9995 - val_head_6_loss: 0.0017 - val_head_7_accuracy: 0.9995 - val_head_7_loss: 0.0018 - val_loss: 0.0129 - learning_rate: 0.0092\n","\n","Epoch 66: LearningRateScheduler setting learning rate to 0.008957966939204882.\n","Epoch 66/100\n","\u001b[1m315/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m━\u001b[0m \u001b[1m0s\u001b[0m 32ms/step - head_1_accuracy: 1.0000 - head_1_loss: 3.3695e-04 - head_2_accuracy: 0.9999 - head_2_loss: 3.2741e-04 - head_3_accuracy: 1.0000 - head_3_loss: 2.4522e-04 - head_4_accuracy: 1.0000 - head_4_loss: 2.6849e-04 - head_5_accuracy: 1.0000 - head_5_loss: 3.2804e-04 - head_6_accuracy: 1.0000 - head_6_loss: 3.4035e-04 - head_7_accuracy: 1.0000 - head_7_loss: 3.7799e-04 - loss: 0.0022\n","Epoch 66: val_head_1_accuracy did not improve from 0.99948\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m12s\u001b[0m 37ms/step - head_1_accuracy: 1.0000 - head_1_loss: 2.7377e-04 - head_2_accuracy: 1.0000 - head_2_loss: 2.4513e-04 - head_3_accuracy: 1.0000 - head_3_loss: 2.2828e-04 - head_4_accuracy: 1.0000 - head_4_loss: 2.3024e-04 - head_5_accuracy: 1.0000 - head_5_loss: 2.6930e-04 - head_6_accuracy: 1.0000 - head_6_loss: 3.0593e-04 - head_7_accuracy: 1.0000 - head_7_loss: 3.1669e-04 - loss: 0.0019 - val_head_1_accuracy: 0.9994 - val_head_1_loss: 0.0020 - val_head_2_accuracy: 0.9994 - val_head_2_loss: 0.0019 - val_head_3_accuracy: 0.9994 - val_head_3_loss: 0.0019 - val_head_4_accuracy: 0.9995 - val_head_4_loss: 0.0020 - val_head_5_accuracy: 0.9995 - val_head_5_loss: 0.0020 - val_head_6_accuracy: 0.9995 - val_head_6_loss: 0.0019 - val_head_7_accuracy: 0.9995 - val_head_7_loss: 0.0019 - val_loss: 0.0137 - learning_rate: 0.0090\n","\n","Epoch 67: LearningRateScheduler setting learning rate to 0.00875.\n","Epoch 67/100\n","\u001b[1m315/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m━\u001b[0m \u001b[1m0s\u001b[0m 32ms/step - head_1_accuracy: 1.0000 - head_1_loss: 2.6780e-04 - head_2_accuracy: 1.0000 - head_2_loss: 2.3594e-04 - head_3_accuracy: 1.0000 - head_3_loss: 2.2385e-04 - head_4_accuracy: 1.0000 - head_4_loss: 2.7473e-04 - head_5_accuracy: 1.0000 - head_5_loss: 2.6693e-04 - head_6_accuracy: 1.0000 - head_6_loss: 3.2316e-04 - head_7_accuracy: 0.9999 - head_7_loss: 3.1304e-04 - loss: 0.0019\n","Epoch 67: val_head_1_accuracy did not improve from 0.99948\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m12s\u001b[0m 37ms/step - head_1_accuracy: 1.0000 - head_1_loss: 2.7823e-04 - head_2_accuracy: 1.0000 - head_2_loss: 2.5611e-04 - head_3_accuracy: 1.0000 - head_3_loss: 2.3859e-04 - head_4_accuracy: 1.0000 - head_4_loss: 2.8741e-04 - head_5_accuracy: 1.0000 - head_5_loss: 2.6945e-04 - head_6_accuracy: 1.0000 - head_6_loss: 3.0568e-04 - head_7_accuracy: 0.9999 - head_7_loss: 3.1490e-04 - loss: 0.0020 - val_head_1_accuracy: 0.9994 - val_head_1_loss: 0.0021 - val_head_2_accuracy: 0.9995 - val_head_2_loss: 0.0021 - val_head_3_accuracy: 0.9993 - val_head_3_loss: 0.0022 - val_head_4_accuracy: 0.9994 - val_head_4_loss: 0.0021 - val_head_5_accuracy: 0.9994 - val_head_5_loss: 0.0021 - val_head_6_accuracy: 0.9994 - val_head_6_loss: 0.0020 - val_head_7_accuracy: 0.9994 - val_head_7_loss: 0.0021 - val_loss: 0.0148 - learning_rate: 0.0088\n","\n","Epoch 68: LearningRateScheduler setting learning rate to 0.008545808992921232.\n","Epoch 68/100\n","\u001b[1m315/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m━\u001b[0m \u001b[1m0s\u001b[0m 32ms/step - head_1_accuracy: 1.0000 - head_1_loss: 2.5771e-04 - head_2_accuracy: 1.0000 - head_2_loss: 2.2816e-04 - head_3_accuracy: 1.0000 - head_3_loss: 2.5811e-04 - head_4_accuracy: 1.0000 - head_4_loss: 2.4922e-04 - head_5_accuracy: 1.0000 - head_5_loss: 2.8762e-04 - head_6_accuracy: 1.0000 - head_6_loss: 3.0563e-04 - head_7_accuracy: 1.0000 - head_7_loss: 2.7158e-04 - loss: 0.0019\n","Epoch 68: val_head_1_accuracy did not improve from 0.99948\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m12s\u001b[0m 37ms/step - head_1_accuracy: 1.0000 - head_1_loss: 2.5414e-04 - head_2_accuracy: 1.0000 - head_2_loss: 2.4251e-04 - head_3_accuracy: 1.0000 - head_3_loss: 2.3895e-04 - head_4_accuracy: 1.0000 - head_4_loss: 2.3266e-04 - head_5_accuracy: 1.0000 - head_5_loss: 2.4541e-04 - head_6_accuracy: 1.0000 - head_6_loss: 2.8198e-04 - head_7_accuracy: 1.0000 - head_7_loss: 2.6783e-04 - loss: 0.0018 - val_head_1_accuracy: 0.9994 - val_head_1_loss: 0.0020 - val_head_2_accuracy: 0.9995 - val_head_2_loss: 0.0020 - val_head_3_accuracy: 0.9994 - val_head_3_loss: 0.0020 - val_head_4_accuracy: 0.9995 - val_head_4_loss: 0.0020 - val_head_5_accuracy: 0.9995 - val_head_5_loss: 0.0021 - val_head_6_accuracy: 0.9994 - val_head_6_loss: 0.0019 - val_head_7_accuracy: 0.9994 - val_head_7_loss: 0.0020 - val_loss: 0.0141 - learning_rate: 0.0085\n","\n","Epoch 69: LearningRateScheduler setting learning rate to 0.008345599521004173.\n","Epoch 69/100\n","\u001b[1m315/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m━\u001b[0m \u001b[1m0s\u001b[0m 32ms/step - head_1_accuracy: 1.0000 - head_1_loss: 1.8460e-04 - head_2_accuracy: 1.0000 - head_2_loss: 1.5159e-04 - head_3_accuracy: 1.0000 - head_3_loss: 1.8056e-04 - head_4_accuracy: 1.0000 - head_4_loss: 1.8323e-04 - head_5_accuracy: 1.0000 - head_5_loss: 1.7181e-04 - head_6_accuracy: 1.0000 - head_6_loss: 1.8694e-04 - head_7_accuracy: 1.0000 - head_7_loss: 1.9497e-04 - loss: 0.0013\n","Epoch 69: val_head_1_accuracy did not improve from 0.99948\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m12s\u001b[0m 37ms/step - head_1_accuracy: 1.0000 - head_1_loss: 1.9895e-04 - head_2_accuracy: 1.0000 - head_2_loss: 1.6843e-04 - head_3_accuracy: 1.0000 - head_3_loss: 2.0832e-04 - head_4_accuracy: 1.0000 - head_4_loss: 2.0520e-04 - head_5_accuracy: 1.0000 - head_5_loss: 1.9859e-04 - head_6_accuracy: 1.0000 - head_6_loss: 2.1816e-04 - head_7_accuracy: 1.0000 - head_7_loss: 2.1273e-04 - loss: 0.0014 - val_head_1_accuracy: 0.9994 - val_head_1_loss: 0.0020 - val_head_2_accuracy: 0.9994 - val_head_2_loss: 0.0020 - val_head_3_accuracy: 0.9994 - val_head_3_loss: 0.0019 - val_head_4_accuracy: 0.9994 - val_head_4_loss: 0.0020 - val_head_5_accuracy: 0.9994 - val_head_5_loss: 0.0020 - val_head_6_accuracy: 0.9994 - val_head_6_loss: 0.0019 - val_head_7_accuracy: 0.9993 - val_head_7_loss: 0.0020 - val_loss: 0.0139 - learning_rate: 0.0083\n","\n","Epoch 70: LearningRateScheduler setting learning rate to 0.008149573178216015.\n","Epoch 70/100\n","\u001b[1m315/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m━\u001b[0m \u001b[1m0s\u001b[0m 32ms/step - head_1_accuracy: 1.0000 - head_1_loss: 2.0871e-04 - head_2_accuracy: 1.0000 - head_2_loss: 1.4127e-04 - head_3_accuracy: 1.0000 - head_3_loss: 1.5654e-04 - head_4_accuracy: 1.0000 - head_4_loss: 1.6583e-04 - head_5_accuracy: 1.0000 - head_5_loss: 1.8338e-04 - head_6_accuracy: 1.0000 - head_6_loss: 2.3227e-04 - head_7_accuracy: 1.0000 - head_7_loss: 1.8102e-04 - loss: 0.0013\n","Epoch 70: val_head_1_accuracy did not improve from 0.99948\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m12s\u001b[0m 37ms/step - head_1_accuracy: 1.0000 - head_1_loss: 2.3465e-04 - head_2_accuracy: 1.0000 - head_2_loss: 1.5688e-04 - head_3_accuracy: 1.0000 - head_3_loss: 1.9034e-04 - head_4_accuracy: 1.0000 - head_4_loss: 1.8820e-04 - head_5_accuracy: 1.0000 - head_5_loss: 2.2672e-04 - head_6_accuracy: 1.0000 - head_6_loss: 3.0818e-04 - head_7_accuracy: 1.0000 - head_7_loss: 2.2390e-04 - loss: 0.0015 - val_head_1_accuracy: 0.9994 - val_head_1_loss: 0.0020 - val_head_2_accuracy: 0.9994 - val_head_2_loss: 0.0020 - val_head_3_accuracy: 0.9994 - val_head_3_loss: 0.0019 - val_head_4_accuracy: 0.9995 - val_head_4_loss: 0.0020 - val_head_5_accuracy: 0.9995 - val_head_5_loss: 0.0020 - val_head_6_accuracy: 0.9994 - val_head_6_loss: 0.0019 - val_head_7_accuracy: 0.9995 - val_head_7_loss: 0.0019 - val_loss: 0.0137 - learning_rate: 0.0081\n","\n","Epoch 71: LearningRateScheduler setting learning rate to 0.0079579273464675.\n","Epoch 71/100\n","\u001b[1m315/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m━\u001b[0m \u001b[1m0s\u001b[0m 32ms/step - head_1_accuracy: 1.0000 - head_1_loss: 2.1233e-04 - head_2_accuracy: 1.0000 - head_2_loss: 1.8812e-04 - head_3_accuracy: 1.0000 - head_3_loss: 1.9273e-04 - head_4_accuracy: 1.0000 - head_4_loss: 2.1525e-04 - head_5_accuracy: 1.0000 - head_5_loss: 2.1391e-04 - head_6_accuracy: 1.0000 - head_6_loss: 3.0658e-04 - head_7_accuracy: 1.0000 - head_7_loss: 2.6141e-04 - loss: 0.0016\n","Epoch 71: val_head_1_accuracy did not improve from 0.99948\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m12s\u001b[0m 37ms/step - head_1_accuracy: 1.0000 - head_1_loss: 2.0693e-04 - head_2_accuracy: 1.0000 - head_2_loss: 1.8670e-04 - head_3_accuracy: 1.0000 - head_3_loss: 1.9721e-04 - head_4_accuracy: 1.0000 - head_4_loss: 2.0926e-04 - head_5_accuracy: 1.0000 - head_5_loss: 2.1380e-04 - head_6_accuracy: 1.0000 - head_6_loss: 2.4651e-04 - head_7_accuracy: 1.0000 - head_7_loss: 2.3556e-04 - loss: 0.0015 - val_head_1_accuracy: 0.9993 - val_head_1_loss: 0.0019 - val_head_2_accuracy: 0.9994 - val_head_2_loss: 0.0019 - val_head_3_accuracy: 0.9993 - val_head_3_loss: 0.0019 - val_head_4_accuracy: 0.9994 - val_head_4_loss: 0.0019 - val_head_5_accuracy: 0.9994 - val_head_5_loss: 0.0020 - val_head_6_accuracy: 0.9994 - val_head_6_loss: 0.0018 - val_head_7_accuracy: 0.9994 - val_head_7_loss: 0.0019 - val_loss: 0.0135 - learning_rate: 0.0080\n","\n","Epoch 72: LearningRateScheduler setting learning rate to 0.007770854996866082.\n","Epoch 72/100\n","\u001b[1m315/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m━\u001b[0m \u001b[1m0s\u001b[0m 32ms/step - head_1_accuracy: 1.0000 - head_1_loss: 1.3555e-04 - head_2_accuracy: 1.0000 - head_2_loss: 1.0693e-04 - head_3_accuracy: 1.0000 - head_3_loss: 1.1384e-04 - head_4_accuracy: 1.0000 - head_4_loss: 1.2179e-04 - head_5_accuracy: 1.0000 - head_5_loss: 1.1953e-04 - head_6_accuracy: 1.0000 - head_6_loss: 1.4762e-04 - head_7_accuracy: 1.0000 - head_7_loss: 1.5571e-04 - loss: 9.0098e-04\n","Epoch 72: val_head_1_accuracy did not improve from 0.99948\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m12s\u001b[0m 37ms/step - head_1_accuracy: 1.0000 - head_1_loss: 1.3756e-04 - head_2_accuracy: 1.0000 - head_2_loss: 1.0888e-04 - head_3_accuracy: 1.0000 - head_3_loss: 1.1698e-04 - head_4_accuracy: 1.0000 - head_4_loss: 1.3148e-04 - head_5_accuracy: 1.0000 - head_5_loss: 1.2834e-04 - head_6_accuracy: 1.0000 - head_6_loss: 1.4696e-04 - head_7_accuracy: 1.0000 - head_7_loss: 1.5221e-04 - loss: 9.2291e-04 - val_head_1_accuracy: 0.9995 - val_head_1_loss: 0.0018 - val_head_2_accuracy: 0.9995 - val_head_2_loss: 0.0017 - val_head_3_accuracy: 0.9994 - val_head_3_loss: 0.0017 - val_head_4_accuracy: 0.9995 - val_head_4_loss: 0.0018 - val_head_5_accuracy: 0.9995 - val_head_5_loss: 0.0018 - val_head_6_accuracy: 0.9995 - val_head_6_loss: 0.0017 - val_head_7_accuracy: 0.9995 - val_head_7_loss: 0.0018 - val_loss: 0.0123 - learning_rate: 0.0078\n","\n","Epoch 73: LearningRateScheduler setting learning rate to 0.007588544495410363.\n","Epoch 73/100\n","\u001b[1m315/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m━\u001b[0m \u001b[1m0s\u001b[0m 32ms/step - head_1_accuracy: 1.0000 - head_1_loss: 1.5735e-04 - head_2_accuracy: 1.0000 - head_2_loss: 1.4581e-04 - head_3_accuracy: 1.0000 - head_3_loss: 1.2855e-04 - head_4_accuracy: 1.0000 - head_4_loss: 1.4295e-04 - head_5_accuracy: 1.0000 - head_5_loss: 1.3410e-04 - head_6_accuracy: 1.0000 - head_6_loss: 1.6943e-04 - head_7_accuracy: 1.0000 - head_7_loss: 1.4995e-04 - loss: 0.0010\n","Epoch 73: val_head_1_accuracy did not improve from 0.99948\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m12s\u001b[0m 37ms/step - head_1_accuracy: 1.0000 - head_1_loss: 2.5131e-04 - head_2_accuracy: 1.0000 - head_2_loss: 3.2342e-04 - head_3_accuracy: 1.0000 - head_3_loss: 2.4736e-04 - head_4_accuracy: 1.0000 - head_4_loss: 2.6295e-04 - head_5_accuracy: 1.0000 - head_5_loss: 2.2432e-04 - head_6_accuracy: 0.9999 - head_6_loss: 2.5942e-04 - head_7_accuracy: 1.0000 - head_7_loss: 2.7312e-04 - loss: 0.0018 - val_head_1_accuracy: 0.9993 - val_head_1_loss: 0.0022 - val_head_2_accuracy: 0.9992 - val_head_2_loss: 0.0022 - val_head_3_accuracy: 0.9993 - val_head_3_loss: 0.0021 - val_head_4_accuracy: 0.9993 - val_head_4_loss: 0.0022 - val_head_5_accuracy: 0.9993 - val_head_5_loss: 0.0022 - val_head_6_accuracy: 0.9993 - val_head_6_loss: 0.0020 - val_head_7_accuracy: 0.9993 - val_head_7_loss: 0.0021 - val_loss: 0.0152 - learning_rate: 0.0076\n","\n","Epoch 74: LearningRateScheduler setting learning rate to 0.007411179413321509.\n","Epoch 74/100\n","\u001b[1m315/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m━\u001b[0m \u001b[1m0s\u001b[0m 32ms/step - head_1_accuracy: 1.0000 - head_1_loss: 2.1151e-04 - head_2_accuracy: 1.0000 - head_2_loss: 1.6891e-04 - head_3_accuracy: 1.0000 - head_3_loss: 2.0305e-04 - head_4_accuracy: 1.0000 - head_4_loss: 1.9878e-04 - head_5_accuracy: 1.0000 - head_5_loss: 2.2695e-04 - head_6_accuracy: 1.0000 - head_6_loss: 2.1860e-04 - head_7_accuracy: 0.9999 - head_7_loss: 3.2185e-04 - loss: 0.0015\n","Epoch 74: val_head_1_accuracy did not improve from 0.99948\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m12s\u001b[0m 37ms/step - head_1_accuracy: 1.0000 - head_1_loss: 2.7060e-04 - head_2_accuracy: 1.0000 - head_2_loss: 2.7172e-04 - head_3_accuracy: 0.9999 - head_3_loss: 2.9285e-04 - head_4_accuracy: 1.0000 - head_4_loss: 2.6277e-04 - head_5_accuracy: 1.0000 - head_5_loss: 2.8730e-04 - head_6_accuracy: 1.0000 - head_6_loss: 2.7960e-04 - head_7_accuracy: 0.9999 - head_7_loss: 3.5720e-04 - loss: 0.0020 - val_head_1_accuracy: 0.9993 - val_head_1_loss: 0.0024 - val_head_2_accuracy: 0.9993 - val_head_2_loss: 0.0023 - val_head_3_accuracy: 0.9994 - val_head_3_loss: 0.0022 - val_head_4_accuracy: 0.9994 - val_head_4_loss: 0.0023 - val_head_5_accuracy: 0.9995 - val_head_5_loss: 0.0023 - val_head_6_accuracy: 0.9995 - val_head_6_loss: 0.0023 - val_head_7_accuracy: 0.9994 - val_head_7_loss: 0.0022 - val_loss: 0.0162 - learning_rate: 0.0074\n","\n","Epoch 75: LearningRateScheduler setting learning rate to 0.0072389383422025895.\n","Epoch 75/100\n","\u001b[1m315/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m━\u001b[0m \u001b[1m0s\u001b[0m 32ms/step - head_1_accuracy: 1.0000 - head_1_loss: 1.6013e-04 - head_2_accuracy: 1.0000 - head_2_loss: 1.2520e-04 - head_3_accuracy: 1.0000 - head_3_loss: 1.2237e-04 - head_4_accuracy: 1.0000 - head_4_loss: 1.4198e-04 - head_5_accuracy: 1.0000 - head_5_loss: 1.6981e-04 - head_6_accuracy: 1.0000 - head_6_loss: 1.9646e-04 - head_7_accuracy: 1.0000 - head_7_loss: 1.8827e-04 - loss: 0.0011\n","Epoch 75: val_head_1_accuracy did not improve from 0.99948\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m12s\u001b[0m 37ms/step - head_1_accuracy: 1.0000 - head_1_loss: 1.6288e-04 - head_2_accuracy: 1.0000 - head_2_loss: 1.4196e-04 - head_3_accuracy: 1.0000 - head_3_loss: 1.3011e-04 - head_4_accuracy: 1.0000 - head_4_loss: 1.5674e-04 - head_5_accuracy: 1.0000 - head_5_loss: 2.1673e-04 - head_6_accuracy: 1.0000 - head_6_loss: 1.8239e-04 - head_7_accuracy: 1.0000 - head_7_loss: 1.9507e-04 - loss: 0.0012 - val_head_1_accuracy: 0.9994 - val_head_1_loss: 0.0024 - val_head_2_accuracy: 0.9994 - val_head_2_loss: 0.0024 - val_head_3_accuracy: 0.9994 - val_head_3_loss: 0.0023 - val_head_4_accuracy: 0.9995 - val_head_4_loss: 0.0023 - val_head_5_accuracy: 0.9995 - val_head_5_loss: 0.0024 - val_head_6_accuracy: 0.9994 - val_head_6_loss: 0.0023 - val_head_7_accuracy: 0.9995 - val_head_7_loss: 0.0023 - val_loss: 0.0166 - learning_rate: 0.0072\n","\n","Epoch 76: LearningRateScheduler setting learning rate to 0.007071994714211974.\n","Epoch 76/100\n","\u001b[1m315/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m━\u001b[0m \u001b[1m0s\u001b[0m 32ms/step - head_1_accuracy: 1.0000 - head_1_loss: 2.2909e-04 - head_2_accuracy: 0.9999 - head_2_loss: 2.6283e-04 - head_3_accuracy: 0.9999 - head_3_loss: 1.7862e-04 - head_4_accuracy: 0.9999 - head_4_loss: 2.0850e-04 - head_5_accuracy: 1.0000 - head_5_loss: 1.7418e-04 - head_6_accuracy: 1.0000 - head_6_loss: 2.2844e-04 - head_7_accuracy: 1.0000 - head_7_loss: 1.8869e-04 - loss: 0.0015\n","Epoch 76: val_head_1_accuracy did not improve from 0.99948\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m12s\u001b[0m 37ms/step - head_1_accuracy: 1.0000 - head_1_loss: 2.1121e-04 - head_2_accuracy: 1.0000 - head_2_loss: 2.0543e-04 - head_3_accuracy: 1.0000 - head_3_loss: 1.8033e-04 - head_4_accuracy: 1.0000 - head_4_loss: 1.9427e-04 - head_5_accuracy: 1.0000 - head_5_loss: 1.7922e-04 - head_6_accuracy: 1.0000 - head_6_loss: 2.8429e-04 - head_7_accuracy: 1.0000 - head_7_loss: 2.1462e-04 - loss: 0.0015 - val_head_1_accuracy: 0.9990 - val_head_1_loss: 0.0025 - val_head_2_accuracy: 0.9992 - val_head_2_loss: 0.0024 - val_head_3_accuracy: 0.9993 - val_head_3_loss: 0.0023 - val_head_4_accuracy: 0.9993 - val_head_4_loss: 0.0024 - val_head_5_accuracy: 0.9993 - val_head_5_loss: 0.0024 - val_head_6_accuracy: 0.9993 - val_head_6_loss: 0.0022 - val_head_7_accuracy: 0.9993 - val_head_7_loss: 0.0023 - val_loss: 0.0167 - learning_rate: 0.0071\n","\n","Epoch 77: LearningRateScheduler setting learning rate to 0.00691051662743184.\n","Epoch 77/100\n","\u001b[1m315/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m━\u001b[0m \u001b[1m0s\u001b[0m 32ms/step - head_1_accuracy: 1.0000 - head_1_loss: 1.5101e-04 - head_2_accuracy: 1.0000 - head_2_loss: 1.2001e-04 - head_3_accuracy: 1.0000 - head_3_loss: 1.2925e-04 - head_4_accuracy: 1.0000 - head_4_loss: 1.4483e-04 - head_5_accuracy: 1.0000 - head_5_loss: 1.5478e-04 - head_6_accuracy: 1.0000 - head_6_loss: 1.8348e-04 - head_7_accuracy: 1.0000 - head_7_loss: 1.6283e-04 - loss: 0.0010\n","Epoch 77: val_head_1_accuracy did not improve from 0.99948\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m12s\u001b[0m 36ms/step - head_1_accuracy: 1.0000 - head_1_loss: 1.5644e-04 - head_2_accuracy: 1.0000 - head_2_loss: 1.2775e-04 - head_3_accuracy: 1.0000 - head_3_loss: 1.3533e-04 - head_4_accuracy: 1.0000 - head_4_loss: 1.5200e-04 - head_5_accuracy: 1.0000 - head_5_loss: 1.5640e-04 - head_6_accuracy: 1.0000 - head_6_loss: 1.7696e-04 - head_7_accuracy: 1.0000 - head_7_loss: 1.6198e-04 - loss: 0.0011 - val_head_1_accuracy: 0.9991 - val_head_1_loss: 0.0027 - val_head_2_accuracy: 0.9991 - val_head_2_loss: 0.0026 - val_head_3_accuracy: 0.9992 - val_head_3_loss: 0.0025 - val_head_4_accuracy: 0.9992 - val_head_4_loss: 0.0025 - val_head_5_accuracy: 0.9992 - val_head_5_loss: 0.0026 - val_head_6_accuracy: 0.9993 - val_head_6_loss: 0.0024 - val_head_7_accuracy: 0.9993 - val_head_7_loss: 0.0025 - val_loss: 0.0180 - learning_rate: 0.0069\n","\n","Epoch 78: LearningRateScheduler setting learning rate to 0.006754666676607666.\n","Epoch 78/100\n","\u001b[1m315/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m━\u001b[0m \u001b[1m0s\u001b[0m 32ms/step - head_1_accuracy: 1.0000 - head_1_loss: 1.8950e-04 - head_2_accuracy: 1.0000 - head_2_loss: 1.2964e-04 - head_3_accuracy: 1.0000 - head_3_loss: 1.4896e-04 - head_4_accuracy: 1.0000 - head_4_loss: 1.3591e-04 - head_5_accuracy: 1.0000 - head_5_loss: 1.3524e-04 - head_6_accuracy: 1.0000 - head_6_loss: 1.5743e-04 - head_7_accuracy: 1.0000 - head_7_loss: 1.3718e-04 - loss: 0.0010\n","Epoch 78: val_head_1_accuracy did not improve from 0.99948\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m12s\u001b[0m 36ms/step - head_1_accuracy: 1.0000 - head_1_loss: 2.1105e-04 - head_2_accuracy: 1.0000 - head_2_loss: 1.3416e-04 - head_3_accuracy: 1.0000 - head_3_loss: 1.4128e-04 - head_4_accuracy: 1.0000 - head_4_loss: 1.3336e-04 - head_5_accuracy: 1.0000 - head_5_loss: 1.3377e-04 - head_6_accuracy: 1.0000 - head_6_loss: 1.5134e-04 - head_7_accuracy: 1.0000 - head_7_loss: 1.4314e-04 - loss: 0.0010 - val_head_1_accuracy: 0.9993 - val_head_1_loss: 0.0023 - val_head_2_accuracy: 0.9993 - val_head_2_loss: 0.0022 - val_head_3_accuracy: 0.9993 - val_head_3_loss: 0.0021 - val_head_4_accuracy: 0.9993 - val_head_4_loss: 0.0022 - val_head_5_accuracy: 0.9993 - val_head_5_loss: 0.0022 - val_head_6_accuracy: 0.9994 - val_head_6_loss: 0.0021 - val_head_7_accuracy: 0.9993 - val_head_7_loss: 0.0021 - val_loss: 0.0153 - learning_rate: 0.0068\n","\n","Epoch 79: LearningRateScheduler setting learning rate to 0.006604601789429095.\n","Epoch 79/100\n","\u001b[1m315/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m━\u001b[0m \u001b[1m0s\u001b[0m 32ms/step - head_1_accuracy: 1.0000 - head_1_loss: 1.3258e-04 - head_2_accuracy: 1.0000 - head_2_loss: 1.0194e-04 - head_3_accuracy: 1.0000 - head_3_loss: 1.1808e-04 - head_4_accuracy: 1.0000 - head_4_loss: 1.4042e-04 - head_5_accuracy: 1.0000 - head_5_loss: 1.3868e-04 - head_6_accuracy: 1.0000 - head_6_loss: 1.6487e-04 - head_7_accuracy: 1.0000 - head_7_loss: 1.5706e-04 - loss: 9.5363e-04\n","Epoch 79: val_head_1_accuracy did not improve from 0.99948\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m12s\u001b[0m 36ms/step - head_1_accuracy: 1.0000 - head_1_loss: 1.3756e-04 - head_2_accuracy: 1.0000 - head_2_loss: 1.1326e-04 - head_3_accuracy: 1.0000 - head_3_loss: 1.3468e-04 - head_4_accuracy: 1.0000 - head_4_loss: 1.3619e-04 - head_5_accuracy: 1.0000 - head_5_loss: 1.2620e-04 - head_6_accuracy: 1.0000 - head_6_loss: 1.4065e-04 - head_7_accuracy: 1.0000 - head_7_loss: 1.5475e-04 - loss: 9.4385e-04 - val_head_1_accuracy: 0.9993 - val_head_1_loss: 0.0021 - val_head_2_accuracy: 0.9993 - val_head_2_loss: 0.0020 - val_head_3_accuracy: 0.9993 - val_head_3_loss: 0.0020 - val_head_4_accuracy: 0.9993 - val_head_4_loss: 0.0020 - val_head_5_accuracy: 0.9993 - val_head_5_loss: 0.0021 - val_head_6_accuracy: 0.9994 - val_head_6_loss: 0.0019 - val_head_7_accuracy: 0.9993 - val_head_7_loss: 0.0020 - val_loss: 0.0141 - learning_rate: 0.0066\n","\n","Epoch 80: LearningRateScheduler setting learning rate to 0.0064604730685170605.\n","Epoch 80/100\n","\u001b[1m315/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m━\u001b[0m \u001b[1m0s\u001b[0m 32ms/step - head_1_accuracy: 1.0000 - head_1_loss: 1.3859e-04 - head_2_accuracy: 1.0000 - head_2_loss: 1.5965e-04 - head_3_accuracy: 0.9999 - head_3_loss: 2.0816e-04 - head_4_accuracy: 1.0000 - head_4_loss: 1.5321e-04 - head_5_accuracy: 0.9999 - head_5_loss: 1.8043e-04 - head_6_accuracy: 1.0000 - head_6_loss: 1.6466e-04 - head_7_accuracy: 1.0000 - head_7_loss: 1.6482e-04 - loss: 0.0012\n","Epoch 80: val_head_1_accuracy did not improve from 0.99948\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m12s\u001b[0m 36ms/step - head_1_accuracy: 1.0000 - head_1_loss: 1.5400e-04 - head_2_accuracy: 1.0000 - head_2_loss: 2.7123e-04 - head_3_accuracy: 1.0000 - head_3_loss: 1.6813e-04 - head_4_accuracy: 1.0000 - head_4_loss: 2.3111e-04 - head_5_accuracy: 1.0000 - head_5_loss: 1.4043e-04 - head_6_accuracy: 1.0000 - head_6_loss: 2.2036e-04 - head_7_accuracy: 1.0000 - head_7_loss: 1.9211e-04 - loss: 0.0014 - val_head_1_accuracy: 0.9993 - val_head_1_loss: 0.0020 - val_head_2_accuracy: 0.9995 - val_head_2_loss: 0.0019 - val_head_3_accuracy: 0.9994 - val_head_3_loss: 0.0019 - val_head_4_accuracy: 0.9993 - val_head_4_loss: 0.0019 - val_head_5_accuracy: 0.9995 - val_head_5_loss: 0.0020 - val_head_6_accuracy: 0.9994 - val_head_6_loss: 0.0017 - val_head_7_accuracy: 0.9994 - val_head_7_loss: 0.0019 - val_loss: 0.0133 - learning_rate: 0.0065\n","\n","Epoch 81: LearningRateScheduler setting learning rate to 0.006322425639276255.\n","Epoch 81/100\n","\u001b[1m315/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m━\u001b[0m \u001b[1m0s\u001b[0m 32ms/step - head_1_accuracy: 1.0000 - head_1_loss: 1.5102e-04 - head_2_accuracy: 1.0000 - head_2_loss: 1.1322e-04 - head_3_accuracy: 1.0000 - head_3_loss: 1.1211e-04 - head_4_accuracy: 1.0000 - head_4_loss: 1.3199e-04 - head_5_accuracy: 1.0000 - head_5_loss: 1.4926e-04 - head_6_accuracy: 1.0000 - head_6_loss: 1.4948e-04 - head_7_accuracy: 1.0000 - head_7_loss: 1.6377e-04 - loss: 9.7084e-04\n","Epoch 81: val_head_1_accuracy did not improve from 0.99948\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m12s\u001b[0m 37ms/step - head_1_accuracy: 1.0000 - head_1_loss: 1.4733e-04 - head_2_accuracy: 1.0000 - head_2_loss: 1.1936e-04 - head_3_accuracy: 1.0000 - head_3_loss: 1.1244e-04 - head_4_accuracy: 1.0000 - head_4_loss: 1.4525e-04 - head_5_accuracy: 1.0000 - head_5_loss: 1.3504e-04 - head_6_accuracy: 1.0000 - head_6_loss: 1.3851e-04 - head_7_accuracy: 1.0000 - head_7_loss: 1.5254e-04 - loss: 9.5098e-04 - val_head_1_accuracy: 0.9993 - val_head_1_loss: 0.0019 - val_head_2_accuracy: 0.9995 - val_head_2_loss: 0.0018 - val_head_3_accuracy: 0.9994 - val_head_3_loss: 0.0019 - val_head_4_accuracy: 0.9994 - val_head_4_loss: 0.0018 - val_head_5_accuracy: 0.9995 - val_head_5_loss: 0.0019 - val_head_6_accuracy: 0.9994 - val_head_6_loss: 0.0017 - val_head_7_accuracy: 0.9994 - val_head_7_loss: 0.0018 - val_loss: 0.0129 - learning_rate: 0.0063\n","\n","Epoch 82: LearningRateScheduler setting learning rate to 0.006190598503766142.\n","Epoch 82/100\n","\u001b[1m315/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m━\u001b[0m \u001b[1m0s\u001b[0m 32ms/step - head_1_accuracy: 1.0000 - head_1_loss: 1.6356e-04 - head_2_accuracy: 1.0000 - head_2_loss: 1.1812e-04 - head_3_accuracy: 1.0000 - head_3_loss: 1.5406e-04 - head_4_accuracy: 1.0000 - head_4_loss: 1.5468e-04 - head_5_accuracy: 1.0000 - head_5_loss: 1.7861e-04 - head_6_accuracy: 1.0000 - head_6_loss: 1.8237e-04 - head_7_accuracy: 1.0000 - head_7_loss: 1.6831e-04 - loss: 0.0011\n","Epoch 82: val_head_1_accuracy did not improve from 0.99948\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m12s\u001b[0m 37ms/step - head_1_accuracy: 1.0000 - head_1_loss: 1.3928e-04 - head_2_accuracy: 1.0000 - head_2_loss: 9.8888e-05 - head_3_accuracy: 1.0000 - head_3_loss: 1.2503e-04 - head_4_accuracy: 1.0000 - head_4_loss: 1.2983e-04 - head_5_accuracy: 1.0000 - head_5_loss: 1.4645e-04 - head_6_accuracy: 1.0000 - head_6_loss: 1.6107e-04 - head_7_accuracy: 1.0000 - head_7_loss: 1.5090e-04 - loss: 9.5197e-04 - val_head_1_accuracy: 0.9994 - val_head_1_loss: 0.0019 - val_head_2_accuracy: 0.9996 - val_head_2_loss: 0.0018 - val_head_3_accuracy: 0.9994 - val_head_3_loss: 0.0018 - val_head_4_accuracy: 0.9994 - val_head_4_loss: 0.0018 - val_head_5_accuracy: 0.9995 - val_head_5_loss: 0.0019 - val_head_6_accuracy: 0.9994 - val_head_6_loss: 0.0017 - val_head_7_accuracy: 0.9994 - val_head_7_loss: 0.0018 - val_loss: 0.0126 - learning_rate: 0.0062\n","\n","Epoch 83: LearningRateScheduler setting learning rate to 0.006065124400737674.\n","Epoch 83/100\n","\u001b[1m315/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m━\u001b[0m \u001b[1m0s\u001b[0m 32ms/step - head_1_accuracy: 1.0000 - head_1_loss: 1.2334e-04 - head_2_accuracy: 1.0000 - head_2_loss: 9.8930e-05 - head_3_accuracy: 1.0000 - head_3_loss: 9.9890e-05 - head_4_accuracy: 1.0000 - head_4_loss: 9.8968e-05 - head_5_accuracy: 1.0000 - head_5_loss: 1.1167e-04 - head_6_accuracy: 1.0000 - head_6_loss: 1.3764e-04 - head_7_accuracy: 1.0000 - head_7_loss: 1.2334e-04 - loss: 7.9377e-04\n","Epoch 83: val_head_1_accuracy did not improve from 0.99948\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m12s\u001b[0m 37ms/step - head_1_accuracy: 1.0000 - head_1_loss: 1.3762e-04 - head_2_accuracy: 1.0000 - head_2_loss: 1.2896e-04 - head_3_accuracy: 1.0000 - head_3_loss: 1.3535e-04 - head_4_accuracy: 1.0000 - head_4_loss: 1.1381e-04 - head_5_accuracy: 1.0000 - head_5_loss: 1.3638e-04 - head_6_accuracy: 1.0000 - head_6_loss: 1.3677e-04 - head_7_accuracy: 1.0000 - head_7_loss: 1.5802e-04 - loss: 9.4739e-04 - val_head_1_accuracy: 0.9993 - val_head_1_loss: 0.0019 - val_head_2_accuracy: 0.9993 - val_head_2_loss: 0.0018 - val_head_3_accuracy: 0.9994 - val_head_3_loss: 0.0018 - val_head_4_accuracy: 0.9994 - val_head_4_loss: 0.0018 - val_head_5_accuracy: 0.9995 - val_head_5_loss: 0.0019 - val_head_6_accuracy: 0.9993 - val_head_6_loss: 0.0018 - val_head_7_accuracy: 0.9995 - val_head_7_loss: 0.0018 - val_loss: 0.0129 - learning_rate: 0.0061\n","\n","Epoch 84: LearningRateScheduler setting learning rate to 0.005946129671976613.\n","Epoch 84/100\n","\u001b[1m315/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m━\u001b[0m \u001b[1m0s\u001b[0m 32ms/step - head_1_accuracy: 1.0000 - head_1_loss: 1.6904e-04 - head_2_accuracy: 1.0000 - head_2_loss: 1.2326e-04 - head_3_accuracy: 1.0000 - head_3_loss: 1.5038e-04 - head_4_accuracy: 1.0000 - head_4_loss: 1.5395e-04 - head_5_accuracy: 1.0000 - head_5_loss: 1.4492e-04 - head_6_accuracy: 1.0000 - head_6_loss: 1.5690e-04 - head_7_accuracy: 0.9999 - head_7_loss: 2.1762e-04 - loss: 0.0011\n","Epoch 84: val_head_1_accuracy did not improve from 0.99948\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m12s\u001b[0m 37ms/step - head_1_accuracy: 1.0000 - head_1_loss: 1.6391e-04 - head_2_accuracy: 1.0000 - head_2_loss: 1.5923e-04 - head_3_accuracy: 1.0000 - head_3_loss: 1.7454e-04 - head_4_accuracy: 1.0000 - head_4_loss: 1.5216e-04 - head_5_accuracy: 1.0000 - head_5_loss: 1.5769e-04 - head_6_accuracy: 1.0000 - head_6_loss: 1.8282e-04 - head_7_accuracy: 1.0000 - head_7_loss: 1.9752e-04 - loss: 0.0012 - val_head_1_accuracy: 0.9994 - val_head_1_loss: 0.0021 - val_head_2_accuracy: 0.9995 - val_head_2_loss: 0.0019 - val_head_3_accuracy: 0.9994 - val_head_3_loss: 0.0020 - val_head_4_accuracy: 0.9995 - val_head_4_loss: 0.0020 - val_head_5_accuracy: 0.9996 - val_head_5_loss: 0.0021 - val_head_6_accuracy: 0.9994 - val_head_6_loss: 0.0019 - val_head_7_accuracy: 0.9995 - val_head_7_loss: 0.0020 - val_loss: 0.0139 - learning_rate: 0.0059\n","\n","Epoch 85: LearningRateScheduler setting learning rate to 0.0058337341350880745.\n","Epoch 85/100\n","\u001b[1m315/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m━\u001b[0m \u001b[1m0s\u001b[0m 32ms/step - head_1_accuracy: 1.0000 - head_1_loss: 1.1555e-04 - head_2_accuracy: 1.0000 - head_2_loss: 9.2609e-05 - head_3_accuracy: 1.0000 - head_3_loss: 1.3042e-04 - head_4_accuracy: 1.0000 - head_4_loss: 1.1790e-04 - head_5_accuracy: 1.0000 - head_5_loss: 1.3039e-04 - head_6_accuracy: 1.0000 - head_6_loss: 1.1348e-04 - head_7_accuracy: 1.0000 - head_7_loss: 1.2247e-04 - loss: 8.2281e-04\n","Epoch 85: val_head_1_accuracy did not improve from 0.99948\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m12s\u001b[0m 37ms/step - head_1_accuracy: 1.0000 - head_1_loss: 1.0017e-04 - head_2_accuracy: 1.0000 - head_2_loss: 8.1517e-05 - head_3_accuracy: 1.0000 - head_3_loss: 1.1396e-04 - head_4_accuracy: 1.0000 - head_4_loss: 9.9347e-05 - head_5_accuracy: 1.0000 - head_5_loss: 1.1557e-04 - head_6_accuracy: 1.0000 - head_6_loss: 1.0428e-04 - head_7_accuracy: 1.0000 - head_7_loss: 1.1205e-04 - loss: 7.2732e-04 - val_head_1_accuracy: 0.9993 - val_head_1_loss: 0.0019 - val_head_2_accuracy: 0.9994 - val_head_2_loss: 0.0019 - val_head_3_accuracy: 0.9995 - val_head_3_loss: 0.0019 - val_head_4_accuracy: 0.9993 - val_head_4_loss: 0.0019 - val_head_5_accuracy: 0.9995 - val_head_5_loss: 0.0020 - val_head_6_accuracy: 0.9993 - val_head_6_loss: 0.0018 - val_head_7_accuracy: 0.9994 - val_head_7_loss: 0.0019 - val_loss: 0.0133 - learning_rate: 0.0058\n","\n","Epoch 86: LearningRateScheduler setting learning rate to 0.005728050962850342.\n","Epoch 86/100\n","\u001b[1m315/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m━\u001b[0m \u001b[1m0s\u001b[0m 32ms/step - head_1_accuracy: 1.0000 - head_1_loss: 1.1313e-04 - head_2_accuracy: 1.0000 - head_2_loss: 7.8902e-05 - head_3_accuracy: 1.0000 - head_3_loss: 9.4111e-05 - head_4_accuracy: 1.0000 - head_4_loss: 1.0804e-04 - head_5_accuracy: 1.0000 - head_5_loss: 2.0179e-04 - head_6_accuracy: 1.0000 - head_6_loss: 1.0072e-04 - head_7_accuracy: 1.0000 - head_7_loss: 1.2331e-04 - loss: 8.2000e-04\n","Epoch 86: val_head_1_accuracy did not improve from 0.99948\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m12s\u001b[0m 37ms/step - head_1_accuracy: 1.0000 - head_1_loss: 1.1192e-04 - head_2_accuracy: 1.0000 - head_2_loss: 8.3819e-05 - head_3_accuracy: 1.0000 - head_3_loss: 1.0411e-04 - head_4_accuracy: 1.0000 - head_4_loss: 1.1957e-04 - head_5_accuracy: 1.0000 - head_5_loss: 2.0691e-04 - head_6_accuracy: 1.0000 - head_6_loss: 9.9841e-05 - head_7_accuracy: 1.0000 - head_7_loss: 1.2951e-04 - loss: 8.5586e-04 - val_head_1_accuracy: 0.9994 - val_head_1_loss: 0.0019 - val_head_2_accuracy: 0.9995 - val_head_2_loss: 0.0018 - val_head_3_accuracy: 0.9995 - val_head_3_loss: 0.0018 - val_head_4_accuracy: 0.9994 - val_head_4_loss: 0.0019 - val_head_5_accuracy: 0.9995 - val_head_5_loss: 0.0020 - val_head_6_accuracy: 0.9993 - val_head_6_loss: 0.0017 - val_head_7_accuracy: 0.9994 - val_head_7_loss: 0.0019 - val_loss: 0.0131 - learning_rate: 0.0057\n","\n","Epoch 87: LearningRateScheduler setting learning rate to 0.005629186569259479.\n","Epoch 87/100\n","\u001b[1m315/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m━\u001b[0m \u001b[1m0s\u001b[0m 32ms/step - head_1_accuracy: 1.0000 - head_1_loss: 9.8922e-05 - head_2_accuracy: 1.0000 - head_2_loss: 9.0739e-05 - head_3_accuracy: 1.0000 - head_3_loss: 2.3955e-04 - head_4_accuracy: 1.0000 - head_4_loss: 1.0430e-04 - head_5_accuracy: 1.0000 - head_5_loss: 1.1784e-04 - head_6_accuracy: 1.0000 - head_6_loss: 1.2582e-04 - head_7_accuracy: 1.0000 - head_7_loss: 1.5512e-04 - loss: 9.3229e-04\n","Epoch 87: val_head_1_accuracy did not improve from 0.99948\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m12s\u001b[0m 37ms/step - head_1_accuracy: 1.0000 - head_1_loss: 1.0370e-04 - head_2_accuracy: 1.0000 - head_2_loss: 1.0552e-04 - head_3_accuracy: 1.0000 - head_3_loss: 1.8586e-04 - head_4_accuracy: 1.0000 - head_4_loss: 1.0552e-04 - head_5_accuracy: 1.0000 - head_5_loss: 1.1346e-04 - head_6_accuracy: 1.0000 - head_6_loss: 1.3624e-04 - head_7_accuracy: 1.0000 - head_7_loss: 1.5104e-04 - loss: 9.0186e-04 - val_head_1_accuracy: 0.9993 - val_head_1_loss: 0.0019 - val_head_2_accuracy: 0.9996 - val_head_2_loss: 0.0018 - val_head_3_accuracy: 0.9995 - val_head_3_loss: 0.0018 - val_head_4_accuracy: 0.9994 - val_head_4_loss: 0.0018 - val_head_5_accuracy: 0.9995 - val_head_5_loss: 0.0020 - val_head_6_accuracy: 0.9994 - val_head_6_loss: 0.0017 - val_head_7_accuracy: 0.9993 - val_head_7_loss: 0.0019 - val_loss: 0.0130 - learning_rate: 0.0056\n","\n","Epoch 88: LearningRateScheduler setting learning rate to 0.005537240502379456.\n","Epoch 88/100\n","\u001b[1m315/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m━\u001b[0m \u001b[1m0s\u001b[0m 32ms/step - head_1_accuracy: 1.0000 - head_1_loss: 2.3346e-04 - head_2_accuracy: 1.0000 - head_2_loss: 2.2088e-04 - head_3_accuracy: 1.0000 - head_3_loss: 1.9793e-04 - head_4_accuracy: 1.0000 - head_4_loss: 2.0384e-04 - head_5_accuracy: 1.0000 - head_5_loss: 2.1907e-04 - head_6_accuracy: 1.0000 - head_6_loss: 2.6080e-04 - head_7_accuracy: 1.0000 - head_7_loss: 2.1519e-04 - loss: 0.0016\n","Epoch 88: val_head_1_accuracy did not improve from 0.99948\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m12s\u001b[0m 37ms/step - head_1_accuracy: 1.0000 - head_1_loss: 2.9759e-04 - head_2_accuracy: 0.9999 - head_2_loss: 3.3682e-04 - head_3_accuracy: 0.9999 - head_3_loss: 2.9222e-04 - head_4_accuracy: 1.0000 - head_4_loss: 2.7653e-04 - head_5_accuracy: 1.0000 - head_5_loss: 2.8995e-04 - head_6_accuracy: 0.9999 - head_6_loss: 3.2540e-04 - head_7_accuracy: 0.9999 - head_7_loss: 3.0227e-04 - loss: 0.0021 - val_head_1_accuracy: 0.9993 - val_head_1_loss: 0.0023 - val_head_2_accuracy: 0.9994 - val_head_2_loss: 0.0023 - val_head_3_accuracy: 0.9994 - val_head_3_loss: 0.0022 - val_head_4_accuracy: 0.9993 - val_head_4_loss: 0.0022 - val_head_5_accuracy: 0.9994 - val_head_5_loss: 0.0023 - val_head_6_accuracy: 0.9993 - val_head_6_loss: 0.0021 - val_head_7_accuracy: 0.9993 - val_head_7_loss: 0.0022 - val_loss: 0.0157 - learning_rate: 0.0055\n","\n","Epoch 89: LearningRateScheduler setting learning rate to 0.0054523053441056886.\n","Epoch 89/100\n","\u001b[1m315/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m━\u001b[0m \u001b[1m0s\u001b[0m 32ms/step - head_1_accuracy: 1.0000 - head_1_loss: 1.1503e-04 - head_2_accuracy: 1.0000 - head_2_loss: 1.2128e-04 - head_3_accuracy: 1.0000 - head_3_loss: 9.6094e-05 - head_4_accuracy: 1.0000 - head_4_loss: 1.0965e-04 - head_5_accuracy: 1.0000 - head_5_loss: 1.0112e-04 - head_6_accuracy: 1.0000 - head_6_loss: 1.5791e-04 - head_7_accuracy: 1.0000 - head_7_loss: 1.1283e-04 - loss: 8.1392e-04\n","Epoch 89: val_head_1_accuracy did not improve from 0.99948\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m12s\u001b[0m 37ms/step - head_1_accuracy: 1.0000 - head_1_loss: 1.3876e-04 - head_2_accuracy: 1.0000 - head_2_loss: 1.0479e-04 - head_3_accuracy: 1.0000 - head_3_loss: 1.0035e-04 - head_4_accuracy: 1.0000 - head_4_loss: 1.0897e-04 - head_5_accuracy: 1.0000 - head_5_loss: 1.2242e-04 - head_6_accuracy: 1.0000 - head_6_loss: 1.6665e-04 - head_7_accuracy: 1.0000 - head_7_loss: 1.3367e-04 - loss: 8.7615e-04 - val_head_1_accuracy: 0.9993 - val_head_1_loss: 0.0021 - val_head_2_accuracy: 0.9994 - val_head_2_loss: 0.0021 - val_head_3_accuracy: 0.9995 - val_head_3_loss: 0.0020 - val_head_4_accuracy: 0.9994 - val_head_4_loss: 0.0020 - val_head_5_accuracy: 0.9994 - val_head_5_loss: 0.0021 - val_head_6_accuracy: 0.9994 - val_head_6_loss: 0.0020 - val_head_7_accuracy: 0.9994 - val_head_7_loss: 0.0021 - val_loss: 0.0144 - learning_rate: 0.0055\n","\n","Epoch 90: LearningRateScheduler setting learning rate to 0.00537446661694291.\n","Epoch 90/100\n","\u001b[1m315/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m━\u001b[0m \u001b[1m0s\u001b[0m 32ms/step - head_1_accuracy: 1.0000 - head_1_loss: 1.4118e-04 - head_2_accuracy: 1.0000 - head_2_loss: 1.1353e-04 - head_3_accuracy: 1.0000 - head_3_loss: 1.6487e-04 - head_4_accuracy: 1.0000 - head_4_loss: 1.4746e-04 - head_5_accuracy: 1.0000 - head_5_loss: 1.8185e-04 - head_6_accuracy: 1.0000 - head_6_loss: 1.7238e-04 - head_7_accuracy: 0.9999 - head_7_loss: 2.0658e-04 - loss: 0.0011\n","Epoch 90: val_head_1_accuracy did not improve from 0.99948\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m12s\u001b[0m 37ms/step - head_1_accuracy: 1.0000 - head_1_loss: 1.6596e-04 - head_2_accuracy: 1.0000 - head_2_loss: 1.1339e-04 - head_3_accuracy: 1.0000 - head_3_loss: 1.4669e-04 - head_4_accuracy: 1.0000 - head_4_loss: 1.6208e-04 - head_5_accuracy: 1.0000 - head_5_loss: 1.4965e-04 - head_6_accuracy: 1.0000 - head_6_loss: 2.0034e-04 - head_7_accuracy: 1.0000 - head_7_loss: 1.8976e-04 - loss: 0.0011 - val_head_1_accuracy: 0.9993 - val_head_1_loss: 0.0021 - val_head_2_accuracy: 0.9993 - val_head_2_loss: 0.0021 - val_head_3_accuracy: 0.9995 - val_head_3_loss: 0.0020 - val_head_4_accuracy: 0.9993 - val_head_4_loss: 0.0020 - val_head_5_accuracy: 0.9994 - val_head_5_loss: 0.0021 - val_head_6_accuracy: 0.9993 - val_head_6_loss: 0.0019 - val_head_7_accuracy: 0.9994 - val_head_7_loss: 0.0020 - val_loss: 0.0145 - learning_rate: 0.0054\n","\n","Epoch 91: LearningRateScheduler setting learning rate to 0.00530380269789127.\n","Epoch 91/100\n","\u001b[1m315/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m━\u001b[0m \u001b[1m0s\u001b[0m 32ms/step - head_1_accuracy: 1.0000 - head_1_loss: 1.0435e-04 - head_2_accuracy: 1.0000 - head_2_loss: 8.9685e-05 - head_3_accuracy: 1.0000 - head_3_loss: 7.9087e-05 - head_4_accuracy: 1.0000 - head_4_loss: 9.1651e-05 - head_5_accuracy: 1.0000 - head_5_loss: 1.0462e-04 - head_6_accuracy: 1.0000 - head_6_loss: 1.0044e-04 - head_7_accuracy: 1.0000 - head_7_loss: 1.2077e-04 - loss: 6.9060e-04\n","Epoch 91: val_head_1_accuracy did not improve from 0.99948\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m12s\u001b[0m 37ms/step - head_1_accuracy: 1.0000 - head_1_loss: 1.1386e-04 - head_2_accuracy: 1.0000 - head_2_loss: 1.0589e-04 - head_3_accuracy: 1.0000 - head_3_loss: 9.3045e-05 - head_4_accuracy: 1.0000 - head_4_loss: 1.0494e-04 - head_5_accuracy: 1.0000 - head_5_loss: 1.1245e-04 - head_6_accuracy: 1.0000 - head_6_loss: 1.2104e-04 - head_7_accuracy: 1.0000 - head_7_loss: 1.4178e-04 - loss: 7.9354e-04 - val_head_1_accuracy: 0.9993 - val_head_1_loss: 0.0021 - val_head_2_accuracy: 0.9993 - val_head_2_loss: 0.0020 - val_head_3_accuracy: 0.9994 - val_head_3_loss: 0.0020 - val_head_4_accuracy: 0.9993 - val_head_4_loss: 0.0020 - val_head_5_accuracy: 0.9993 - val_head_5_loss: 0.0021 - val_head_6_accuracy: 0.9994 - val_head_6_loss: 0.0019 - val_head_7_accuracy: 0.9993 - val_head_7_loss: 0.0020 - val_loss: 0.0143 - learning_rate: 0.0053\n","\n","Epoch 92: LearningRateScheduler setting learning rate to 0.005240384739527328.\n","Epoch 92/100\n","\u001b[1m315/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m━\u001b[0m \u001b[1m0s\u001b[0m 32ms/step - head_1_accuracy: 1.0000 - head_1_loss: 1.0875e-04 - head_2_accuracy: 1.0000 - head_2_loss: 1.0360e-04 - head_3_accuracy: 1.0000 - head_3_loss: 8.7263e-05 - head_4_accuracy: 1.0000 - head_4_loss: 1.2273e-04 - head_5_accuracy: 1.0000 - head_5_loss: 1.0469e-04 - head_6_accuracy: 1.0000 - head_6_loss: 1.4120e-04 - head_7_accuracy: 1.0000 - head_7_loss: 1.3611e-04 - loss: 8.0435e-04\n","Epoch 92: val_head_1_accuracy did not improve from 0.99948\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m12s\u001b[0m 37ms/step - head_1_accuracy: 1.0000 - head_1_loss: 9.6472e-05 - head_2_accuracy: 1.0000 - head_2_loss: 8.7868e-05 - head_3_accuracy: 1.0000 - head_3_loss: 8.5625e-05 - head_4_accuracy: 1.0000 - head_4_loss: 9.6117e-05 - head_5_accuracy: 1.0000 - head_5_loss: 9.3086e-05 - head_6_accuracy: 1.0000 - head_6_loss: 1.1204e-04 - head_7_accuracy: 1.0000 - head_7_loss: 1.1931e-04 - loss: 6.9096e-04 - val_head_1_accuracy: 0.9993 - val_head_1_loss: 0.0019 - val_head_2_accuracy: 0.9994 - val_head_2_loss: 0.0018 - val_head_3_accuracy: 0.9994 - val_head_3_loss: 0.0018 - val_head_4_accuracy: 0.9994 - val_head_4_loss: 0.0018 - val_head_5_accuracy: 0.9995 - val_head_5_loss: 0.0019 - val_head_6_accuracy: 0.9993 - val_head_6_loss: 0.0017 - val_head_7_accuracy: 0.9993 - val_head_7_loss: 0.0018 - val_loss: 0.0130 - learning_rate: 0.0052\n","\n","Epoch 93: LearningRateScheduler setting learning rate to 0.005184276598359447.\n","Epoch 93/100\n","\u001b[1m315/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m━\u001b[0m \u001b[1m0s\u001b[0m 32ms/step - head_1_accuracy: 1.0000 - head_1_loss: 9.2980e-05 - head_2_accuracy: 1.0000 - head_2_loss: 7.7929e-05 - head_3_accuracy: 1.0000 - head_3_loss: 8.8888e-05 - head_4_accuracy: 1.0000 - head_4_loss: 9.5538e-05 - head_5_accuracy: 1.0000 - head_5_loss: 9.1141e-05 - head_6_accuracy: 1.0000 - head_6_loss: 1.2140e-04 - head_7_accuracy: 1.0000 - head_7_loss: 1.1933e-04 - loss: 6.8720e-04\n","Epoch 93: val_head_1_accuracy did not improve from 0.99948\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m12s\u001b[0m 37ms/step - head_1_accuracy: 1.0000 - head_1_loss: 9.3494e-05 - head_2_accuracy: 1.0000 - head_2_loss: 8.2094e-05 - head_3_accuracy: 1.0000 - head_3_loss: 8.6464e-05 - head_4_accuracy: 1.0000 - head_4_loss: 8.6850e-05 - head_5_accuracy: 1.0000 - head_5_loss: 8.7925e-05 - head_6_accuracy: 1.0000 - head_6_loss: 1.1404e-04 - head_7_accuracy: 1.0000 - head_7_loss: 1.1498e-04 - loss: 6.6623e-04 - val_head_1_accuracy: 0.9993 - val_head_1_loss: 0.0019 - val_head_2_accuracy: 0.9994 - val_head_2_loss: 0.0018 - val_head_3_accuracy: 0.9994 - val_head_3_loss: 0.0018 - val_head_4_accuracy: 0.9993 - val_head_4_loss: 0.0017 - val_head_5_accuracy: 0.9994 - val_head_5_loss: 0.0019 - val_head_6_accuracy: 0.9994 - val_head_6_loss: 0.0016 - val_head_7_accuracy: 0.9993 - val_head_7_loss: 0.0018 - val_loss: 0.0124 - learning_rate: 0.0052\n","\n","Epoch 94: LearningRateScheduler setting learning rate to 0.005135534770529701.\n","Epoch 94/100\n","\u001b[1m315/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m━\u001b[0m \u001b[1m0s\u001b[0m 32ms/step - head_1_accuracy: 1.0000 - head_1_loss: 1.5088e-04 - head_2_accuracy: 1.0000 - head_2_loss: 1.1826e-04 - head_3_accuracy: 1.0000 - head_3_loss: 1.1441e-04 - head_4_accuracy: 1.0000 - head_4_loss: 1.1924e-04 - head_5_accuracy: 1.0000 - head_5_loss: 1.7093e-04 - head_6_accuracy: 1.0000 - head_6_loss: 1.3509e-04 - head_7_accuracy: 1.0000 - head_7_loss: 1.3689e-04 - loss: 9.4570e-04\n","Epoch 94: val_head_1_accuracy did not improve from 0.99948\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m12s\u001b[0m 37ms/step - head_1_accuracy: 1.0000 - head_1_loss: 1.9579e-04 - head_2_accuracy: 0.9999 - head_2_loss: 1.7064e-04 - head_3_accuracy: 1.0000 - head_3_loss: 1.3675e-04 - head_4_accuracy: 1.0000 - head_4_loss: 1.4546e-04 - head_5_accuracy: 1.0000 - head_5_loss: 1.8092e-04 - head_6_accuracy: 1.0000 - head_6_loss: 1.5994e-04 - head_7_accuracy: 1.0000 - head_7_loss: 1.5968e-04 - loss: 0.0011 - val_head_1_accuracy: 0.9992 - val_head_1_loss: 0.0019 - val_head_2_accuracy: 0.9993 - val_head_2_loss: 0.0018 - val_head_3_accuracy: 0.9993 - val_head_3_loss: 0.0018 - val_head_4_accuracy: 0.9993 - val_head_4_loss: 0.0018 - val_head_5_accuracy: 0.9994 - val_head_5_loss: 0.0019 - val_head_6_accuracy: 0.9993 - val_head_6_loss: 0.0017 - val_head_7_accuracy: 0.9993 - val_head_7_loss: 0.0018 - val_loss: 0.0129 - learning_rate: 0.0051\n","\n","Epoch 95: LearningRateScheduler setting learning rate to 0.005094208334927043.\n","Epoch 95/100\n","\u001b[1m315/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m━\u001b[0m \u001b[1m0s\u001b[0m 32ms/step - head_1_accuracy: 1.0000 - head_1_loss: 1.4935e-04 - head_2_accuracy: 1.0000 - head_2_loss: 1.1575e-04 - head_3_accuracy: 1.0000 - head_3_loss: 1.2238e-04 - head_4_accuracy: 0.9999 - head_4_loss: 2.8303e-04 - head_5_accuracy: 1.0000 - head_5_loss: 1.5358e-04 - head_6_accuracy: 1.0000 - head_6_loss: 1.3947e-04 - head_7_accuracy: 1.0000 - head_7_loss: 1.4185e-04 - loss: 0.0011\n","Epoch 95: val_head_1_accuracy did not improve from 0.99948\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m12s\u001b[0m 37ms/step - head_1_accuracy: 1.0000 - head_1_loss: 1.3398e-04 - head_2_accuracy: 1.0000 - head_2_loss: 9.2466e-05 - head_3_accuracy: 1.0000 - head_3_loss: 1.1936e-04 - head_4_accuracy: 1.0000 - head_4_loss: 1.6123e-04 - head_5_accuracy: 1.0000 - head_5_loss: 1.2697e-04 - head_6_accuracy: 1.0000 - head_6_loss: 1.2716e-04 - head_7_accuracy: 1.0000 - head_7_loss: 1.2889e-04 - loss: 8.9042e-04 - val_head_1_accuracy: 0.9991 - val_head_1_loss: 0.0021 - val_head_2_accuracy: 0.9993 - val_head_2_loss: 0.0021 - val_head_3_accuracy: 0.9993 - val_head_3_loss: 0.0020 - val_head_4_accuracy: 0.9993 - val_head_4_loss: 0.0020 - val_head_5_accuracy: 0.9994 - val_head_5_loss: 0.0021 - val_head_6_accuracy: 0.9993 - val_head_6_loss: 0.0019 - val_head_7_accuracy: 0.9993 - val_head_7_loss: 0.0020 - val_loss: 0.0143 - learning_rate: 0.0051\n","\n","Epoch 96: LearningRateScheduler setting learning rate to 0.0050603389037690355.\n","Epoch 96/100\n","\u001b[1m315/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m━\u001b[0m \u001b[1m0s\u001b[0m 32ms/step - head_1_accuracy: 1.0000 - head_1_loss: 1.1725e-04 - head_2_accuracy: 1.0000 - head_2_loss: 1.0787e-04 - head_3_accuracy: 1.0000 - head_3_loss: 1.0903e-04 - head_4_accuracy: 1.0000 - head_4_loss: 1.2004e-04 - head_5_accuracy: 1.0000 - head_5_loss: 1.3524e-04 - head_6_accuracy: 1.0000 - head_6_loss: 1.4102e-04 - head_7_accuracy: 1.0000 - head_7_loss: 1.3792e-04 - loss: 8.6838e-04\n","Epoch 96: val_head_1_accuracy did not improve from 0.99948\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m12s\u001b[0m 37ms/step - head_1_accuracy: 1.0000 - head_1_loss: 1.4755e-04 - head_2_accuracy: 1.0000 - head_2_loss: 1.3285e-04 - head_3_accuracy: 1.0000 - head_3_loss: 1.3260e-04 - head_4_accuracy: 1.0000 - head_4_loss: 1.3994e-04 - head_5_accuracy: 1.0000 - head_5_loss: 1.6530e-04 - head_6_accuracy: 1.0000 - head_6_loss: 1.8415e-04 - head_7_accuracy: 1.0000 - head_7_loss: 1.5019e-04 - loss: 0.0011 - val_head_1_accuracy: 0.9991 - val_head_1_loss: 0.0024 - val_head_2_accuracy: 0.9992 - val_head_2_loss: 0.0023 - val_head_3_accuracy: 0.9992 - val_head_3_loss: 0.0023 - val_head_4_accuracy: 0.9993 - val_head_4_loss: 0.0022 - val_head_5_accuracy: 0.9992 - val_head_5_loss: 0.0022 - val_head_6_accuracy: 0.9993 - val_head_6_loss: 0.0021 - val_head_7_accuracy: 0.9993 - val_head_7_loss: 0.0022 - val_loss: 0.0157 - learning_rate: 0.0051\n","\n","Epoch 97: LearningRateScheduler setting learning rate to 0.005033960580701866.\n","Epoch 97/100\n","\u001b[1m315/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m━\u001b[0m \u001b[1m0s\u001b[0m 32ms/step - head_1_accuracy: 0.9999 - head_1_loss: 2.3891e-04 - head_2_accuracy: 0.9999 - head_2_loss: 2.4737e-04 - head_3_accuracy: 1.0000 - head_3_loss: 1.6828e-04 - head_4_accuracy: 1.0000 - head_4_loss: 1.6392e-04 - head_5_accuracy: 1.0000 - head_5_loss: 1.4085e-04 - head_6_accuracy: 1.0000 - head_6_loss: 2.3959e-04 - head_7_accuracy: 1.0000 - head_7_loss: 1.4484e-04 - loss: 0.0013\n","Epoch 97: val_head_1_accuracy did not improve from 0.99948\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m12s\u001b[0m 37ms/step - head_1_accuracy: 1.0000 - head_1_loss: 1.4708e-04 - head_2_accuracy: 1.0000 - head_2_loss: 1.3732e-04 - head_3_accuracy: 1.0000 - head_3_loss: 1.3711e-04 - head_4_accuracy: 1.0000 - head_4_loss: 1.1819e-04 - head_5_accuracy: 1.0000 - head_5_loss: 1.0901e-04 - head_6_accuracy: 1.0000 - head_6_loss: 1.5427e-04 - head_7_accuracy: 1.0000 - head_7_loss: 1.2895e-04 - loss: 9.3246e-04 - val_head_1_accuracy: 0.9993 - val_head_1_loss: 0.0021 - val_head_2_accuracy: 0.9993 - val_head_2_loss: 0.0021 - val_head_3_accuracy: 0.9993 - val_head_3_loss: 0.0020 - val_head_4_accuracy: 0.9993 - val_head_4_loss: 0.0020 - val_head_5_accuracy: 0.9993 - val_head_5_loss: 0.0021 - val_head_6_accuracy: 0.9993 - val_head_6_loss: 0.0019 - val_head_7_accuracy: 0.9993 - val_head_7_loss: 0.0020 - val_loss: 0.0143 - learning_rate: 0.0050\n","\n","Epoch 98: LearningRateScheduler setting learning rate to 0.0050150999264608675.\n","Epoch 98/100\n","\u001b[1m315/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m━\u001b[0m \u001b[1m0s\u001b[0m 32ms/step - head_1_accuracy: 1.0000 - head_1_loss: 1.1384e-04 - head_2_accuracy: 1.0000 - head_2_loss: 8.1419e-05 - head_3_accuracy: 1.0000 - head_3_loss: 8.1153e-05 - head_4_accuracy: 1.0000 - head_4_loss: 1.1083e-04 - head_5_accuracy: 1.0000 - head_5_loss: 9.4056e-05 - head_6_accuracy: 1.0000 - head_6_loss: 1.0138e-04 - head_7_accuracy: 1.0000 - head_7_loss: 9.8911e-05 - loss: 6.8160e-04\n","Epoch 98: val_head_1_accuracy did not improve from 0.99948\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m12s\u001b[0m 37ms/step - head_1_accuracy: 1.0000 - head_1_loss: 1.0861e-04 - head_2_accuracy: 1.0000 - head_2_loss: 8.7349e-05 - head_3_accuracy: 1.0000 - head_3_loss: 8.7218e-05 - head_4_accuracy: 1.0000 - head_4_loss: 1.1323e-04 - head_5_accuracy: 1.0000 - head_5_loss: 1.0507e-04 - head_6_accuracy: 1.0000 - head_6_loss: 1.0839e-04 - head_7_accuracy: 1.0000 - head_7_loss: 1.0313e-04 - loss: 7.1340e-04 - val_head_1_accuracy: 0.9993 - val_head_1_loss: 0.0020 - val_head_2_accuracy: 0.9993 - val_head_2_loss: 0.0020 - val_head_3_accuracy: 0.9993 - val_head_3_loss: 0.0019 - val_head_4_accuracy: 0.9993 - val_head_4_loss: 0.0019 - val_head_5_accuracy: 0.9993 - val_head_5_loss: 0.0020 - val_head_6_accuracy: 0.9993 - val_head_6_loss: 0.0018 - val_head_7_accuracy: 0.9993 - val_head_7_loss: 0.0019 - val_loss: 0.0136 - learning_rate: 0.0050\n","\n","Epoch 99: LearningRateScheduler setting learning rate to 0.005003775932126112.\n","Epoch 99/100\n","\u001b[1m315/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m━\u001b[0m \u001b[1m0s\u001b[0m 32ms/step - head_1_accuracy: 1.0000 - head_1_loss: 1.5561e-04 - head_2_accuracy: 1.0000 - head_2_loss: 9.7202e-05 - head_3_accuracy: 1.0000 - head_3_loss: 8.3762e-05 - head_4_accuracy: 1.0000 - head_4_loss: 9.1128e-05 - head_5_accuracy: 1.0000 - head_5_loss: 9.0702e-05 - head_6_accuracy: 1.0000 - head_6_loss: 1.0862e-04 - head_7_accuracy: 1.0000 - head_7_loss: 1.2747e-04 - loss: 7.5448e-04\n","Epoch 99: val_head_1_accuracy did not improve from 0.99948\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m12s\u001b[0m 37ms/step - head_1_accuracy: 1.0000 - head_1_loss: 1.8433e-04 - head_2_accuracy: 1.0000 - head_2_loss: 8.7546e-05 - head_3_accuracy: 1.0000 - head_3_loss: 8.2426e-05 - head_4_accuracy: 1.0000 - head_4_loss: 9.5181e-05 - head_5_accuracy: 1.0000 - head_5_loss: 9.5836e-05 - head_6_accuracy: 1.0000 - head_6_loss: 1.2117e-04 - head_7_accuracy: 1.0000 - head_7_loss: 1.2388e-04 - loss: 7.9073e-04 - val_head_1_accuracy: 0.9993 - val_head_1_loss: 0.0023 - val_head_2_accuracy: 0.9993 - val_head_2_loss: 0.0022 - val_head_3_accuracy: 0.9993 - val_head_3_loss: 0.0021 - val_head_4_accuracy: 0.9993 - val_head_4_loss: 0.0022 - val_head_5_accuracy: 0.9994 - val_head_5_loss: 0.0022 - val_head_6_accuracy: 0.9993 - val_head_6_loss: 0.0020 - val_head_7_accuracy: 0.9993 - val_head_7_loss: 0.0022 - val_loss: 0.0154 - learning_rate: 0.0050\n","\n","Epoch 100: LearningRateScheduler setting learning rate to 0.005.\n","Epoch 100/100\n","\u001b[1m315/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m━\u001b[0m \u001b[1m0s\u001b[0m 32ms/step - head_1_accuracy: 1.0000 - head_1_loss: 7.2669e-05 - head_2_accuracy: 1.0000 - head_2_loss: 6.2626e-05 - head_3_accuracy: 1.0000 - head_3_loss: 6.6714e-05 - head_4_accuracy: 1.0000 - head_4_loss: 7.5061e-05 - head_5_accuracy: 1.0000 - head_5_loss: 7.8065e-05 - head_6_accuracy: 1.0000 - head_6_loss: 7.5674e-05 - head_7_accuracy: 1.0000 - head_7_loss: 8.7619e-05 - loss: 5.1843e-04\n","Epoch 100: val_head_1_accuracy did not improve from 0.99948\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m12s\u001b[0m 37ms/step - head_1_accuracy: 1.0000 - head_1_loss: 7.7865e-05 - head_2_accuracy: 1.0000 - head_2_loss: 6.8537e-05 - head_3_accuracy: 1.0000 - head_3_loss: 6.8809e-05 - head_4_accuracy: 1.0000 - head_4_loss: 7.9690e-05 - head_5_accuracy: 1.0000 - head_5_loss: 8.6793e-05 - head_6_accuracy: 1.0000 - head_6_loss: 8.3099e-05 - head_7_accuracy: 1.0000 - head_7_loss: 9.0220e-05 - loss: 5.5530e-04 - val_head_1_accuracy: 0.9991 - val_head_1_loss: 0.0020 - val_head_2_accuracy: 0.9993 - val_head_2_loss: 0.0020 - val_head_3_accuracy: 0.9993 - val_head_3_loss: 0.0019 - val_head_4_accuracy: 0.9993 - val_head_4_loss: 0.0020 - val_head_5_accuracy: 0.9994 - val_head_5_loss: 0.0020 - val_head_6_accuracy: 0.9993 - val_head_6_loss: 0.0018 - val_head_7_accuracy: 0.9993 - val_head_7_loss: 0.0019 - val_loss: 0.0137 - learning_rate: 0.0050\n"]},{"name":"stderr","output_type":"stream","text":["WARNING:absl:You are saving your model as an HDF5 file via `model.save()` or `keras.saving.save_model(model)`. This file format is considered legacy. We recommend using instead the native Keras format, e.g. `model.save('my_model.keras')` or `keras.saving.save_model(model, 'my_model.keras')`. \n"]},{"name":"stdout","output_type":"stream","text":["Saved .h5 models: AlexNet_CNN_MultiHead_best.h5, AlexNet_CNN_MultiHead_final.h5\n"]},{"data":{"image/png":"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\n","text/plain":["
"]},"metadata":{},"output_type":"display_data"},{"name":"stdout","output_type":"stream","text":["Saved training plots directly to /kaggle/working/results\n","\n","============================== Training GFNet_MultiHead ==============================\n"]},{"data":{"image/png":"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\n","text/plain":["
"]},"metadata":{},"output_type":"display_data"},{"data":{"image/png":"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\n","text/plain":["
"]},"metadata":{},"output_type":"display_data"},{"name":"stdout","output_type":"stream","text":["\n","Epoch 1: LearningRateScheduler setting learning rate to 0.0006.\n","Epoch 1/100\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 174ms/step - head_1_accuracy: 0.6389 - head_1_loss: 0.9949 - head_2_accuracy: 0.6409 - head_2_loss: 0.9590 - head_3_accuracy: 0.6382 - head_3_loss: 1.0245 - head_4_accuracy: 0.6423 - head_4_loss: 0.9430 - head_5_accuracy: 0.6421 - head_5_loss: 0.9379 - head_6_accuracy: 0.6447 - head_6_loss: 0.9983 - head_7_accuracy: 0.6325 - head_7_loss: 1.0081 - loss: 6.8657\n","Epoch 1: val_head_1_accuracy improved from None to 0.75627, saving model to /kaggle/working/results/models/GFNet_MultiHead_best.keras\n","\n","Epoch 1: finished saving model to /kaggle/working/results/models/GFNet_MultiHead_best.keras\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m122s\u001b[0m 214ms/step - head_1_accuracy: 0.7292 - head_1_loss: 0.7354 - head_2_accuracy: 0.7312 - head_2_loss: 0.7260 - head_3_accuracy: 0.7297 - head_3_loss: 0.7423 - head_4_accuracy: 0.7299 - head_4_loss: 0.7207 - head_5_accuracy: 0.7304 - head_5_loss: 0.7208 - head_6_accuracy: 0.7336 - head_6_loss: 0.7330 - head_7_accuracy: 0.7252 - head_7_loss: 0.7444 - loss: 5.1240 - val_head_1_accuracy: 0.7563 - val_head_1_loss: 0.6554 - val_head_2_accuracy: 0.7491 - val_head_2_loss: 0.6826 - val_head_3_accuracy: 0.7702 - val_head_3_loss: 0.6366 - val_head_4_accuracy: 0.7595 - val_head_4_loss: 0.6646 - val_head_5_accuracy: 0.7605 - val_head_5_loss: 0.6462 - val_head_6_accuracy: 0.7719 - val_head_6_loss: 0.6344 - val_head_7_accuracy: 0.7560 - val_head_7_loss: 0.6712 - val_loss: 4.6066 - learning_rate: 6.0000e-04\n","\n","Epoch 2: LearningRateScheduler setting learning rate to 0.0005998489878878364.\n","Epoch 2/100\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 89ms/step - head_1_accuracy: 0.8050 - head_1_loss: 0.5356 - head_2_accuracy: 0.8058 - head_2_loss: 0.5327 - head_3_accuracy: 0.8060 - head_3_loss: 0.5321 - head_4_accuracy: 0.8064 - head_4_loss: 0.5354 - head_5_accuracy: 0.8054 - head_5_loss: 0.5353 - head_6_accuracy: 0.8070 - head_6_loss: 0.5331 - head_7_accuracy: 0.8047 - head_7_loss: 0.5360 - loss: 3.7403\n","Epoch 2: val_head_1_accuracy improved from 0.75627 to 0.82121, saving model to /kaggle/working/results/models/GFNet_MultiHead_best.keras\n","\n","Epoch 2: finished saving model to /kaggle/working/results/models/GFNet_MultiHead_best.keras\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m32s\u001b[0m 100ms/step - head_1_accuracy: 0.8158 - head_1_loss: 0.5079 - head_2_accuracy: 0.8154 - head_2_loss: 0.5077 - head_3_accuracy: 0.8164 - head_3_loss: 0.5060 - head_4_accuracy: 0.8170 - head_4_loss: 0.5072 - head_5_accuracy: 0.8150 - head_5_loss: 0.5091 - head_6_accuracy: 0.8161 - head_6_loss: 0.5059 - head_7_accuracy: 0.8157 - head_7_loss: 0.5089 - loss: 3.5534 - val_head_1_accuracy: 0.8212 - val_head_1_loss: 0.4566 - val_head_2_accuracy: 0.8231 - val_head_2_loss: 0.4564 - val_head_3_accuracy: 0.8160 - val_head_3_loss: 0.4680 - val_head_4_accuracy: 0.8263 - val_head_4_loss: 0.4547 - val_head_5_accuracy: 0.8130 - val_head_5_loss: 0.4773 - val_head_6_accuracy: 0.8254 - val_head_6_loss: 0.4498 - val_head_7_accuracy: 0.8201 - val_head_7_loss: 0.4672 - val_loss: 3.2409 - learning_rate: 5.9985e-04\n","\n","Epoch 3: LearningRateScheduler setting learning rate to 0.0005993961036077416.\n","Epoch 3/100\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 89ms/step - head_1_accuracy: 0.8391 - head_1_loss: 0.4508 - head_2_accuracy: 0.8392 - head_2_loss: 0.4482 - head_3_accuracy: 0.8373 - head_3_loss: 0.4491 - head_4_accuracy: 0.8402 - head_4_loss: 0.4490 - head_5_accuracy: 0.8409 - head_5_loss: 0.4473 - head_6_accuracy: 0.8386 - head_6_loss: 0.4472 - head_7_accuracy: 0.8402 - head_7_loss: 0.4472 - loss: 3.1389\n","Epoch 3: val_head_1_accuracy improved from 0.82121 to 0.87071, saving model to /kaggle/working/results/models/GFNet_MultiHead_best.keras\n","\n","Epoch 3: finished saving model to /kaggle/working/results/models/GFNet_MultiHead_best.keras\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m32s\u001b[0m 101ms/step - head_1_accuracy: 0.8453 - head_1_loss: 0.4285 - head_2_accuracy: 0.8448 - head_2_loss: 0.4272 - head_3_accuracy: 0.8445 - head_3_loss: 0.4276 - head_4_accuracy: 0.8461 - head_4_loss: 0.4267 - head_5_accuracy: 0.8464 - head_5_loss: 0.4266 - head_6_accuracy: 0.8449 - head_6_loss: 0.4265 - head_7_accuracy: 0.8453 - head_7_loss: 0.4271 - loss: 2.9903 - val_head_1_accuracy: 0.8707 - val_head_1_loss: 0.4147 - val_head_2_accuracy: 0.8711 - val_head_2_loss: 0.4119 - val_head_3_accuracy: 0.8711 - val_head_3_loss: 0.4130 - val_head_4_accuracy: 0.8706 - val_head_4_loss: 0.4201 - val_head_5_accuracy: 0.8705 - val_head_5_loss: 0.4201 - val_head_6_accuracy: 0.8712 - val_head_6_loss: 0.4155 - val_head_7_accuracy: 0.8706 - val_head_7_loss: 0.4070 - val_loss: 2.9123 - learning_rate: 5.9940e-04\n","\n","Epoch 4: LearningRateScheduler setting learning rate to 0.0005986418031757967.\n","Epoch 4/100\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 89ms/step - head_1_accuracy: 0.8782 - head_1_loss: 0.3347 - head_2_accuracy: 0.8810 - head_2_loss: 0.3323 - head_3_accuracy: 0.8766 - head_3_loss: 0.3355 - head_4_accuracy: 0.8768 - head_4_loss: 0.3361 - head_5_accuracy: 0.8777 - head_5_loss: 0.3342 - head_6_accuracy: 0.8796 - head_6_loss: 0.3338 - head_7_accuracy: 0.8765 - head_7_loss: 0.3354 - loss: 2.3419\n","Epoch 4: val_head_1_accuracy improved from 0.87071 to 0.89862, saving model to /kaggle/working/results/models/GFNet_MultiHead_best.keras\n","\n","Epoch 4: finished saving model to /kaggle/working/results/models/GFNet_MultiHead_best.keras\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m32s\u001b[0m 101ms/step - head_1_accuracy: 0.8816 - head_1_loss: 0.3226 - head_2_accuracy: 0.8825 - head_2_loss: 0.3226 - head_3_accuracy: 0.8799 - head_3_loss: 0.3234 - head_4_accuracy: 0.8801 - head_4_loss: 0.3237 - head_5_accuracy: 0.8803 - head_5_loss: 0.3230 - head_6_accuracy: 0.8819 - head_6_loss: 0.3222 - head_7_accuracy: 0.8799 - head_7_loss: 0.3245 - loss: 2.2620 - val_head_1_accuracy: 0.8986 - val_head_1_loss: 0.2738 - val_head_2_accuracy: 0.9007 - val_head_2_loss: 0.2730 - val_head_3_accuracy: 0.8977 - val_head_3_loss: 0.2800 - val_head_4_accuracy: 0.8963 - val_head_4_loss: 0.2788 - val_head_5_accuracy: 0.8988 - val_head_5_loss: 0.2734 - val_head_6_accuracy: 0.8980 - val_head_6_loss: 0.2778 - val_head_7_accuracy: 0.9005 - val_head_7_loss: 0.2745 - val_loss: 1.9393 - learning_rate: 5.9864e-04\n","\n","Epoch 5: LearningRateScheduler setting learning rate to 0.000597586846108597.\n","Epoch 5/100\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 89ms/step - head_1_accuracy: 0.8966 - head_1_loss: 0.2767 - head_2_accuracy: 0.8955 - head_2_loss: 0.2780 - head_3_accuracy: 0.8960 - head_3_loss: 0.2778 - head_4_accuracy: 0.8960 - head_4_loss: 0.2765 - head_5_accuracy: 0.8964 - head_5_loss: 0.2760 - head_6_accuracy: 0.8962 - head_6_loss: 0.2757 - head_7_accuracy: 0.8975 - head_7_loss: 0.2769 - loss: 1.9377\n","Epoch 5: val_head_1_accuracy improved from 0.89862 to 0.92170, saving model to /kaggle/working/results/models/GFNet_MultiHead_best.keras\n","\n","Epoch 5: finished saving model to /kaggle/working/results/models/GFNet_MultiHead_best.keras\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m32s\u001b[0m 100ms/step - head_1_accuracy: 0.8989 - head_1_loss: 0.2712 - head_2_accuracy: 0.8980 - head_2_loss: 0.2718 - head_3_accuracy: 0.8985 - head_3_loss: 0.2721 - head_4_accuracy: 0.8989 - head_4_loss: 0.2717 - head_5_accuracy: 0.8991 - head_5_loss: 0.2704 - head_6_accuracy: 0.8982 - head_6_loss: 0.2708 - head_7_accuracy: 0.8994 - head_7_loss: 0.2707 - loss: 1.8989 - val_head_1_accuracy: 0.9217 - val_head_1_loss: 0.2157 - val_head_2_accuracy: 0.9204 - val_head_2_loss: 0.2195 - val_head_3_accuracy: 0.9198 - val_head_3_loss: 0.2201 - val_head_4_accuracy: 0.9197 - val_head_4_loss: 0.2204 - val_head_5_accuracy: 0.9219 - val_head_5_loss: 0.2167 - val_head_6_accuracy: 0.9203 - val_head_6_loss: 0.2191 - val_head_7_accuracy: 0.9199 - val_head_7_loss: 0.2229 - val_loss: 1.5408 - learning_rate: 5.9759e-04\n","\n","Epoch 6: LearningRateScheduler setting learning rate to 0.0005962322946584844.\n","Epoch 6/100\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 89ms/step - head_1_accuracy: 0.9136 - head_1_loss: 0.2336 - head_2_accuracy: 0.9142 - head_2_loss: 0.2334 - head_3_accuracy: 0.9142 - head_3_loss: 0.2326 - head_4_accuracy: 0.9150 - head_4_loss: 0.2330 - head_5_accuracy: 0.9142 - head_5_loss: 0.2332 - head_6_accuracy: 0.9158 - head_6_loss: 0.2312 - head_7_accuracy: 0.9133 - head_7_loss: 0.2340 - loss: 1.6310\n","Epoch 6: val_head_1_accuracy improved from 0.92170 to 0.94241, saving model to /kaggle/working/results/models/GFNet_MultiHead_best.keras\n","\n","Epoch 6: finished saving model to /kaggle/working/results/models/GFNet_MultiHead_best.keras\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m32s\u001b[0m 101ms/step - head_1_accuracy: 0.9161 - head_1_loss: 0.2277 - head_2_accuracy: 0.9164 - head_2_loss: 0.2281 - head_3_accuracy: 0.9165 - head_3_loss: 0.2278 - head_4_accuracy: 0.9171 - head_4_loss: 0.2276 - head_5_accuracy: 0.9161 - head_5_loss: 0.2277 - head_6_accuracy: 0.9177 - head_6_loss: 0.2268 - head_7_accuracy: 0.9156 - head_7_loss: 0.2278 - loss: 1.5940 - val_head_1_accuracy: 0.9424 - val_head_1_loss: 0.1635 - val_head_2_accuracy: 0.9430 - val_head_2_loss: 0.1602 - val_head_3_accuracy: 0.9429 - val_head_3_loss: 0.1621 - val_head_4_accuracy: 0.9431 - val_head_4_loss: 0.1617 - val_head_5_accuracy: 0.9445 - val_head_5_loss: 0.1583 - val_head_6_accuracy: 0.9449 - val_head_6_loss: 0.1590 - val_head_7_accuracy: 0.9435 - val_head_7_loss: 0.1622 - val_loss: 1.1323 - learning_rate: 5.9623e-04\n","\n","Epoch 7: LearningRateScheduler setting learning rate to 0.0005945795127439488.\n","Epoch 7/100\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 89ms/step - head_1_accuracy: 0.9339 - head_1_loss: 0.1785 - head_2_accuracy: 0.9344 - head_2_loss: 0.1780 - head_3_accuracy: 0.9341 - head_3_loss: 0.1791 - head_4_accuracy: 0.9340 - head_4_loss: 0.1787 - head_5_accuracy: 0.9341 - head_5_loss: 0.1788 - head_6_accuracy: 0.9361 - head_6_loss: 0.1789 - head_7_accuracy: 0.9342 - head_7_loss: 0.1789 - loss: 1.2510\n","Epoch 7: val_head_1_accuracy did not improve from 0.94241\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m30s\u001b[0m 95ms/step - head_1_accuracy: 0.9330 - head_1_loss: 0.1808 - head_2_accuracy: 0.9331 - head_2_loss: 0.1810 - head_3_accuracy: 0.9331 - head_3_loss: 0.1810 - head_4_accuracy: 0.9335 - head_4_loss: 0.1810 - head_5_accuracy: 0.9332 - head_5_loss: 0.1811 - head_6_accuracy: 0.9344 - head_6_loss: 0.1810 - head_7_accuracy: 0.9330 - head_7_loss: 0.1817 - loss: 1.2670 - val_head_1_accuracy: 0.9245 - val_head_1_loss: 0.2225 - val_head_2_accuracy: 0.9249 - val_head_2_loss: 0.2206 - val_head_3_accuracy: 0.9213 - val_head_3_loss: 0.2243 - val_head_4_accuracy: 0.9238 - val_head_4_loss: 0.2213 - val_head_5_accuracy: 0.9216 - val_head_5_loss: 0.2281 - val_head_6_accuracy: 0.9255 - val_head_6_loss: 0.2162 - val_head_7_accuracy: 0.9233 - val_head_7_loss: 0.2205 - val_loss: 1.5557 - learning_rate: 5.9458e-04\n","\n","Epoch 8: LearningRateScheduler setting learning rate to 0.0005926301645762779.\n","Epoch 8/100\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 89ms/step - head_1_accuracy: 0.9359 - head_1_loss: 0.1779 - head_2_accuracy: 0.9353 - head_2_loss: 0.1762 - head_3_accuracy: 0.9360 - head_3_loss: 0.1771 - head_4_accuracy: 0.9359 - head_4_loss: 0.1763 - head_5_accuracy: 0.9356 - head_5_loss: 0.1768 - head_6_accuracy: 0.9358 - head_6_loss: 0.1766 - head_7_accuracy: 0.9357 - head_7_loss: 0.1763 - loss: 1.2372\n","Epoch 8: val_head_1_accuracy improved from 0.94241 to 0.96133, saving model to /kaggle/working/results/models/GFNet_MultiHead_best.keras\n","\n","Epoch 8: finished saving model to /kaggle/working/results/models/GFNet_MultiHead_best.keras\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m32s\u001b[0m 100ms/step - head_1_accuracy: 0.9392 - head_1_loss: 0.1673 - head_2_accuracy: 0.9392 - head_2_loss: 0.1658 - head_3_accuracy: 0.9387 - head_3_loss: 0.1665 - head_4_accuracy: 0.9388 - head_4_loss: 0.1666 - head_5_accuracy: 0.9386 - head_5_loss: 0.1665 - head_6_accuracy: 0.9392 - head_6_loss: 0.1657 - head_7_accuracy: 0.9388 - head_7_loss: 0.1663 - loss: 1.1651 - val_head_1_accuracy: 0.9613 - val_head_1_loss: 0.1092 - val_head_2_accuracy: 0.9615 - val_head_2_loss: 0.1112 - val_head_3_accuracy: 0.9614 - val_head_3_loss: 0.1098 - val_head_4_accuracy: 0.9612 - val_head_4_loss: 0.1098 - val_head_5_accuracy: 0.9621 - val_head_5_loss: 0.1084 - val_head_6_accuracy: 0.9614 - val_head_6_loss: 0.1105 - val_head_7_accuracy: 0.9608 - val_head_7_loss: 0.1106 - val_loss: 0.7721 - learning_rate: 5.9263e-04\n","\n","Epoch 9: LearningRateScheduler setting learning rate to 0.0005903862129838371.\n","Epoch 9/100\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 89ms/step - head_1_accuracy: 0.9457 - head_1_loss: 0.1518 - head_2_accuracy: 0.9453 - head_2_loss: 0.1515 - head_3_accuracy: 0.9452 - head_3_loss: 0.1517 - head_4_accuracy: 0.9456 - head_4_loss: 0.1519 - head_5_accuracy: 0.9453 - head_5_loss: 0.1507 - head_6_accuracy: 0.9453 - head_6_loss: 0.1515 - head_7_accuracy: 0.9456 - head_7_loss: 0.1524 - loss: 1.0615\n","Epoch 9: val_head_1_accuracy did not improve from 0.96133\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m30s\u001b[0m 95ms/step - head_1_accuracy: 0.9419 - head_1_loss: 0.1592 - head_2_accuracy: 0.9415 - head_2_loss: 0.1582 - head_3_accuracy: 0.9414 - head_3_loss: 0.1586 - head_4_accuracy: 0.9420 - head_4_loss: 0.1587 - head_5_accuracy: 0.9418 - head_5_loss: 0.1586 - head_6_accuracy: 0.9412 - head_6_loss: 0.1585 - head_7_accuracy: 0.9414 - head_7_loss: 0.1592 - loss: 1.1113 - val_head_1_accuracy: 0.9607 - val_head_1_loss: 0.1267 - val_head_2_accuracy: 0.9613 - val_head_2_loss: 0.1288 - val_head_3_accuracy: 0.9599 - val_head_3_loss: 0.1268 - val_head_4_accuracy: 0.9611 - val_head_4_loss: 0.1260 - val_head_5_accuracy: 0.9610 - val_head_5_loss: 0.1265 - val_head_6_accuracy: 0.9605 - val_head_6_loss: 0.1286 - val_head_7_accuracy: 0.9620 - val_head_7_loss: 0.1256 - val_loss: 0.8920 - learning_rate: 5.9039e-04\n","\n","Epoch 10: LearningRateScheduler setting learning rate to 0.0005878499174356684.\n","Epoch 10/100\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 89ms/step - head_1_accuracy: 0.9522 - head_1_loss: 0.1322 - head_2_accuracy: 0.9529 - head_2_loss: 0.1314 - head_3_accuracy: 0.9527 - head_3_loss: 0.1319 - head_4_accuracy: 0.9517 - head_4_loss: 0.1312 - head_5_accuracy: 0.9526 - head_5_loss: 0.1309 - head_6_accuracy: 0.9530 - head_6_loss: 0.1307 - head_7_accuracy: 0.9533 - head_7_loss: 0.1314 - loss: 0.9198\n","Epoch 10: val_head_1_accuracy improved from 0.96133 to 0.97937, saving model to /kaggle/working/results/models/GFNet_MultiHead_best.keras\n","\n","Epoch 10: finished saving model to /kaggle/working/results/models/GFNet_MultiHead_best.keras\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m32s\u001b[0m 100ms/step - head_1_accuracy: 0.9478 - head_1_loss: 0.1430 - head_2_accuracy: 0.9490 - head_2_loss: 0.1429 - head_3_accuracy: 0.9486 - head_3_loss: 0.1431 - head_4_accuracy: 0.9478 - head_4_loss: 0.1424 - head_5_accuracy: 0.9484 - head_5_loss: 0.1425 - head_6_accuracy: 0.9486 - head_6_loss: 0.1425 - head_7_accuracy: 0.9496 - head_7_loss: 0.1422 - loss: 0.9991 - val_head_1_accuracy: 0.9794 - val_head_1_loss: 0.0600 - val_head_2_accuracy: 0.9788 - val_head_2_loss: 0.0604 - val_head_3_accuracy: 0.9795 - val_head_3_loss: 0.0608 - val_head_4_accuracy: 0.9792 - val_head_4_loss: 0.0599 - val_head_5_accuracy: 0.9789 - val_head_5_loss: 0.0601 - val_head_6_accuracy: 0.9790 - val_head_6_loss: 0.0604 - val_head_7_accuracy: 0.9787 - val_head_7_loss: 0.0602 - val_loss: 0.4204 - learning_rate: 5.8785e-04\n","\n","Epoch 11: LearningRateScheduler setting learning rate to 0.0005850238317663965.\n","Epoch 11/100\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 89ms/step - head_1_accuracy: 0.9549 - head_1_loss: 0.1252 - head_2_accuracy: 0.9544 - head_2_loss: 0.1258 - head_3_accuracy: 0.9542 - head_3_loss: 0.1254 - head_4_accuracy: 0.9538 - head_4_loss: 0.1260 - head_5_accuracy: 0.9543 - head_5_loss: 0.1252 - head_6_accuracy: 0.9551 - head_6_loss: 0.1250 - head_7_accuracy: 0.9542 - head_7_loss: 0.1248 - loss: 0.8775\n","Epoch 11: val_head_1_accuracy did not improve from 0.97937\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m30s\u001b[0m 95ms/step - head_1_accuracy: 0.9547 - head_1_loss: 0.1228 - head_2_accuracy: 0.9546 - head_2_loss: 0.1230 - head_3_accuracy: 0.9547 - head_3_loss: 0.1234 - head_4_accuracy: 0.9544 - head_4_loss: 0.1231 - head_5_accuracy: 0.9547 - head_5_loss: 0.1226 - head_6_accuracy: 0.9547 - head_6_loss: 0.1228 - head_7_accuracy: 0.9546 - head_7_loss: 0.1229 - loss: 0.8609 - val_head_1_accuracy: 0.9737 - val_head_1_loss: 0.0712 - val_head_2_accuracy: 0.9731 - val_head_2_loss: 0.0741 - val_head_3_accuracy: 0.9732 - val_head_3_loss: 0.0733 - val_head_4_accuracy: 0.9738 - val_head_4_loss: 0.0711 - val_head_5_accuracy: 0.9737 - val_head_5_loss: 0.0710 - val_head_6_accuracy: 0.9739 - val_head_6_loss: 0.0723 - val_head_7_accuracy: 0.9737 - val_head_7_loss: 0.0714 - val_loss: 0.5061 - learning_rate: 5.8502e-04\n","\n","Epoch 12: LearningRateScheduler setting learning rate to 0.0005819108016047332.\n","Epoch 12/100\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 89ms/step - head_1_accuracy: 0.9575 - head_1_loss: 0.1192 - head_2_accuracy: 0.9575 - head_2_loss: 0.1201 - head_3_accuracy: 0.9573 - head_3_loss: 0.1196 - head_4_accuracy: 0.9561 - head_4_loss: 0.1197 - head_5_accuracy: 0.9565 - head_5_loss: 0.1200 - head_6_accuracy: 0.9570 - head_6_loss: 0.1191 - head_7_accuracy: 0.9569 - head_7_loss: 0.1194 - loss: 0.8370\n","Epoch 12: val_head_1_accuracy improved from 0.97937 to 0.98100, saving model to /kaggle/working/results/models/GFNet_MultiHead_best.keras\n","\n","Epoch 12: finished saving model to /kaggle/working/results/models/GFNet_MultiHead_best.keras\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m32s\u001b[0m 100ms/step - head_1_accuracy: 0.9582 - head_1_loss: 0.1145 - head_2_accuracy: 0.9584 - head_2_loss: 0.1144 - head_3_accuracy: 0.9587 - head_3_loss: 0.1145 - head_4_accuracy: 0.9577 - head_4_loss: 0.1150 - head_5_accuracy: 0.9582 - head_5_loss: 0.1150 - head_6_accuracy: 0.9580 - head_6_loss: 0.1146 - head_7_accuracy: 0.9580 - head_7_loss: 0.1142 - loss: 0.8026 - val_head_1_accuracy: 0.9810 - val_head_1_loss: 0.0556 - val_head_2_accuracy: 0.9800 - val_head_2_loss: 0.0567 - val_head_3_accuracy: 0.9813 - val_head_3_loss: 0.0561 - val_head_4_accuracy: 0.9803 - val_head_4_loss: 0.0560 - val_head_5_accuracy: 0.9795 - val_head_5_loss: 0.0571 - val_head_6_accuracy: 0.9807 - val_head_6_loss: 0.0561 - val_head_7_accuracy: 0.9810 - val_head_7_loss: 0.0558 - val_loss: 0.3916 - learning_rate: 5.8191e-04\n","\n","Epoch 13: LearningRateScheduler setting learning rate to 0.0005785139615081709.\n","Epoch 13/100\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 89ms/step - head_1_accuracy: 0.9651 - head_1_loss: 0.0979 - head_2_accuracy: 0.9655 - head_2_loss: 0.0983 - head_3_accuracy: 0.9645 - head_3_loss: 0.0982 - head_4_accuracy: 0.9656 - head_4_loss: 0.0978 - head_5_accuracy: 0.9648 - head_5_loss: 0.0982 - head_6_accuracy: 0.9646 - head_6_loss: 0.0981 - head_7_accuracy: 0.9649 - head_7_loss: 0.0977 - loss: 0.6860\n","Epoch 13: val_head_1_accuracy did not improve from 0.98100\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m30s\u001b[0m 95ms/step - head_1_accuracy: 0.9631 - head_1_loss: 0.1024 - head_2_accuracy: 0.9631 - head_2_loss: 0.1023 - head_3_accuracy: 0.9625 - head_3_loss: 0.1027 - head_4_accuracy: 0.9633 - head_4_loss: 0.1023 - head_5_accuracy: 0.9629 - head_5_loss: 0.1024 - head_6_accuracy: 0.9630 - head_6_loss: 0.1023 - head_7_accuracy: 0.9632 - head_7_loss: 0.1022 - loss: 0.7166 - val_head_1_accuracy: 0.9768 - val_head_1_loss: 0.0719 - val_head_2_accuracy: 0.9770 - val_head_2_loss: 0.0723 - val_head_3_accuracy: 0.9767 - val_head_3_loss: 0.0715 - val_head_4_accuracy: 0.9764 - val_head_4_loss: 0.0732 - val_head_5_accuracy: 0.9770 - val_head_5_loss: 0.0716 - val_head_6_accuracy: 0.9767 - val_head_6_loss: 0.0725 - val_head_7_accuracy: 0.9765 - val_head_7_loss: 0.0730 - val_loss: 0.5069 - learning_rate: 5.7851e-04\n","\n","Epoch 14: LearningRateScheduler setting learning rate to 0.0005748367318067492.\n","Epoch 14/100\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 89ms/step - head_1_accuracy: 0.9666 - head_1_loss: 0.0914 - head_2_accuracy: 0.9671 - head_2_loss: 0.0918 - head_3_accuracy: 0.9672 - head_3_loss: 0.0915 - head_4_accuracy: 0.9680 - head_4_loss: 0.0917 - head_5_accuracy: 0.9679 - head_5_loss: 0.0907 - head_6_accuracy: 0.9672 - head_6_loss: 0.0911 - head_7_accuracy: 0.9675 - head_7_loss: 0.0911 - loss: 0.6394\n","Epoch 14: val_head_1_accuracy did not improve from 0.98100\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m30s\u001b[0m 95ms/step - head_1_accuracy: 0.9697 - head_1_loss: 0.0844 - head_2_accuracy: 0.9696 - head_2_loss: 0.0844 - head_3_accuracy: 0.9700 - head_3_loss: 0.0840 - head_4_accuracy: 0.9705 - head_4_loss: 0.0841 - head_5_accuracy: 0.9706 - head_5_loss: 0.0841 - head_6_accuracy: 0.9703 - head_6_loss: 0.0839 - head_7_accuracy: 0.9706 - head_7_loss: 0.0841 - loss: 0.5894 - val_head_1_accuracy: 0.9754 - val_head_1_loss: 0.0722 - val_head_2_accuracy: 0.9752 - val_head_2_loss: 0.0728 - val_head_3_accuracy: 0.9752 - val_head_3_loss: 0.0730 - val_head_4_accuracy: 0.9751 - val_head_4_loss: 0.0730 - val_head_5_accuracy: 0.9754 - val_head_5_loss: 0.0727 - val_head_6_accuracy: 0.9751 - val_head_6_loss: 0.0725 - val_head_7_accuracy: 0.9757 - val_head_7_loss: 0.0730 - val_loss: 0.5123 - learning_rate: 5.7484e-04\n","\n","Epoch 15: LearningRateScheduler setting learning rate to 0.0005708828151590721.\n","Epoch 15/100\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 89ms/step - head_1_accuracy: 0.9707 - head_1_loss: 0.0802 - head_2_accuracy: 0.9715 - head_2_loss: 0.0799 - head_3_accuracy: 0.9714 - head_3_loss: 0.0800 - head_4_accuracy: 0.9714 - head_4_loss: 0.0798 - head_5_accuracy: 0.9709 - head_5_loss: 0.0800 - head_6_accuracy: 0.9714 - head_6_loss: 0.0796 - head_7_accuracy: 0.9709 - head_7_loss: 0.0800 - loss: 0.5595\n","Epoch 15: val_head_1_accuracy did not improve from 0.98100\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m30s\u001b[0m 95ms/step - head_1_accuracy: 0.9702 - head_1_loss: 0.0826 - head_2_accuracy: 0.9705 - head_2_loss: 0.0825 - head_3_accuracy: 0.9701 - head_3_loss: 0.0828 - head_4_accuracy: 0.9708 - head_4_loss: 0.0821 - head_5_accuracy: 0.9704 - head_5_loss: 0.0824 - head_6_accuracy: 0.9707 - head_6_loss: 0.0821 - head_7_accuracy: 0.9702 - head_7_loss: 0.0822 - loss: 0.5768 - val_head_1_accuracy: 0.9742 - val_head_1_loss: 0.0811 - val_head_2_accuracy: 0.9743 - val_head_2_loss: 0.0818 - val_head_3_accuracy: 0.9744 - val_head_3_loss: 0.0821 - val_head_4_accuracy: 0.9747 - val_head_4_loss: 0.0821 - val_head_5_accuracy: 0.9748 - val_head_5_loss: 0.0795 - val_head_6_accuracy: 0.9745 - val_head_6_loss: 0.0816 - val_head_7_accuracy: 0.9748 - val_head_7_loss: 0.0809 - val_loss: 0.5728 - learning_rate: 5.7088e-04\n","\n","Epoch 16: LearningRateScheduler setting learning rate to 0.0005666561928240443.\n","Epoch 16/100\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 89ms/step - head_1_accuracy: 0.9691 - head_1_loss: 0.0860 - head_2_accuracy: 0.9693 - head_2_loss: 0.0853 - head_3_accuracy: 0.9691 - head_3_loss: 0.0862 - head_4_accuracy: 0.9695 - head_4_loss: 0.0860 - head_5_accuracy: 0.9684 - head_5_loss: 0.0855 - head_6_accuracy: 0.9688 - head_6_loss: 0.0864 - head_7_accuracy: 0.9683 - head_7_loss: 0.0863 - loss: 0.6016\n","Epoch 16: val_head_1_accuracy improved from 0.98100 to 0.98672, saving model to /kaggle/working/results/models/GFNet_MultiHead_best.keras\n","\n","Epoch 16: finished saving model to /kaggle/working/results/models/GFNet_MultiHead_best.keras\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m32s\u001b[0m 100ms/step - head_1_accuracy: 0.9715 - head_1_loss: 0.0808 - head_2_accuracy: 0.9719 - head_2_loss: 0.0804 - head_3_accuracy: 0.9716 - head_3_loss: 0.0807 - head_4_accuracy: 0.9719 - head_4_loss: 0.0807 - head_5_accuracy: 0.9712 - head_5_loss: 0.0807 - head_6_accuracy: 0.9714 - head_6_loss: 0.0810 - head_7_accuracy: 0.9711 - head_7_loss: 0.0807 - loss: 0.5653 - val_head_1_accuracy: 0.9867 - val_head_1_loss: 0.0440 - val_head_2_accuracy: 0.9866 - val_head_2_loss: 0.0435 - val_head_3_accuracy: 0.9869 - val_head_3_loss: 0.0436 - val_head_4_accuracy: 0.9870 - val_head_4_loss: 0.0436 - val_head_5_accuracy: 0.9871 - val_head_5_loss: 0.0439 - val_head_6_accuracy: 0.9868 - val_head_6_loss: 0.0434 - val_head_7_accuracy: 0.9867 - val_head_7_loss: 0.0432 - val_loss: 0.3049 - learning_rate: 5.6666e-04\n","\n","Epoch 17: LearningRateScheduler setting learning rate to 0.000562161120652082.\n","Epoch 17/100\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 89ms/step - head_1_accuracy: 0.9779 - head_1_loss: 0.0672 - head_2_accuracy: 0.9771 - head_2_loss: 0.0674 - head_3_accuracy: 0.9772 - head_3_loss: 0.0674 - head_4_accuracy: 0.9772 - head_4_loss: 0.0675 - head_5_accuracy: 0.9775 - head_5_loss: 0.0670 - head_6_accuracy: 0.9769 - head_6_loss: 0.0674 - head_7_accuracy: 0.9771 - head_7_loss: 0.0676 - loss: 0.4713\n","Epoch 17: val_head_1_accuracy improved from 0.98672 to 0.99117, saving model to /kaggle/working/results/models/GFNet_MultiHead_best.keras\n","\n","Epoch 17: finished saving model to /kaggle/working/results/models/GFNet_MultiHead_best.keras\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m32s\u001b[0m 101ms/step - head_1_accuracy: 0.9727 - head_1_loss: 0.0802 - head_2_accuracy: 0.9722 - head_2_loss: 0.0807 - head_3_accuracy: 0.9727 - head_3_loss: 0.0805 - head_4_accuracy: 0.9722 - head_4_loss: 0.0807 - head_5_accuracy: 0.9725 - head_5_loss: 0.0799 - head_6_accuracy: 0.9720 - head_6_loss: 0.0804 - head_7_accuracy: 0.9724 - head_7_loss: 0.0807 - loss: 0.5635 - val_head_1_accuracy: 0.9912 - val_head_1_loss: 0.0259 - val_head_2_accuracy: 0.9909 - val_head_2_loss: 0.0265 - val_head_3_accuracy: 0.9909 - val_head_3_loss: 0.0265 - val_head_4_accuracy: 0.9909 - val_head_4_loss: 0.0262 - val_head_5_accuracy: 0.9908 - val_head_5_loss: 0.0263 - val_head_6_accuracy: 0.9911 - val_head_6_loss: 0.0261 - val_head_7_accuracy: 0.9911 - val_head_7_loss: 0.0261 - val_loss: 0.1846 - learning_rate: 5.6216e-04\n","\n","Epoch 18: LearningRateScheduler setting learning rate to 0.0005574021247998313.\n","Epoch 18/100\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 89ms/step - head_1_accuracy: 0.9809 - head_1_loss: 0.0569 - head_2_accuracy: 0.9812 - head_2_loss: 0.0571 - head_3_accuracy: 0.9809 - head_3_loss: 0.0572 - head_4_accuracy: 0.9811 - head_4_loss: 0.0568 - head_5_accuracy: 0.9808 - head_5_loss: 0.0571 - head_6_accuracy: 0.9809 - head_6_loss: 0.0571 - head_7_accuracy: 0.9804 - head_7_loss: 0.0571 - loss: 0.3993\n","Epoch 18: val_head_1_accuracy improved from 0.99117 to 0.99295, saving model to /kaggle/working/results/models/GFNet_MultiHead_best.keras\n","\n","Epoch 18: finished saving model to /kaggle/working/results/models/GFNet_MultiHead_best.keras\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m32s\u001b[0m 100ms/step - head_1_accuracy: 0.9795 - head_1_loss: 0.0600 - head_2_accuracy: 0.9796 - head_2_loss: 0.0601 - head_3_accuracy: 0.9798 - head_3_loss: 0.0600 - head_4_accuracy: 0.9798 - head_4_loss: 0.0599 - head_5_accuracy: 0.9796 - head_5_loss: 0.0599 - head_6_accuracy: 0.9795 - head_6_loss: 0.0602 - head_7_accuracy: 0.9794 - head_7_loss: 0.0599 - loss: 0.4200 - val_head_1_accuracy: 0.9929 - val_head_1_loss: 0.0225 - val_head_2_accuracy: 0.9928 - val_head_2_loss: 0.0228 - val_head_3_accuracy: 0.9927 - val_head_3_loss: 0.0231 - val_head_4_accuracy: 0.9930 - val_head_4_loss: 0.0227 - val_head_5_accuracy: 0.9932 - val_head_5_loss: 0.0223 - val_head_6_accuracy: 0.9930 - val_head_6_loss: 0.0227 - val_head_7_accuracy: 0.9929 - val_head_7_loss: 0.0223 - val_loss: 0.1595 - learning_rate: 5.5740e-04\n","\n","Epoch 19: LearningRateScheduler setting learning rate to 0.0005523839971727127.\n","Epoch 19/100\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 89ms/step - head_1_accuracy: 0.9765 - head_1_loss: 0.0681 - head_2_accuracy: 0.9773 - head_2_loss: 0.0680 - head_3_accuracy: 0.9770 - head_3_loss: 0.0683 - head_4_accuracy: 0.9766 - head_4_loss: 0.0682 - head_5_accuracy: 0.9767 - head_5_loss: 0.0682 - head_6_accuracy: 0.9768 - head_6_loss: 0.0682 - head_7_accuracy: 0.9768 - head_7_loss: 0.0682 - loss: 0.4772\n","Epoch 19: val_head_1_accuracy did not improve from 0.99295\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m30s\u001b[0m 95ms/step - head_1_accuracy: 0.9766 - head_1_loss: 0.0662 - head_2_accuracy: 0.9769 - head_2_loss: 0.0661 - head_3_accuracy: 0.9768 - head_3_loss: 0.0665 - head_4_accuracy: 0.9765 - head_4_loss: 0.0664 - head_5_accuracy: 0.9764 - head_5_loss: 0.0663 - head_6_accuracy: 0.9767 - head_6_loss: 0.0665 - head_7_accuracy: 0.9768 - head_7_loss: 0.0662 - loss: 0.4646 - val_head_1_accuracy: 0.9921 - val_head_1_loss: 0.0242 - val_head_2_accuracy: 0.9918 - val_head_2_loss: 0.0249 - val_head_3_accuracy: 0.9921 - val_head_3_loss: 0.0252 - val_head_4_accuracy: 0.9924 - val_head_4_loss: 0.0242 - val_head_5_accuracy: 0.9921 - val_head_5_loss: 0.0246 - val_head_6_accuracy: 0.9922 - val_head_6_loss: 0.0249 - val_head_7_accuracy: 0.9921 - val_head_7_loss: 0.0243 - val_loss: 0.1735 - learning_rate: 5.5238e-04\n","\n","Epoch 20: LearningRateScheduler setting learning rate to 0.0005471117905998783.\n","Epoch 20/100\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 89ms/step - head_1_accuracy: 0.9784 - head_1_loss: 0.0632 - head_2_accuracy: 0.9777 - head_2_loss: 0.0635 - head_3_accuracy: 0.9781 - head_3_loss: 0.0634 - head_4_accuracy: 0.9777 - head_4_loss: 0.0633 - head_5_accuracy: 0.9776 - head_5_loss: 0.0636 - head_6_accuracy: 0.9785 - head_6_loss: 0.0631 - head_7_accuracy: 0.9773 - head_7_loss: 0.0636 - loss: 0.4437\n","Epoch 20: val_head_1_accuracy did not improve from 0.99295\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m30s\u001b[0m 96ms/step - head_1_accuracy: 0.9769 - head_1_loss: 0.0662 - head_2_accuracy: 0.9766 - head_2_loss: 0.0661 - head_3_accuracy: 0.9767 - head_3_loss: 0.0662 - head_4_accuracy: 0.9765 - head_4_loss: 0.0661 - head_5_accuracy: 0.9761 - head_5_loss: 0.0663 - head_6_accuracy: 0.9769 - head_6_loss: 0.0661 - head_7_accuracy: 0.9762 - head_7_loss: 0.0664 - loss: 0.4635 - val_head_1_accuracy: 0.9683 - val_head_1_loss: 0.0969 - val_head_2_accuracy: 0.9685 - val_head_2_loss: 0.0981 - val_head_3_accuracy: 0.9683 - val_head_3_loss: 0.0982 - val_head_4_accuracy: 0.9687 - val_head_4_loss: 0.0972 - val_head_5_accuracy: 0.9685 - val_head_5_loss: 0.0992 - val_head_6_accuracy: 0.9684 - val_head_6_loss: 0.0973 - val_head_7_accuracy: 0.9682 - val_head_7_loss: 0.0985 - val_loss: 0.6887 - learning_rate: 5.4711e-04\n","\n","Epoch 21: LearningRateScheduler setting learning rate to 0.000541590813746441.\n","Epoch 21/100\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 89ms/step - head_1_accuracy: 0.9747 - head_1_loss: 0.0704 - head_2_accuracy: 0.9746 - head_2_loss: 0.0698 - head_3_accuracy: 0.9744 - head_3_loss: 0.0708 - head_4_accuracy: 0.9743 - head_4_loss: 0.0704 - head_5_accuracy: 0.9751 - head_5_loss: 0.0701 - head_6_accuracy: 0.9743 - head_6_loss: 0.0700 - head_7_accuracy: 0.9741 - head_7_loss: 0.0702 - loss: 0.4916\n","Epoch 21: val_head_1_accuracy did not improve from 0.99295\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m30s\u001b[0m 95ms/step - head_1_accuracy: 0.9759 - head_1_loss: 0.0679 - head_2_accuracy: 0.9757 - head_2_loss: 0.0676 - head_3_accuracy: 0.9758 - head_3_loss: 0.0681 - head_4_accuracy: 0.9757 - head_4_loss: 0.0681 - head_5_accuracy: 0.9763 - head_5_loss: 0.0679 - head_6_accuracy: 0.9758 - head_6_loss: 0.0676 - head_7_accuracy: 0.9756 - head_7_loss: 0.0676 - loss: 0.4750 - val_head_1_accuracy: 0.9904 - val_head_1_loss: 0.0280 - val_head_2_accuracy: 0.9905 - val_head_2_loss: 0.0280 - val_head_3_accuracy: 0.9904 - val_head_3_loss: 0.0280 - val_head_4_accuracy: 0.9905 - val_head_4_loss: 0.0278 - val_head_5_accuracy: 0.9906 - val_head_5_loss: 0.0277 - val_head_6_accuracy: 0.9904 - val_head_6_loss: 0.0280 - val_head_7_accuracy: 0.9904 - val_head_7_loss: 0.0279 - val_loss: 0.1967 - learning_rate: 5.4159e-04\n","\n","Epoch 22: LearningRateScheduler setting learning rate to 0.0005358266257680991.\n","Epoch 22/100\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 89ms/step - head_1_accuracy: 0.9814 - head_1_loss: 0.0515 - head_2_accuracy: 0.9815 - head_2_loss: 0.0519 - head_3_accuracy: 0.9812 - head_3_loss: 0.0513 - head_4_accuracy: 0.9815 - head_4_loss: 0.0515 - head_5_accuracy: 0.9810 - head_5_loss: 0.0518 - head_6_accuracy: 0.9816 - head_6_loss: 0.0516 - head_7_accuracy: 0.9816 - head_7_loss: 0.0522 - loss: 0.3617\n","Epoch 22: val_head_1_accuracy did not improve from 0.99295\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m30s\u001b[0m 95ms/step - head_1_accuracy: 0.9823 - head_1_loss: 0.0509 - head_2_accuracy: 0.9822 - head_2_loss: 0.0510 - head_3_accuracy: 0.9823 - head_3_loss: 0.0508 - head_4_accuracy: 0.9821 - head_4_loss: 0.0510 - head_5_accuracy: 0.9821 - head_5_loss: 0.0510 - head_6_accuracy: 0.9824 - head_6_loss: 0.0507 - head_7_accuracy: 0.9821 - head_7_loss: 0.0515 - loss: 0.3563 - val_head_1_accuracy: 0.9555 - val_head_1_loss: 0.1375 - val_head_2_accuracy: 0.9551 - val_head_2_loss: 0.1378 - val_head_3_accuracy: 0.9561 - val_head_3_loss: 0.1346 - val_head_4_accuracy: 0.9563 - val_head_4_loss: 0.1341 - val_head_5_accuracy: 0.9552 - val_head_5_loss: 0.1377 - val_head_6_accuracy: 0.9550 - val_head_6_loss: 0.1375 - val_head_7_accuracy: 0.9547 - val_head_7_loss: 0.1389 - val_loss: 0.9618 - learning_rate: 5.3583e-04\n","\n","Epoch 23: LearningRateScheduler setting learning rate to 0.0005298250307135374.\n","Epoch 23/100\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 89ms/step - head_1_accuracy: 0.9771 - head_1_loss: 0.0642 - head_2_accuracy: 0.9766 - head_2_loss: 0.0638 - head_3_accuracy: 0.9765 - head_3_loss: 0.0647 - head_4_accuracy: 0.9770 - head_4_loss: 0.0637 - head_5_accuracy: 0.9762 - head_5_loss: 0.0642 - head_6_accuracy: 0.9763 - head_6_loss: 0.0647 - head_7_accuracy: 0.9763 - head_7_loss: 0.0641 - loss: 0.4493\n","Epoch 23: val_head_1_accuracy improved from 0.99295 to 0.99347, saving model to /kaggle/working/results/models/GFNet_MultiHead_best.keras\n","\n","Epoch 23: finished saving model to /kaggle/working/results/models/GFNet_MultiHead_best.keras\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m32s\u001b[0m 100ms/step - head_1_accuracy: 0.9827 - head_1_loss: 0.0501 - head_2_accuracy: 0.9827 - head_2_loss: 0.0497 - head_3_accuracy: 0.9826 - head_3_loss: 0.0502 - head_4_accuracy: 0.9825 - head_4_loss: 0.0499 - head_5_accuracy: 0.9825 - head_5_loss: 0.0502 - head_6_accuracy: 0.9825 - head_6_loss: 0.0502 - head_7_accuracy: 0.9824 - head_7_loss: 0.0500 - loss: 0.3503 - val_head_1_accuracy: 0.9935 - val_head_1_loss: 0.0187 - val_head_2_accuracy: 0.9935 - val_head_2_loss: 0.0188 - val_head_3_accuracy: 0.9933 - val_head_3_loss: 0.0188 - val_head_4_accuracy: 0.9934 - val_head_4_loss: 0.0187 - val_head_5_accuracy: 0.9935 - val_head_5_loss: 0.0186 - val_head_6_accuracy: 0.9935 - val_head_6_loss: 0.0188 - val_head_7_accuracy: 0.9934 - val_head_7_loss: 0.0189 - val_loss: 0.1321 - learning_rate: 5.2983e-04\n","\n","Epoch 24: LearningRateScheduler setting learning rate to 0.0005235920716802426.\n","Epoch 24/100\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 89ms/step - head_1_accuracy: 0.9806 - head_1_loss: 0.0582 - head_2_accuracy: 0.9806 - head_2_loss: 0.0578 - head_3_accuracy: 0.9805 - head_3_loss: 0.0582 - head_4_accuracy: 0.9808 - head_4_loss: 0.0576 - head_5_accuracy: 0.9811 - head_5_loss: 0.0571 - head_6_accuracy: 0.9811 - head_6_loss: 0.0577 - head_7_accuracy: 0.9806 - head_7_loss: 0.0579 - loss: 0.4046\n","Epoch 24: val_head_1_accuracy improved from 0.99347 to 0.99473, saving model to /kaggle/working/results/models/GFNet_MultiHead_best.keras\n","\n","Epoch 24: finished saving model to /kaggle/working/results/models/GFNet_MultiHead_best.keras\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m32s\u001b[0m 101ms/step - head_1_accuracy: 0.9822 - head_1_loss: 0.0530 - head_2_accuracy: 0.9823 - head_2_loss: 0.0526 - head_3_accuracy: 0.9822 - head_3_loss: 0.0528 - head_4_accuracy: 0.9823 - head_4_loss: 0.0525 - head_5_accuracy: 0.9825 - head_5_loss: 0.0523 - head_6_accuracy: 0.9823 - head_6_loss: 0.0524 - head_7_accuracy: 0.9824 - head_7_loss: 0.0527 - loss: 0.3684 - val_head_1_accuracy: 0.9947 - val_head_1_loss: 0.0146 - val_head_2_accuracy: 0.9948 - val_head_2_loss: 0.0148 - val_head_3_accuracy: 0.9948 - val_head_3_loss: 0.0144 - val_head_4_accuracy: 0.9948 - val_head_4_loss: 0.0142 - val_head_5_accuracy: 0.9947 - val_head_5_loss: 0.0145 - val_head_6_accuracy: 0.9948 - val_head_6_loss: 0.0145 - val_head_7_accuracy: 0.9950 - val_head_7_loss: 0.0145 - val_loss: 0.1023 - learning_rate: 5.2359e-04\n","\n","Epoch 25: LearningRateScheduler setting learning rate to 0.0005171340247296157.\n","Epoch 25/100\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 89ms/step - head_1_accuracy: 0.9862 - head_1_loss: 0.0401 - head_2_accuracy: 0.9862 - head_2_loss: 0.0404 - head_3_accuracy: 0.9862 - head_3_loss: 0.0404 - head_4_accuracy: 0.9870 - head_4_loss: 0.0403 - head_5_accuracy: 0.9864 - head_5_loss: 0.0403 - head_6_accuracy: 0.9866 - head_6_loss: 0.0402 - head_7_accuracy: 0.9868 - head_7_loss: 0.0403 - loss: 0.2818\n","Epoch 25: val_head_1_accuracy did not improve from 0.99473\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m30s\u001b[0m 95ms/step - head_1_accuracy: 0.9775 - head_1_loss: 0.0697 - head_2_accuracy: 0.9772 - head_2_loss: 0.0696 - head_3_accuracy: 0.9775 - head_3_loss: 0.0699 - head_4_accuracy: 0.9780 - head_4_loss: 0.0696 - head_5_accuracy: 0.9775 - head_5_loss: 0.0701 - head_6_accuracy: 0.9777 - head_6_loss: 0.0696 - head_7_accuracy: 0.9776 - head_7_loss: 0.0695 - loss: 0.4883 - val_head_1_accuracy: 0.9947 - val_head_1_loss: 0.0157 - val_head_2_accuracy: 0.9947 - val_head_2_loss: 0.0155 - val_head_3_accuracy: 0.9941 - val_head_3_loss: 0.0162 - val_head_4_accuracy: 0.9947 - val_head_4_loss: 0.0159 - val_head_5_accuracy: 0.9946 - val_head_5_loss: 0.0155 - val_head_6_accuracy: 0.9947 - val_head_6_loss: 0.0158 - val_head_7_accuracy: 0.9946 - val_head_7_loss: 0.0157 - val_loss: 0.1111 - learning_rate: 5.1713e-04\n","\n","Epoch 26: LearningRateScheduler setting learning rate to 0.000510457392567511.\n","Epoch 26/100\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 89ms/step - head_1_accuracy: 0.9894 - head_1_loss: 0.0305 - head_2_accuracy: 0.9892 - head_2_loss: 0.0308 - head_3_accuracy: 0.9895 - head_3_loss: 0.0308 - head_4_accuracy: 0.9895 - head_4_loss: 0.0305 - head_5_accuracy: 0.9893 - head_5_loss: 0.0309 - head_6_accuracy: 0.9897 - head_6_loss: 0.0309 - head_7_accuracy: 0.9894 - head_7_loss: 0.0306 - loss: 0.2150\n","Epoch 26: val_head_1_accuracy improved from 0.99473 to 0.99510, saving model to /kaggle/working/results/models/GFNet_MultiHead_best.keras\n","\n","Epoch 26: finished saving model to /kaggle/working/results/models/GFNet_MultiHead_best.keras\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m32s\u001b[0m 100ms/step - head_1_accuracy: 0.9885 - head_1_loss: 0.0330 - head_2_accuracy: 0.9884 - head_2_loss: 0.0332 - head_3_accuracy: 0.9886 - head_3_loss: 0.0332 - head_4_accuracy: 0.9885 - head_4_loss: 0.0331 - head_5_accuracy: 0.9881 - head_5_loss: 0.0335 - head_6_accuracy: 0.9887 - head_6_loss: 0.0334 - head_7_accuracy: 0.9886 - head_7_loss: 0.0331 - loss: 0.2325 - val_head_1_accuracy: 0.9951 - val_head_1_loss: 0.0135 - val_head_2_accuracy: 0.9952 - val_head_2_loss: 0.0133 - val_head_3_accuracy: 0.9952 - val_head_3_loss: 0.0133 - val_head_4_accuracy: 0.9952 - val_head_4_loss: 0.0131 - val_head_5_accuracy: 0.9952 - val_head_5_loss: 0.0130 - val_head_6_accuracy: 0.9953 - val_head_6_loss: 0.0129 - val_head_7_accuracy: 0.9950 - val_head_7_loss: 0.0134 - val_loss: 0.0930 - learning_rate: 5.1046e-04\n","\n","Epoch 27: LearningRateScheduler setting learning rate to 0.0005035688979965618.\n","Epoch 27/100\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 89ms/step - head_1_accuracy: 0.9915 - head_1_loss: 0.0250 - head_2_accuracy: 0.9913 - head_2_loss: 0.0249 - head_3_accuracy: 0.9917 - head_3_loss: 0.0247 - head_4_accuracy: 0.9913 - head_4_loss: 0.0252 - head_5_accuracy: 0.9915 - head_5_loss: 0.0249 - head_6_accuracy: 0.9912 - head_6_loss: 0.0253 - head_7_accuracy: 0.9910 - head_7_loss: 0.0252 - loss: 0.1751\n","Epoch 27: val_head_1_accuracy did not improve from 0.99510\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m30s\u001b[0m 96ms/step - head_1_accuracy: 0.9892 - head_1_loss: 0.0317 - head_2_accuracy: 0.9893 - head_2_loss: 0.0317 - head_3_accuracy: 0.9894 - head_3_loss: 0.0317 - head_4_accuracy: 0.9890 - head_4_loss: 0.0319 - head_5_accuracy: 0.9895 - head_5_loss: 0.0317 - head_6_accuracy: 0.9893 - head_6_loss: 0.0318 - head_7_accuracy: 0.9890 - head_7_loss: 0.0317 - loss: 0.2223 - val_head_1_accuracy: 0.9773 - val_head_1_loss: 0.0761 - val_head_2_accuracy: 0.9774 - val_head_2_loss: 0.0743 - val_head_3_accuracy: 0.9772 - val_head_3_loss: 0.0752 - val_head_4_accuracy: 0.9777 - val_head_4_loss: 0.0739 - val_head_5_accuracy: 0.9776 - val_head_5_loss: 0.0736 - val_head_6_accuracy: 0.9774 - val_head_6_loss: 0.0739 - val_head_7_accuracy: 0.9776 - val_head_7_loss: 0.0744 - val_loss: 0.5244 - learning_rate: 5.0357e-04\n","\n","Epoch 28: LearningRateScheduler setting learning rate to 0.0004964754771468882.\n","Epoch 28/100\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 89ms/step - head_1_accuracy: 0.9808 - head_1_loss: 0.0615 - head_2_accuracy: 0.9804 - head_2_loss: 0.0616 - head_3_accuracy: 0.9805 - head_3_loss: 0.0613 - head_4_accuracy: 0.9803 - head_4_loss: 0.0615 - head_5_accuracy: 0.9805 - head_5_loss: 0.0614 - head_6_accuracy: 0.9807 - head_6_loss: 0.0614 - head_7_accuracy: 0.9803 - head_7_loss: 0.0618 - loss: 0.4305\n","Epoch 28: val_head_1_accuracy improved from 0.99510 to 0.99904, saving model to /kaggle/working/results/models/GFNet_MultiHead_best.keras\n","\n","Epoch 28: finished saving model to /kaggle/working/results/models/GFNet_MultiHead_best.keras\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m32s\u001b[0m 100ms/step - head_1_accuracy: 0.9822 - head_1_loss: 0.0545 - head_2_accuracy: 0.9817 - head_2_loss: 0.0546 - head_3_accuracy: 0.9816 - head_3_loss: 0.0544 - head_4_accuracy: 0.9815 - head_4_loss: 0.0546 - head_5_accuracy: 0.9817 - head_5_loss: 0.0542 - head_6_accuracy: 0.9820 - head_6_loss: 0.0544 - head_7_accuracy: 0.9819 - head_7_loss: 0.0546 - loss: 0.3816 - val_head_1_accuracy: 0.9990 - val_head_1_loss: 0.0034 - val_head_2_accuracy: 0.9990 - val_head_2_loss: 0.0035 - val_head_3_accuracy: 0.9989 - val_head_3_loss: 0.0034 - val_head_4_accuracy: 0.9990 - val_head_4_loss: 0.0033 - val_head_5_accuracy: 0.9990 - val_head_5_loss: 0.0034 - val_head_6_accuracy: 0.9990 - val_head_6_loss: 0.0034 - val_head_7_accuracy: 0.9990 - val_head_7_loss: 0.0034 - val_loss: 0.0240 - learning_rate: 4.9648e-04\n","\n","Epoch 29: LearningRateScheduler setting learning rate to 0.0004891842724920025.\n","Epoch 29/100\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 89ms/step - head_1_accuracy: 0.9925 - head_1_loss: 0.0219 - head_2_accuracy: 0.9924 - head_2_loss: 0.0218 - head_3_accuracy: 0.9927 - head_3_loss: 0.0216 - head_4_accuracy: 0.9922 - head_4_loss: 0.0218 - head_5_accuracy: 0.9924 - head_5_loss: 0.0219 - head_6_accuracy: 0.9925 - head_6_loss: 0.0218 - head_7_accuracy: 0.9924 - head_7_loss: 0.0220 - loss: 0.1528\n","Epoch 29: val_head_1_accuracy did not improve from 0.99904\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m30s\u001b[0m 95ms/step - head_1_accuracy: 0.9921 - head_1_loss: 0.0232 - head_2_accuracy: 0.9919 - head_2_loss: 0.0230 - head_3_accuracy: 0.9921 - head_3_loss: 0.0230 - head_4_accuracy: 0.9918 - head_4_loss: 0.0231 - head_5_accuracy: 0.9918 - head_5_loss: 0.0232 - head_6_accuracy: 0.9918 - head_6_loss: 0.0231 - head_7_accuracy: 0.9921 - head_7_loss: 0.0231 - loss: 0.1617 - val_head_1_accuracy: 0.9982 - val_head_1_loss: 0.0045 - val_head_2_accuracy: 0.9981 - val_head_2_loss: 0.0045 - val_head_3_accuracy: 0.9981 - val_head_3_loss: 0.0047 - val_head_4_accuracy: 0.9981 - val_head_4_loss: 0.0046 - val_head_5_accuracy: 0.9982 - val_head_5_loss: 0.0045 - val_head_6_accuracy: 0.9981 - val_head_6_loss: 0.0047 - val_head_7_accuracy: 0.9981 - val_head_7_loss: 0.0045 - val_loss: 0.0323 - learning_rate: 4.8918e-04\n","\n","Epoch 30: LearningRateScheduler setting learning rate to 0.0004817026256569431.\n","Epoch 30/100\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 89ms/step - head_1_accuracy: 0.9921 - head_1_loss: 0.0242 - head_2_accuracy: 0.9920 - head_2_loss: 0.0241 - head_3_accuracy: 0.9918 - head_3_loss: 0.0241 - head_4_accuracy: 0.9920 - head_4_loss: 0.0241 - head_5_accuracy: 0.9918 - head_5_loss: 0.0241 - head_6_accuracy: 0.9918 - head_6_loss: 0.0241 - head_7_accuracy: 0.9919 - head_7_loss: 0.0238 - loss: 0.1683\n","Epoch 30: val_head_1_accuracy did not improve from 0.99904\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m30s\u001b[0m 95ms/step - head_1_accuracy: 0.9906 - head_1_loss: 0.0294 - head_2_accuracy: 0.9903 - head_2_loss: 0.0293 - head_3_accuracy: 0.9905 - head_3_loss: 0.0293 - head_4_accuracy: 0.9904 - head_4_loss: 0.0292 - head_5_accuracy: 0.9903 - head_5_loss: 0.0294 - head_6_accuracy: 0.9903 - head_6_loss: 0.0292 - head_7_accuracy: 0.9904 - head_7_loss: 0.0291 - loss: 0.2050 - val_head_1_accuracy: 0.9981 - val_head_1_loss: 0.0050 - val_head_2_accuracy: 0.9982 - val_head_2_loss: 0.0050 - val_head_3_accuracy: 0.9982 - val_head_3_loss: 0.0050 - val_head_4_accuracy: 0.9981 - val_head_4_loss: 0.0052 - val_head_5_accuracy: 0.9982 - val_head_5_loss: 0.0050 - val_head_6_accuracy: 0.9982 - val_head_6_loss: 0.0049 - val_head_7_accuracy: 0.9982 - val_head_7_loss: 0.0050 - val_loss: 0.0353 - learning_rate: 4.8170e-04\n","\n","Epoch 31: LearningRateScheduler setting learning rate to 0.0004740380700258809.\n","Epoch 31/100\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 89ms/step - head_1_accuracy: 0.9879 - head_1_loss: 0.0359 - head_2_accuracy: 0.9875 - head_2_loss: 0.0362 - head_3_accuracy: 0.9876 - head_3_loss: 0.0363 - head_4_accuracy: 0.9876 - head_4_loss: 0.0361 - head_5_accuracy: 0.9877 - head_5_loss: 0.0361 - head_6_accuracy: 0.9878 - head_6_loss: 0.0361 - head_7_accuracy: 0.9881 - head_7_loss: 0.0361 - loss: 0.2528\n","Epoch 31: val_head_1_accuracy did not improve from 0.99904\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m30s\u001b[0m 95ms/step - head_1_accuracy: 0.9839 - head_1_loss: 0.0483 - head_2_accuracy: 0.9836 - head_2_loss: 0.0484 - head_3_accuracy: 0.9836 - head_3_loss: 0.0485 - head_4_accuracy: 0.9838 - head_4_loss: 0.0484 - head_5_accuracy: 0.9837 - head_5_loss: 0.0484 - head_6_accuracy: 0.9836 - head_6_loss: 0.0485 - head_7_accuracy: 0.9839 - head_7_loss: 0.0483 - loss: 0.3390 - val_head_1_accuracy: 0.9927 - val_head_1_loss: 0.0199 - val_head_2_accuracy: 0.9929 - val_head_2_loss: 0.0195 - val_head_3_accuracy: 0.9929 - val_head_3_loss: 0.0198 - val_head_4_accuracy: 0.9930 - val_head_4_loss: 0.0195 - val_head_5_accuracy: 0.9927 - val_head_5_loss: 0.0200 - val_head_6_accuracy: 0.9929 - val_head_6_loss: 0.0195 - val_head_7_accuracy: 0.9929 - val_head_7_loss: 0.0194 - val_loss: 0.1372 - learning_rate: 4.7404e-04\n","\n","Epoch 32: LearningRateScheduler setting learning rate to 0.0004661983231566398.\n","Epoch 32/100\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 89ms/step - head_1_accuracy: 0.9931 - head_1_loss: 0.0191 - head_2_accuracy: 0.9931 - head_2_loss: 0.0193 - head_3_accuracy: 0.9931 - head_3_loss: 0.0193 - head_4_accuracy: 0.9929 - head_4_loss: 0.0192 - head_5_accuracy: 0.9934 - head_5_loss: 0.0192 - head_6_accuracy: 0.9931 - head_6_loss: 0.0193 - head_7_accuracy: 0.9932 - head_7_loss: 0.0191 - loss: 0.1345\n","Epoch 32: val_head_1_accuracy did not improve from 0.99904\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m30s\u001b[0m 95ms/step - head_1_accuracy: 0.9918 - head_1_loss: 0.0237 - head_2_accuracy: 0.9917 - head_2_loss: 0.0238 - head_3_accuracy: 0.9917 - head_3_loss: 0.0238 - head_4_accuracy: 0.9916 - head_4_loss: 0.0237 - head_5_accuracy: 0.9917 - head_5_loss: 0.0238 - head_6_accuracy: 0.9917 - head_6_loss: 0.0238 - head_7_accuracy: 0.9920 - head_7_loss: 0.0236 - loss: 0.1660 - val_head_1_accuracy: 0.9887 - val_head_1_loss: 0.0351 - val_head_2_accuracy: 0.9884 - val_head_2_loss: 0.0356 - val_head_3_accuracy: 0.9887 - val_head_3_loss: 0.0352 - val_head_4_accuracy: 0.9884 - val_head_4_loss: 0.0351 - val_head_5_accuracy: 0.9885 - val_head_5_loss: 0.0351 - val_head_6_accuracy: 0.9886 - val_head_6_loss: 0.0346 - val_head_7_accuracy: 0.9886 - val_head_7_loss: 0.0355 - val_loss: 0.2479 - learning_rate: 4.6620e-04\n","\n","Epoch 33: LearningRateScheduler setting learning rate to 0.0004581912790097702.\n","Epoch 33/100\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 89ms/step - head_1_accuracy: 0.9872 - head_1_loss: 0.0333 - head_2_accuracy: 0.9872 - head_2_loss: 0.0334 - head_3_accuracy: 0.9869 - head_3_loss: 0.0335 - head_4_accuracy: 0.9873 - head_4_loss: 0.0335 - head_5_accuracy: 0.9870 - head_5_loss: 0.0338 - head_6_accuracy: 0.9871 - head_6_loss: 0.0336 - head_7_accuracy: 0.9872 - head_7_loss: 0.0333 - loss: 0.2345\n","Epoch 33: val_head_1_accuracy did not improve from 0.99904\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m30s\u001b[0m 96ms/step - head_1_accuracy: 0.9884 - head_1_loss: 0.0318 - head_2_accuracy: 0.9886 - head_2_loss: 0.0318 - head_3_accuracy: 0.9883 - head_3_loss: 0.0318 - head_4_accuracy: 0.9886 - head_4_loss: 0.0318 - head_5_accuracy: 0.9885 - head_5_loss: 0.0320 - head_6_accuracy: 0.9883 - head_6_loss: 0.0320 - head_7_accuracy: 0.9884 - head_7_loss: 0.0318 - loss: 0.2232 - val_head_1_accuracy: 0.9987 - val_head_1_loss: 0.0038 - val_head_2_accuracy: 0.9986 - val_head_2_loss: 0.0038 - val_head_3_accuracy: 0.9986 - val_head_3_loss: 0.0038 - val_head_4_accuracy: 0.9985 - val_head_4_loss: 0.0038 - val_head_5_accuracy: 0.9986 - val_head_5_loss: 0.0038 - val_head_6_accuracy: 0.9986 - val_head_6_loss: 0.0039 - val_head_7_accuracy: 0.9986 - val_head_7_loss: 0.0039 - val_loss: 0.0270 - learning_rate: 4.5819e-04\n","\n","Epoch 34: LearningRateScheduler setting learning rate to 0.000450025.\n","Epoch 34/100\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 89ms/step - head_1_accuracy: 0.9871 - head_1_loss: 0.0372 - head_2_accuracy: 0.9872 - head_2_loss: 0.0369 - head_3_accuracy: 0.9872 - head_3_loss: 0.0372 - head_4_accuracy: 0.9871 - head_4_loss: 0.0370 - head_5_accuracy: 0.9872 - head_5_loss: 0.0371 - head_6_accuracy: 0.9873 - head_6_loss: 0.0370 - head_7_accuracy: 0.9873 - head_7_loss: 0.0370 - loss: 0.2592\n","Epoch 34: val_head_1_accuracy did not improve from 0.99904\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m30s\u001b[0m 95ms/step - head_1_accuracy: 0.9884 - head_1_loss: 0.0326 - head_2_accuracy: 0.9884 - head_2_loss: 0.0325 - head_3_accuracy: 0.9886 - head_3_loss: 0.0326 - head_4_accuracy: 0.9884 - head_4_loss: 0.0325 - head_5_accuracy: 0.9885 - head_5_loss: 0.0326 - head_6_accuracy: 0.9886 - head_6_loss: 0.0324 - head_7_accuracy: 0.9885 - head_7_loss: 0.0325 - loss: 0.2278 - val_head_1_accuracy: 0.9932 - val_head_1_loss: 0.0198 - val_head_2_accuracy: 0.9935 - val_head_2_loss: 0.0190 - val_head_3_accuracy: 0.9935 - val_head_3_loss: 0.0189 - val_head_4_accuracy: 0.9935 - val_head_4_loss: 0.0192 - val_head_5_accuracy: 0.9932 - val_head_5_loss: 0.0191 - val_head_6_accuracy: 0.9934 - val_head_6_loss: 0.0190 - val_head_7_accuracy: 0.9934 - val_head_7_loss: 0.0188 - val_loss: 0.1344 - learning_rate: 4.5002e-04\n","\n","Epoch 35: LearningRateScheduler setting learning rate to 0.0004417077088780662.\n","Epoch 35/100\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 89ms/step - head_1_accuracy: 0.9894 - head_1_loss: 0.0321 - head_2_accuracy: 0.9893 - head_2_loss: 0.0319 - head_3_accuracy: 0.9894 - head_3_loss: 0.0320 - head_4_accuracy: 0.9893 - head_4_loss: 0.0319 - head_5_accuracy: 0.9888 - head_5_loss: 0.0319 - head_6_accuracy: 0.9890 - head_6_loss: 0.0319 - head_7_accuracy: 0.9895 - head_7_loss: 0.0318 - loss: 0.2235\n","Epoch 35: val_head_1_accuracy improved from 0.99904 to 0.99948, saving model to /kaggle/working/results/models/GFNet_MultiHead_best.keras\n","\n","Epoch 35: finished saving model to /kaggle/working/results/models/GFNet_MultiHead_best.keras\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m32s\u001b[0m 100ms/step - head_1_accuracy: 0.9904 - head_1_loss: 0.0287 - head_2_accuracy: 0.9900 - head_2_loss: 0.0287 - head_3_accuracy: 0.9903 - head_3_loss: 0.0288 - head_4_accuracy: 0.9901 - head_4_loss: 0.0287 - head_5_accuracy: 0.9898 - head_5_loss: 0.0287 - head_6_accuracy: 0.9899 - head_6_loss: 0.0286 - head_7_accuracy: 0.9903 - head_7_loss: 0.0285 - loss: 0.2009 - val_head_1_accuracy: 0.9995 - val_head_1_loss: 0.0019 - val_head_2_accuracy: 0.9994 - val_head_2_loss: 0.0020 - val_head_3_accuracy: 0.9995 - val_head_3_loss: 0.0020 - val_head_4_accuracy: 0.9995 - val_head_4_loss: 0.0020 - val_head_5_accuracy: 0.9995 - val_head_5_loss: 0.0020 - val_head_6_accuracy: 0.9995 - val_head_6_loss: 0.0019 - val_head_7_accuracy: 0.9995 - val_head_7_loss: 0.0019 - val_loss: 0.0139 - learning_rate: 4.4171e-04\n","\n","Epoch 36: LearningRateScheduler setting learning rate to 0.00043324778045110196.\n","Epoch 36/100\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 89ms/step - head_1_accuracy: 0.9947 - head_1_loss: 0.0159 - head_2_accuracy: 0.9946 - head_2_loss: 0.0163 - head_3_accuracy: 0.9948 - head_3_loss: 0.0162 - head_4_accuracy: 0.9947 - head_4_loss: 0.0160 - head_5_accuracy: 0.9947 - head_5_loss: 0.0160 - head_6_accuracy: 0.9948 - head_6_loss: 0.0161 - head_7_accuracy: 0.9946 - head_7_loss: 0.0161 - loss: 0.1127\n","Epoch 36: val_head_1_accuracy did not improve from 0.99948\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m30s\u001b[0m 95ms/step - head_1_accuracy: 0.9920 - head_1_loss: 0.0242 - head_2_accuracy: 0.9921 - head_2_loss: 0.0244 - head_3_accuracy: 0.9920 - head_3_loss: 0.0242 - head_4_accuracy: 0.9921 - head_4_loss: 0.0242 - head_5_accuracy: 0.9922 - head_5_loss: 0.0242 - head_6_accuracy: 0.9920 - head_6_loss: 0.0244 - head_7_accuracy: 0.9922 - head_7_loss: 0.0243 - loss: 0.1699 - val_head_1_accuracy: 0.9849 - val_head_1_loss: 0.0425 - val_head_2_accuracy: 0.9848 - val_head_2_loss: 0.0434 - val_head_3_accuracy: 0.9847 - val_head_3_loss: 0.0429 - val_head_4_accuracy: 0.9843 - val_head_4_loss: 0.0440 - val_head_5_accuracy: 0.9844 - val_head_5_loss: 0.0440 - val_head_6_accuracy: 0.9851 - val_head_6_loss: 0.0424 - val_head_7_accuracy: 0.9846 - val_head_7_loss: 0.0428 - val_loss: 0.3040 - learning_rate: 4.3325e-04\n","\n","Epoch 37: LearningRateScheduler setting learning rate to 0.0004246537331499159.\n","Epoch 37/100\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 89ms/step - head_1_accuracy: 0.9917 - head_1_loss: 0.0243 - head_2_accuracy: 0.9913 - head_2_loss: 0.0242 - head_3_accuracy: 0.9917 - head_3_loss: 0.0244 - head_4_accuracy: 0.9920 - head_4_loss: 0.0243 - head_5_accuracy: 0.9915 - head_5_loss: 0.0242 - head_6_accuracy: 0.9921 - head_6_loss: 0.0242 - head_7_accuracy: 0.9916 - head_7_loss: 0.0244 - loss: 0.1700\n","Epoch 37: val_head_1_accuracy did not improve from 0.99948\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m30s\u001b[0m 95ms/step - head_1_accuracy: 0.9922 - head_1_loss: 0.0219 - head_2_accuracy: 0.9922 - head_2_loss: 0.0217 - head_3_accuracy: 0.9922 - head_3_loss: 0.0219 - head_4_accuracy: 0.9924 - head_4_loss: 0.0219 - head_5_accuracy: 0.9922 - head_5_loss: 0.0217 - head_6_accuracy: 0.9925 - head_6_loss: 0.0217 - head_7_accuracy: 0.9925 - head_7_loss: 0.0218 - loss: 0.1525 - val_head_1_accuracy: 0.9981 - val_head_1_loss: 0.0058 - val_head_2_accuracy: 0.9981 - val_head_2_loss: 0.0059 - val_head_3_accuracy: 0.9981 - val_head_3_loss: 0.0058 - val_head_4_accuracy: 0.9981 - val_head_4_loss: 0.0059 - val_head_5_accuracy: 0.9981 - val_head_5_loss: 0.0058 - val_head_6_accuracy: 0.9981 - val_head_6_loss: 0.0059 - val_head_7_accuracy: 0.9981 - val_head_7_loss: 0.0060 - val_loss: 0.0413 - learning_rate: 4.2465e-04\n","\n","Epoch 38: LearningRateScheduler setting learning rate to 0.0004159342204516539.\n","Epoch 38/100\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 89ms/step - head_1_accuracy: 0.9938 - head_1_loss: 0.0184 - head_2_accuracy: 0.9937 - head_2_loss: 0.0182 - head_3_accuracy: 0.9938 - head_3_loss: 0.0182 - head_4_accuracy: 0.9938 - head_4_loss: 0.0182 - head_5_accuracy: 0.9942 - head_5_loss: 0.0183 - head_6_accuracy: 0.9941 - head_6_loss: 0.0182 - head_7_accuracy: 0.9941 - head_7_loss: 0.0182 - loss: 0.1278\n","Epoch 38: val_head_1_accuracy did not improve from 0.99948\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m30s\u001b[0m 95ms/step - head_1_accuracy: 0.9946 - head_1_loss: 0.0154 - head_2_accuracy: 0.9946 - head_2_loss: 0.0154 - head_3_accuracy: 0.9946 - head_3_loss: 0.0154 - head_4_accuracy: 0.9946 - head_4_loss: 0.0154 - head_5_accuracy: 0.9948 - head_5_loss: 0.0154 - head_6_accuracy: 0.9949 - head_6_loss: 0.0154 - head_7_accuracy: 0.9948 - head_7_loss: 0.0153 - loss: 0.1076 - val_head_1_accuracy: 0.9993 - val_head_1_loss: 0.0022 - val_head_2_accuracy: 0.9993 - val_head_2_loss: 0.0021 - val_head_3_accuracy: 0.9993 - val_head_3_loss: 0.0021 - val_head_4_accuracy: 0.9993 - val_head_4_loss: 0.0021 - val_head_5_accuracy: 0.9993 - val_head_5_loss: 0.0022 - val_head_6_accuracy: 0.9993 - val_head_6_loss: 0.0021 - val_head_7_accuracy: 0.9993 - val_head_7_loss: 0.0021 - val_loss: 0.0150 - learning_rate: 4.1593e-04\n","\n","Epoch 39: LearningRateScheduler setting learning rate to 0.0004070980221664819.\n","Epoch 39/100\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 89ms/step - head_1_accuracy: 0.9920 - head_1_loss: 0.0233 - head_2_accuracy: 0.9916 - head_2_loss: 0.0235 - head_3_accuracy: 0.9918 - head_3_loss: 0.0235 - head_4_accuracy: 0.9918 - head_4_loss: 0.0235 - head_5_accuracy: 0.9919 - head_5_loss: 0.0233 - head_6_accuracy: 0.9919 - head_6_loss: 0.0234 - head_7_accuracy: 0.9916 - head_7_loss: 0.0235 - loss: 0.1640\n","Epoch 39: val_head_1_accuracy did not improve from 0.99948\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m30s\u001b[0m 95ms/step - head_1_accuracy: 0.9911 - head_1_loss: 0.0273 - head_2_accuracy: 0.9907 - head_2_loss: 0.0273 - head_3_accuracy: 0.9908 - head_3_loss: 0.0273 - head_4_accuracy: 0.9909 - head_4_loss: 0.0273 - head_5_accuracy: 0.9907 - head_5_loss: 0.0271 - head_6_accuracy: 0.9908 - head_6_loss: 0.0273 - head_7_accuracy: 0.9906 - head_7_loss: 0.0274 - loss: 0.1910 - val_head_1_accuracy: 0.9968 - val_head_1_loss: 0.0093 - val_head_2_accuracy: 0.9967 - val_head_2_loss: 0.0091 - val_head_3_accuracy: 0.9968 - val_head_3_loss: 0.0091 - val_head_4_accuracy: 0.9968 - val_head_4_loss: 0.0090 - val_head_5_accuracy: 0.9968 - val_head_5_loss: 0.0092 - val_head_6_accuracy: 0.9968 - val_head_6_loss: 0.0091 - val_head_7_accuracy: 0.9967 - val_head_7_loss: 0.0091 - val_loss: 0.0642 - learning_rate: 4.0710e-04\n","\n","Epoch 40: LearningRateScheduler setting learning rate to 0.0003981540355970607.\n","Epoch 40/100\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 89ms/step - head_1_accuracy: 0.9933 - head_1_loss: 0.0197 - head_2_accuracy: 0.9935 - head_2_loss: 0.0195 - head_3_accuracy: 0.9932 - head_3_loss: 0.0199 - head_4_accuracy: 0.9935 - head_4_loss: 0.0194 - head_5_accuracy: 0.9934 - head_5_loss: 0.0196 - head_6_accuracy: 0.9932 - head_6_loss: 0.0196 - head_7_accuracy: 0.9934 - head_7_loss: 0.0197 - loss: 0.1373\n","Epoch 40: val_head_1_accuracy did not improve from 0.99948\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m30s\u001b[0m 95ms/step - head_1_accuracy: 0.9940 - head_1_loss: 0.0168 - head_2_accuracy: 0.9941 - head_2_loss: 0.0166 - head_3_accuracy: 0.9940 - head_3_loss: 0.0169 - head_4_accuracy: 0.9943 - head_4_loss: 0.0167 - head_5_accuracy: 0.9941 - head_5_loss: 0.0167 - head_6_accuracy: 0.9940 - head_6_loss: 0.0169 - head_7_accuracy: 0.9942 - head_7_loss: 0.0167 - loss: 0.1172 - val_head_1_accuracy: 0.9990 - val_head_1_loss: 0.0021 - val_head_2_accuracy: 0.9990 - val_head_2_loss: 0.0020 - val_head_3_accuracy: 0.9991 - val_head_3_loss: 0.0021 - val_head_4_accuracy: 0.9990 - val_head_4_loss: 0.0021 - val_head_5_accuracy: 0.9990 - val_head_5_loss: 0.0021 - val_head_6_accuracy: 0.9991 - val_head_6_loss: 0.0021 - val_head_7_accuracy: 0.9990 - val_head_7_loss: 0.0020 - val_loss: 0.0146 - learning_rate: 3.9815e-04\n","\n","Epoch 41: LearningRateScheduler setting learning rate to 0.0003891112665797161.\n","Epoch 41/100\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 89ms/step - head_1_accuracy: 0.9904 - head_1_loss: 0.0299 - head_2_accuracy: 0.9905 - head_2_loss: 0.0297 - head_3_accuracy: 0.9907 - head_3_loss: 0.0295 - head_4_accuracy: 0.9905 - head_4_loss: 0.0299 - head_5_accuracy: 0.9902 - head_5_loss: 0.0297 - head_6_accuracy: 0.9905 - head_6_loss: 0.0296 - head_7_accuracy: 0.9901 - head_7_loss: 0.0295 - loss: 0.2079\n","Epoch 41: val_head_1_accuracy did not improve from 0.99948\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m30s\u001b[0m 95ms/step - head_1_accuracy: 0.9921 - head_1_loss: 0.0246 - head_2_accuracy: 0.9922 - head_2_loss: 0.0245 - head_3_accuracy: 0.9923 - head_3_loss: 0.0244 - head_4_accuracy: 0.9921 - head_4_loss: 0.0248 - head_5_accuracy: 0.9919 - head_5_loss: 0.0246 - head_6_accuracy: 0.9921 - head_6_loss: 0.0244 - head_7_accuracy: 0.9920 - head_7_loss: 0.0245 - loss: 0.1718 - val_head_1_accuracy: 0.9988 - val_head_1_loss: 0.0038 - val_head_2_accuracy: 0.9987 - val_head_2_loss: 0.0037 - val_head_3_accuracy: 0.9987 - val_head_3_loss: 0.0036 - val_head_4_accuracy: 0.9987 - val_head_4_loss: 0.0037 - val_head_5_accuracy: 0.9986 - val_head_5_loss: 0.0039 - val_head_6_accuracy: 0.9987 - val_head_6_loss: 0.0038 - val_head_7_accuracy: 0.9984 - val_head_7_loss: 0.0039 - val_loss: 0.0267 - learning_rate: 3.8911e-04\n","\n","Epoch 42: LearningRateScheduler setting learning rate to 0.000379978820416326.\n","Epoch 42/100\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 89ms/step - head_1_accuracy: 0.9950 - head_1_loss: 0.0145 - head_2_accuracy: 0.9950 - head_2_loss: 0.0146 - head_3_accuracy: 0.9947 - head_3_loss: 0.0145 - head_4_accuracy: 0.9950 - head_4_loss: 0.0145 - head_5_accuracy: 0.9950 - head_5_loss: 0.0145 - head_6_accuracy: 0.9947 - head_6_loss: 0.0145 - head_7_accuracy: 0.9950 - head_7_loss: 0.0145 - loss: 0.1017\n","Epoch 42: val_head_1_accuracy did not improve from 0.99948\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m30s\u001b[0m 95ms/step - head_1_accuracy: 0.9939 - head_1_loss: 0.0183 - head_2_accuracy: 0.9940 - head_2_loss: 0.0183 - head_3_accuracy: 0.9938 - head_3_loss: 0.0183 - head_4_accuracy: 0.9937 - head_4_loss: 0.0182 - head_5_accuracy: 0.9938 - head_5_loss: 0.0182 - head_6_accuracy: 0.9936 - head_6_loss: 0.0183 - head_7_accuracy: 0.9939 - head_7_loss: 0.0183 - loss: 0.1275 - val_head_1_accuracy: 0.9975 - val_head_1_loss: 0.0084 - val_head_2_accuracy: 0.9974 - val_head_2_loss: 0.0085 - val_head_3_accuracy: 0.9974 - val_head_3_loss: 0.0086 - val_head_4_accuracy: 0.9976 - val_head_4_loss: 0.0084 - val_head_5_accuracy: 0.9976 - val_head_5_loss: 0.0082 - val_head_6_accuracy: 0.9976 - val_head_6_loss: 0.0083 - val_head_7_accuracy: 0.9977 - val_head_7_loss: 0.0083 - val_loss: 0.0593 - learning_rate: 3.7998e-04\n","\n","Epoch 43: LearningRateScheduler setting learning rate to 0.00037076589270605275.\n","Epoch 43/100\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 89ms/step - head_1_accuracy: 0.9913 - head_1_loss: 0.0249 - head_2_accuracy: 0.9914 - head_2_loss: 0.0250 - head_3_accuracy: 0.9916 - head_3_loss: 0.0248 - head_4_accuracy: 0.9915 - head_4_loss: 0.0252 - head_5_accuracy: 0.9916 - head_5_loss: 0.0250 - head_6_accuracy: 0.9917 - head_6_loss: 0.0251 - head_7_accuracy: 0.9916 - head_7_loss: 0.0249 - loss: 0.1750\n","Epoch 43: val_head_1_accuracy improved from 0.99948 to 0.99955, saving model to /kaggle/working/results/models/GFNet_MultiHead_best.keras\n","\n","Epoch 43: finished saving model to /kaggle/working/results/models/GFNet_MultiHead_best.keras\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m32s\u001b[0m 100ms/step - head_1_accuracy: 0.9950 - head_1_loss: 0.0153 - head_2_accuracy: 0.9951 - head_2_loss: 0.0153 - head_3_accuracy: 0.9951 - head_3_loss: 0.0153 - head_4_accuracy: 0.9950 - head_4_loss: 0.0154 - head_5_accuracy: 0.9950 - head_5_loss: 0.0154 - head_6_accuracy: 0.9952 - head_6_loss: 0.0154 - head_7_accuracy: 0.9952 - head_7_loss: 0.0153 - loss: 0.1074 - val_head_1_accuracy: 0.9996 - val_head_1_loss: 0.0019 - val_head_2_accuracy: 0.9996 - val_head_2_loss: 0.0019 - val_head_3_accuracy: 0.9996 - val_head_3_loss: 0.0019 - val_head_4_accuracy: 0.9996 - val_head_4_loss: 0.0019 - val_head_5_accuracy: 0.9996 - val_head_5_loss: 0.0018 - val_head_6_accuracy: 0.9996 - val_head_6_loss: 0.0019 - val_head_7_accuracy: 0.9996 - val_head_7_loss: 0.0018 - val_loss: 0.0132 - learning_rate: 3.7077e-04\n","\n","Epoch 44: LearningRateScheduler setting learning rate to 0.000361481760086154.\n","Epoch 44/100\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 89ms/step - head_1_accuracy: 0.9928 - head_1_loss: 0.0222 - head_2_accuracy: 0.9927 - head_2_loss: 0.0221 - head_3_accuracy: 0.9930 - head_3_loss: 0.0220 - head_4_accuracy: 0.9930 - head_4_loss: 0.0224 - head_5_accuracy: 0.9928 - head_5_loss: 0.0222 - head_6_accuracy: 0.9927 - head_6_loss: 0.0221 - head_7_accuracy: 0.9930 - head_7_loss: 0.0222 - loss: 0.1552\n","Epoch 44: val_head_1_accuracy did not improve from 0.99955\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m30s\u001b[0m 95ms/step - head_1_accuracy: 0.9952 - head_1_loss: 0.0150 - head_2_accuracy: 0.9952 - head_2_loss: 0.0149 - head_3_accuracy: 0.9953 - head_3_loss: 0.0148 - head_4_accuracy: 0.9954 - head_4_loss: 0.0150 - head_5_accuracy: 0.9952 - head_5_loss: 0.0149 - head_6_accuracy: 0.9952 - head_6_loss: 0.0150 - head_7_accuracy: 0.9952 - head_7_loss: 0.0150 - loss: 0.1047 - val_head_1_accuracy: 0.9992 - val_head_1_loss: 0.0022 - val_head_2_accuracy: 0.9991 - val_head_2_loss: 0.0023 - val_head_3_accuracy: 0.9992 - val_head_3_loss: 0.0022 - val_head_4_accuracy: 0.9992 - val_head_4_loss: 0.0023 - val_head_5_accuracy: 0.9993 - val_head_5_loss: 0.0022 - val_head_6_accuracy: 0.9992 - val_head_6_loss: 0.0022 - val_head_7_accuracy: 0.9992 - val_head_7_loss: 0.0023 - val_loss: 0.0158 - learning_rate: 3.6148e-04\n","\n","Epoch 45: LearningRateScheduler setting learning rate to 0.0003521357708911958.\n","Epoch 45/100\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 89ms/step - head_1_accuracy: 0.9933 - head_1_loss: 0.0200 - head_2_accuracy: 0.9932 - head_2_loss: 0.0200 - head_3_accuracy: 0.9934 - head_3_loss: 0.0199 - head_4_accuracy: 0.9934 - head_4_loss: 0.0201 - head_5_accuracy: 0.9934 - head_5_loss: 0.0200 - head_6_accuracy: 0.9933 - head_6_loss: 0.0200 - head_7_accuracy: 0.9933 - head_7_loss: 0.0201 - loss: 0.1401\n","Epoch 45: val_head_1_accuracy did not improve from 0.99955\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m30s\u001b[0m 95ms/step - head_1_accuracy: 0.9943 - head_1_loss: 0.0168 - head_2_accuracy: 0.9942 - head_2_loss: 0.0169 - head_3_accuracy: 0.9944 - head_3_loss: 0.0168 - head_4_accuracy: 0.9944 - head_4_loss: 0.0169 - head_5_accuracy: 0.9942 - head_5_loss: 0.0169 - head_6_accuracy: 0.9943 - head_6_loss: 0.0168 - head_7_accuracy: 0.9942 - head_7_loss: 0.0170 - loss: 0.1183 - val_head_1_accuracy: 0.9993 - val_head_1_loss: 0.0019 - val_head_2_accuracy: 0.9993 - val_head_2_loss: 0.0018 - val_head_3_accuracy: 0.9993 - val_head_3_loss: 0.0020 - val_head_4_accuracy: 0.9993 - val_head_4_loss: 0.0019 - val_head_5_accuracy: 0.9993 - val_head_5_loss: 0.0019 - val_head_6_accuracy: 0.9993 - val_head_6_loss: 0.0019 - val_head_7_accuracy: 0.9993 - val_head_7_loss: 0.0019 - val_loss: 0.0133 - learning_rate: 3.5214e-04\n","\n","Epoch 46: LearningRateScheduler setting learning rate to 0.00034273733574007183.\n","Epoch 46/100\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 89ms/step - head_1_accuracy: 0.9978 - head_1_loss: 0.0082 - head_2_accuracy: 0.9978 - head_2_loss: 0.0081 - head_3_accuracy: 0.9977 - head_3_loss: 0.0082 - head_4_accuracy: 0.9978 - head_4_loss: 0.0081 - head_5_accuracy: 0.9978 - head_5_loss: 0.0083 - head_6_accuracy: 0.9977 - head_6_loss: 0.0082 - head_7_accuracy: 0.9977 - head_7_loss: 0.0081 - loss: 0.0572\n","Epoch 46: val_head_1_accuracy did not improve from 0.99955\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m30s\u001b[0m 95ms/step - head_1_accuracy: 0.9966 - head_1_loss: 0.0115 - head_2_accuracy: 0.9965 - head_2_loss: 0.0115 - head_3_accuracy: 0.9964 - head_3_loss: 0.0116 - head_4_accuracy: 0.9964 - head_4_loss: 0.0116 - head_5_accuracy: 0.9965 - head_5_loss: 0.0116 - head_6_accuracy: 0.9963 - head_6_loss: 0.0116 - head_7_accuracy: 0.9964 - head_7_loss: 0.0116 - loss: 0.0810 - val_head_1_accuracy: 0.9967 - val_head_1_loss: 0.0085 - val_head_2_accuracy: 0.9966 - val_head_2_loss: 0.0087 - val_head_3_accuracy: 0.9967 - val_head_3_loss: 0.0084 - val_head_4_accuracy: 0.9967 - val_head_4_loss: 0.0083 - val_head_5_accuracy: 0.9967 - val_head_5_loss: 0.0086 - val_head_6_accuracy: 0.9966 - val_head_6_loss: 0.0086 - val_head_7_accuracy: 0.9967 - val_head_7_loss: 0.0085 - val_loss: 0.0600 - learning_rate: 3.4274e-04\n","\n","Epoch 47: LearningRateScheduler setting learning rate to 0.0003332959180603083.\n","Epoch 47/100\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 89ms/step - head_1_accuracy: 0.9937 - head_1_loss: 0.0186 - head_2_accuracy: 0.9937 - head_2_loss: 0.0187 - head_3_accuracy: 0.9933 - head_3_loss: 0.0185 - head_4_accuracy: 0.9936 - head_4_loss: 0.0185 - head_5_accuracy: 0.9934 - head_5_loss: 0.0186 - head_6_accuracy: 0.9935 - head_6_loss: 0.0185 - head_7_accuracy: 0.9936 - head_7_loss: 0.0185 - loss: 0.1299\n","Epoch 47: val_head_1_accuracy did not improve from 0.99955\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m30s\u001b[0m 95ms/step - head_1_accuracy: 0.9950 - head_1_loss: 0.0143 - head_2_accuracy: 0.9949 - head_2_loss: 0.0144 - head_3_accuracy: 0.9948 - head_3_loss: 0.0144 - head_4_accuracy: 0.9949 - head_4_loss: 0.0143 - head_5_accuracy: 0.9948 - head_5_loss: 0.0144 - head_6_accuracy: 0.9948 - head_6_loss: 0.0144 - head_7_accuracy: 0.9949 - head_7_loss: 0.0143 - loss: 0.1007 - val_head_1_accuracy: 0.9992 - val_head_1_loss: 0.0021 - val_head_2_accuracy: 0.9992 - val_head_2_loss: 0.0021 - val_head_3_accuracy: 0.9992 - val_head_3_loss: 0.0022 - val_head_4_accuracy: 0.9992 - val_head_4_loss: 0.0021 - val_head_5_accuracy: 0.9993 - val_head_5_loss: 0.0021 - val_head_6_accuracy: 0.9992 - val_head_6_loss: 0.0022 - val_head_7_accuracy: 0.9992 - val_head_7_loss: 0.0021 - val_loss: 0.0150 - learning_rate: 3.3330e-04\n","\n","Epoch 48: LearningRateScheduler setting learning rate to 0.00032382102455919367.\n","Epoch 48/100\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 89ms/step - head_1_accuracy: 0.9970 - head_1_loss: 0.0096 - head_2_accuracy: 0.9971 - head_2_loss: 0.0095 - head_3_accuracy: 0.9970 - head_3_loss: 0.0098 - head_4_accuracy: 0.9971 - head_4_loss: 0.0096 - head_5_accuracy: 0.9970 - head_5_loss: 0.0096 - head_6_accuracy: 0.9970 - head_6_loss: 0.0097 - head_7_accuracy: 0.9971 - head_7_loss: 0.0096 - loss: 0.0674\n","Epoch 48: val_head_1_accuracy did not improve from 0.99955\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m30s\u001b[0m 95ms/step - head_1_accuracy: 0.9949 - head_1_loss: 0.0154 - head_2_accuracy: 0.9951 - head_2_loss: 0.0154 - head_3_accuracy: 0.9948 - head_3_loss: 0.0156 - head_4_accuracy: 0.9950 - head_4_loss: 0.0155 - head_5_accuracy: 0.9949 - head_5_loss: 0.0155 - head_6_accuracy: 0.9949 - head_6_loss: 0.0154 - head_7_accuracy: 0.9949 - head_7_loss: 0.0155 - loss: 0.1080 - val_head_1_accuracy: 0.9996 - val_head_1_loss: 0.0013 - val_head_2_accuracy: 0.9995 - val_head_2_loss: 0.0013 - val_head_3_accuracy: 0.9995 - val_head_3_loss: 0.0014 - val_head_4_accuracy: 0.9994 - val_head_4_loss: 0.0013 - val_head_5_accuracy: 0.9995 - val_head_5_loss: 0.0013 - val_head_6_accuracy: 0.9994 - val_head_6_loss: 0.0013 - val_head_7_accuracy: 0.9994 - val_head_7_loss: 0.0013 - val_loss: 0.0094 - learning_rate: 3.2382e-04\n","\n","Epoch 49: LearningRateScheduler setting learning rate to 0.0003143221956513316.\n","Epoch 49/100\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 89ms/step - head_1_accuracy: 0.9966 - head_1_loss: 0.0101 - head_2_accuracy: 0.9967 - head_2_loss: 0.0099 - head_3_accuracy: 0.9966 - head_3_loss: 0.0102 - head_4_accuracy: 0.9966 - head_4_loss: 0.0100 - head_5_accuracy: 0.9967 - head_5_loss: 0.0101 - head_6_accuracy: 0.9968 - head_6_loss: 0.0100 - head_7_accuracy: 0.9968 - head_7_loss: 0.0100 - loss: 0.0702\n","Epoch 49: val_head_1_accuracy did not improve from 0.99955\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m30s\u001b[0m 95ms/step - head_1_accuracy: 0.9962 - head_1_loss: 0.0116 - head_2_accuracy: 0.9963 - head_2_loss: 0.0116 - head_3_accuracy: 0.9961 - head_3_loss: 0.0118 - head_4_accuracy: 0.9962 - head_4_loss: 0.0117 - head_5_accuracy: 0.9962 - head_5_loss: 0.0118 - head_6_accuracy: 0.9963 - head_6_loss: 0.0117 - head_7_accuracy: 0.9963 - head_7_loss: 0.0117 - loss: 0.0818 - val_head_1_accuracy: 0.9996 - val_head_1_loss: 0.0012 - val_head_2_accuracy: 0.9996 - val_head_2_loss: 0.0012 - val_head_3_accuracy: 0.9996 - val_head_3_loss: 0.0012 - val_head_4_accuracy: 0.9997 - val_head_4_loss: 0.0012 - val_head_5_accuracy: 0.9996 - val_head_5_loss: 0.0012 - val_head_6_accuracy: 0.9997 - val_head_6_loss: 0.0012 - val_head_7_accuracy: 0.9997 - val_head_7_loss: 0.0012 - val_loss: 0.0085 - learning_rate: 3.1432e-04\n","\n","Epoch 50: LearningRateScheduler setting learning rate to 0.0003048089958522507.\n","Epoch 50/100\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 89ms/step - head_1_accuracy: 0.9959 - head_1_loss: 0.0126 - head_2_accuracy: 0.9961 - head_2_loss: 0.0125 - head_3_accuracy: 0.9961 - head_3_loss: 0.0125 - head_4_accuracy: 0.9958 - head_4_loss: 0.0125 - head_5_accuracy: 0.9960 - head_5_loss: 0.0125 - head_6_accuracy: 0.9959 - head_6_loss: 0.0124 - head_7_accuracy: 0.9959 - head_7_loss: 0.0126 - loss: 0.0877\n","Epoch 50: val_head_1_accuracy improved from 0.99955 to 0.99970, saving model to /kaggle/working/results/models/GFNet_MultiHead_best.keras\n","\n","Epoch 50: finished saving model to /kaggle/working/results/models/GFNet_MultiHead_best.keras\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m32s\u001b[0m 100ms/step - head_1_accuracy: 0.9940 - head_1_loss: 0.0198 - head_2_accuracy: 0.9941 - head_2_loss: 0.0197 - head_3_accuracy: 0.9941 - head_3_loss: 0.0198 - head_4_accuracy: 0.9940 - head_4_loss: 0.0198 - head_5_accuracy: 0.9941 - head_5_loss: 0.0198 - head_6_accuracy: 0.9940 - head_6_loss: 0.0197 - head_7_accuracy: 0.9939 - head_7_loss: 0.0198 - loss: 0.1387 - val_head_1_accuracy: 0.9997 - val_head_1_loss: 0.0012 - val_head_2_accuracy: 0.9996 - val_head_2_loss: 0.0013 - val_head_3_accuracy: 0.9996 - val_head_3_loss: 0.0013 - val_head_4_accuracy: 0.9997 - val_head_4_loss: 0.0013 - val_head_5_accuracy: 0.9996 - val_head_5_loss: 0.0013 - val_head_6_accuracy: 0.9996 - val_head_6_loss: 0.0013 - val_head_7_accuracy: 0.9997 - val_head_7_loss: 0.0013 - val_loss: 0.0090 - learning_rate: 3.0481e-04\n","\n","Epoch 51: LearningRateScheduler setting learning rate to 0.00029529100414774935.\n","Epoch 51/100\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 89ms/step - head_1_accuracy: 0.9992 - head_1_loss: 0.0036 - head_2_accuracy: 0.9990 - head_2_loss: 0.0037 - head_3_accuracy: 0.9989 - head_3_loss: 0.0037 - head_4_accuracy: 0.9990 - head_4_loss: 0.0037 - head_5_accuracy: 0.9992 - head_5_loss: 0.0037 - head_6_accuracy: 0.9990 - head_6_loss: 0.0037 - head_7_accuracy: 0.9990 - head_7_loss: 0.0038 - loss: 0.0260\n","Epoch 51: val_head_1_accuracy improved from 0.99970 to 0.99993, saving model to /kaggle/working/results/models/GFNet_MultiHead_best.keras\n","\n","Epoch 51: finished saving model to /kaggle/working/results/models/GFNet_MultiHead_best.keras\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m32s\u001b[0m 101ms/step - head_1_accuracy: 0.9991 - head_1_loss: 0.0035 - head_2_accuracy: 0.9990 - head_2_loss: 0.0035 - head_3_accuracy: 0.9990 - head_3_loss: 0.0036 - head_4_accuracy: 0.9991 - head_4_loss: 0.0035 - head_5_accuracy: 0.9991 - head_5_loss: 0.0035 - head_6_accuracy: 0.9991 - head_6_loss: 0.0035 - head_7_accuracy: 0.9990 - head_7_loss: 0.0036 - loss: 0.0248 - val_head_1_accuracy: 0.9999 - val_head_1_loss: 2.9159e-04 - val_head_2_accuracy: 0.9999 - val_head_2_loss: 2.9352e-04 - val_head_3_accuracy: 0.9999 - val_head_3_loss: 2.9223e-04 - val_head_4_accuracy: 0.9999 - val_head_4_loss: 2.7876e-04 - val_head_5_accuracy: 0.9999 - val_head_5_loss: 2.8931e-04 - val_head_6_accuracy: 0.9999 - val_head_6_loss: 3.0661e-04 - val_head_7_accuracy: 0.9999 - val_head_7_loss: 2.9270e-04 - val_loss: 0.0021 - learning_rate: 2.9529e-04\n","\n","Epoch 52: LearningRateScheduler setting learning rate to 0.00028577780434866853.\n","Epoch 52/100\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 89ms/step - head_1_accuracy: 0.9982 - head_1_loss: 0.0063 - head_2_accuracy: 0.9982 - head_2_loss: 0.0063 - head_3_accuracy: 0.9982 - head_3_loss: 0.0062 - head_4_accuracy: 0.9981 - head_4_loss: 0.0063 - head_5_accuracy: 0.9982 - head_5_loss: 0.0062 - head_6_accuracy: 0.9982 - head_6_loss: 0.0062 - head_7_accuracy: 0.9981 - head_7_loss: 0.0062 - loss: 0.0436\n","Epoch 52: val_head_1_accuracy did not improve from 0.99993\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m30s\u001b[0m 95ms/step - head_1_accuracy: 0.9967 - head_1_loss: 0.0109 - head_2_accuracy: 0.9966 - head_2_loss: 0.0110 - head_3_accuracy: 0.9967 - head_3_loss: 0.0109 - head_4_accuracy: 0.9966 - head_4_loss: 0.0110 - head_5_accuracy: 0.9967 - head_5_loss: 0.0108 - head_6_accuracy: 0.9966 - head_6_loss: 0.0110 - head_7_accuracy: 0.9966 - head_7_loss: 0.0109 - loss: 0.0764 - val_head_1_accuracy: 0.9984 - val_head_1_loss: 0.0035 - val_head_2_accuracy: 0.9984 - val_head_2_loss: 0.0034 - val_head_3_accuracy: 0.9983 - val_head_3_loss: 0.0035 - val_head_4_accuracy: 0.9983 - val_head_4_loss: 0.0034 - val_head_5_accuracy: 0.9984 - val_head_5_loss: 0.0034 - val_head_6_accuracy: 0.9984 - val_head_6_loss: 0.0034 - val_head_7_accuracy: 0.9984 - val_head_7_loss: 0.0034 - val_loss: 0.0240 - learning_rate: 2.8578e-04\n","\n","Epoch 53: LearningRateScheduler setting learning rate to 0.0002762789754408063.\n","Epoch 53/100\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 89ms/step - head_1_accuracy: 0.9991 - head_1_loss: 0.0042 - head_2_accuracy: 0.9990 - head_2_loss: 0.0043 - head_3_accuracy: 0.9990 - head_3_loss: 0.0043 - head_4_accuracy: 0.9989 - head_4_loss: 0.0043 - head_5_accuracy: 0.9989 - head_5_loss: 0.0043 - head_6_accuracy: 0.9990 - head_6_loss: 0.0043 - head_7_accuracy: 0.9990 - head_7_loss: 0.0043 - loss: 0.0299\n","Epoch 53: val_head_1_accuracy did not improve from 0.99993\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m30s\u001b[0m 95ms/step - head_1_accuracy: 0.9991 - head_1_loss: 0.0038 - head_2_accuracy: 0.9991 - head_2_loss: 0.0039 - head_3_accuracy: 0.9991 - head_3_loss: 0.0039 - head_4_accuracy: 0.9991 - head_4_loss: 0.0038 - head_5_accuracy: 0.9991 - head_5_loss: 0.0039 - head_6_accuracy: 0.9991 - head_6_loss: 0.0038 - head_7_accuracy: 0.9991 - head_7_loss: 0.0039 - loss: 0.0270 - val_head_1_accuracy: 0.9997 - val_head_1_loss: 6.2127e-04 - val_head_2_accuracy: 0.9998 - val_head_2_loss: 6.4013e-04 - val_head_3_accuracy: 0.9997 - val_head_3_loss: 6.3382e-04 - val_head_4_accuracy: 0.9998 - val_head_4_loss: 6.2714e-04 - val_head_5_accuracy: 0.9998 - val_head_5_loss: 5.9820e-04 - val_head_6_accuracy: 0.9997 - val_head_6_loss: 6.2590e-04 - val_head_7_accuracy: 0.9998 - val_head_7_loss: 5.7239e-04 - val_loss: 0.0043 - learning_rate: 2.7628e-04\n","\n","Epoch 54: LearningRateScheduler setting learning rate to 0.0002668040819396918.\n","Epoch 54/100\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 89ms/step - head_1_accuracy: 0.9982 - head_1_loss: 0.0052 - head_2_accuracy: 0.9983 - head_2_loss: 0.0052 - head_3_accuracy: 0.9982 - head_3_loss: 0.0052 - head_4_accuracy: 0.9983 - head_4_loss: 0.0051 - head_5_accuracy: 0.9982 - head_5_loss: 0.0052 - head_6_accuracy: 0.9984 - head_6_loss: 0.0052 - head_7_accuracy: 0.9984 - head_7_loss: 0.0052 - loss: 0.0363\n","Epoch 54: val_head_1_accuracy did not improve from 0.99993\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m30s\u001b[0m 95ms/step - head_1_accuracy: 0.9979 - head_1_loss: 0.0064 - head_2_accuracy: 0.9979 - head_2_loss: 0.0065 - head_3_accuracy: 0.9978 - head_3_loss: 0.0066 - head_4_accuracy: 0.9979 - head_4_loss: 0.0065 - head_5_accuracy: 0.9979 - head_5_loss: 0.0065 - head_6_accuracy: 0.9979 - head_6_loss: 0.0065 - head_7_accuracy: 0.9979 - head_7_loss: 0.0065 - loss: 0.0456 - val_head_1_accuracy: 0.9986 - val_head_1_loss: 0.0044 - val_head_2_accuracy: 0.9986 - val_head_2_loss: 0.0046 - val_head_3_accuracy: 0.9985 - val_head_3_loss: 0.0045 - val_head_4_accuracy: 0.9986 - val_head_4_loss: 0.0044 - val_head_5_accuracy: 0.9985 - val_head_5_loss: 0.0044 - val_head_6_accuracy: 0.9986 - val_head_6_loss: 0.0045 - val_head_7_accuracy: 0.9986 - val_head_7_loss: 0.0045 - val_loss: 0.0314 - learning_rate: 2.6680e-04\n","\n","Epoch 55: LearningRateScheduler setting learning rate to 0.00025736266425992817.\n","Epoch 55/100\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 89ms/step - head_1_accuracy: 0.9944 - head_1_loss: 0.0173 - head_2_accuracy: 0.9944 - head_2_loss: 0.0172 - head_3_accuracy: 0.9942 - head_3_loss: 0.0171 - head_4_accuracy: 0.9944 - head_4_loss: 0.0172 - head_5_accuracy: 0.9944 - head_5_loss: 0.0173 - head_6_accuracy: 0.9944 - head_6_loss: 0.0172 - head_7_accuracy: 0.9943 - head_7_loss: 0.0173 - loss: 0.1205\n","Epoch 55: val_head_1_accuracy did not improve from 0.99993\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m30s\u001b[0m 95ms/step - head_1_accuracy: 0.9947 - head_1_loss: 0.0163 - head_2_accuracy: 0.9948 - head_2_loss: 0.0162 - head_3_accuracy: 0.9946 - head_3_loss: 0.0162 - head_4_accuracy: 0.9947 - head_4_loss: 0.0162 - head_5_accuracy: 0.9946 - head_5_loss: 0.0162 - head_6_accuracy: 0.9947 - head_6_loss: 0.0161 - head_7_accuracy: 0.9946 - head_7_loss: 0.0162 - loss: 0.1135 - val_head_1_accuracy: 0.9999 - val_head_1_loss: 4.8401e-04 - val_head_2_accuracy: 0.9998 - val_head_2_loss: 4.9212e-04 - val_head_3_accuracy: 0.9998 - val_head_3_loss: 4.8565e-04 - val_head_4_accuracy: 0.9999 - val_head_4_loss: 4.7033e-04 - val_head_5_accuracy: 0.9999 - val_head_5_loss: 4.6836e-04 - val_head_6_accuracy: 0.9998 - val_head_6_loss: 5.0311e-04 - val_head_7_accuracy: 0.9999 - val_head_7_loss: 4.6791e-04 - val_loss: 0.0034 - learning_rate: 2.5736e-04\n","\n","Epoch 56: LearningRateScheduler setting learning rate to 0.00024796422910880423.\n","Epoch 56/100\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 89ms/step - head_1_accuracy: 0.9986 - head_1_loss: 0.0041 - head_2_accuracy: 0.9988 - head_2_loss: 0.0041 - head_3_accuracy: 0.9987 - head_3_loss: 0.0041 - head_4_accuracy: 0.9988 - head_4_loss: 0.0040 - head_5_accuracy: 0.9987 - head_5_loss: 0.0040 - head_6_accuracy: 0.9986 - head_6_loss: 0.0040 - head_7_accuracy: 0.9988 - head_7_loss: 0.0040 - loss: 0.0282\n","Epoch 56: val_head_1_accuracy did not improve from 0.99993\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m30s\u001b[0m 95ms/step - head_1_accuracy: 0.9987 - head_1_loss: 0.0041 - head_2_accuracy: 0.9988 - head_2_loss: 0.0041 - head_3_accuracy: 0.9987 - head_3_loss: 0.0041 - head_4_accuracy: 0.9988 - head_4_loss: 0.0040 - head_5_accuracy: 0.9986 - head_5_loss: 0.0041 - head_6_accuracy: 0.9986 - head_6_loss: 0.0040 - head_7_accuracy: 0.9987 - head_7_loss: 0.0041 - loss: 0.0285 - val_head_1_accuracy: 0.9999 - val_head_1_loss: 2.8732e-04 - val_head_2_accuracy: 0.9999 - val_head_2_loss: 3.0670e-04 - val_head_3_accuracy: 0.9999 - val_head_3_loss: 2.9855e-04 - val_head_4_accuracy: 0.9999 - val_head_4_loss: 2.9195e-04 - val_head_5_accuracy: 0.9999 - val_head_5_loss: 2.7193e-04 - val_head_6_accuracy: 0.9999 - val_head_6_loss: 3.1320e-04 - val_head_7_accuracy: 0.9999 - val_head_7_loss: 2.7449e-04 - val_loss: 0.0021 - learning_rate: 2.4796e-04\n","\n","Epoch 57: LearningRateScheduler setting learning rate to 0.0002386182399138461.\n","Epoch 57/100\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 89ms/step - head_1_accuracy: 0.9978 - head_1_loss: 0.0061 - head_2_accuracy: 0.9978 - head_2_loss: 0.0061 - head_3_accuracy: 0.9978 - head_3_loss: 0.0062 - head_4_accuracy: 0.9978 - head_4_loss: 0.0061 - head_5_accuracy: 0.9978 - head_5_loss: 0.0061 - head_6_accuracy: 0.9978 - head_6_loss: 0.0062 - head_7_accuracy: 0.9978 - head_7_loss: 0.0060 - loss: 0.0429\n","Epoch 57: val_head_1_accuracy did not improve from 0.99993\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m30s\u001b[0m 95ms/step - head_1_accuracy: 0.9961 - head_1_loss: 0.0114 - head_2_accuracy: 0.9960 - head_2_loss: 0.0113 - head_3_accuracy: 0.9961 - head_3_loss: 0.0115 - head_4_accuracy: 0.9961 - head_4_loss: 0.0114 - head_5_accuracy: 0.9961 - head_5_loss: 0.0114 - head_6_accuracy: 0.9961 - head_6_loss: 0.0113 - head_7_accuracy: 0.9961 - head_7_loss: 0.0113 - loss: 0.0797 - val_head_1_accuracy: 0.9998 - val_head_1_loss: 0.0011 - val_head_2_accuracy: 0.9998 - val_head_2_loss: 0.0011 - val_head_3_accuracy: 0.9997 - val_head_3_loss: 0.0011 - val_head_4_accuracy: 0.9998 - val_head_4_loss: 0.0011 - val_head_5_accuracy: 0.9997 - val_head_5_loss: 0.0011 - val_head_6_accuracy: 0.9998 - val_head_6_loss: 0.0011 - val_head_7_accuracy: 0.9998 - val_head_7_loss: 0.0011 - val_loss: 0.0078 - learning_rate: 2.3862e-04\n","\n","Epoch 58: LearningRateScheduler setting learning rate to 0.0002293341072939474.\n","Epoch 58/100\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 89ms/step - head_1_accuracy: 0.9983 - head_1_loss: 0.0047 - head_2_accuracy: 0.9984 - head_2_loss: 0.0046 - head_3_accuracy: 0.9983 - head_3_loss: 0.0048 - head_4_accuracy: 0.9985 - head_4_loss: 0.0047 - head_5_accuracy: 0.9983 - head_5_loss: 0.0047 - head_6_accuracy: 0.9985 - head_6_loss: 0.0046 - head_7_accuracy: 0.9984 - head_7_loss: 0.0046 - loss: 0.0327\n","Epoch 58: val_head_1_accuracy improved from 0.99993 to 1.00000, saving model to /kaggle/working/results/models/GFNet_MultiHead_best.keras\n","\n","Epoch 58: finished saving model to /kaggle/working/results/models/GFNet_MultiHead_best.keras\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m32s\u001b[0m 100ms/step - head_1_accuracy: 0.9986 - head_1_loss: 0.0045 - head_2_accuracy: 0.9986 - head_2_loss: 0.0044 - head_3_accuracy: 0.9985 - head_3_loss: 0.0045 - head_4_accuracy: 0.9986 - head_4_loss: 0.0045 - head_5_accuracy: 0.9985 - head_5_loss: 0.0045 - head_6_accuracy: 0.9986 - head_6_loss: 0.0044 - head_7_accuracy: 0.9986 - head_7_loss: 0.0045 - loss: 0.0313 - val_head_1_accuracy: 1.0000 - val_head_1_loss: 1.4279e-04 - val_head_2_accuracy: 1.0000 - val_head_2_loss: 1.6426e-04 - val_head_3_accuracy: 1.0000 - val_head_3_loss: 1.5358e-04 - val_head_4_accuracy: 1.0000 - val_head_4_loss: 1.4872e-04 - val_head_5_accuracy: 1.0000 - val_head_5_loss: 1.5475e-04 - val_head_6_accuracy: 1.0000 - val_head_6_loss: 1.5808e-04 - val_head_7_accuracy: 1.0000 - val_head_7_loss: 1.4950e-04 - val_loss: 0.0011 - learning_rate: 2.2933e-04\n","\n","Epoch 59: LearningRateScheduler setting learning rate to 0.00022012117958367398.\n","Epoch 59/100\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 89ms/step - head_1_accuracy: 0.9984 - head_1_loss: 0.0046 - head_2_accuracy: 0.9983 - head_2_loss: 0.0047 - head_3_accuracy: 0.9984 - head_3_loss: 0.0047 - head_4_accuracy: 0.9984 - head_4_loss: 0.0047 - head_5_accuracy: 0.9983 - head_5_loss: 0.0047 - head_6_accuracy: 0.9984 - head_6_loss: 0.0047 - head_7_accuracy: 0.9983 - head_7_loss: 0.0047 - loss: 0.0328\n","Epoch 59: val_head_1_accuracy did not improve from 1.00000\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m30s\u001b[0m 95ms/step - head_1_accuracy: 0.9986 - head_1_loss: 0.0041 - head_2_accuracy: 0.9985 - head_2_loss: 0.0041 - head_3_accuracy: 0.9986 - head_3_loss: 0.0041 - head_4_accuracy: 0.9985 - head_4_loss: 0.0041 - head_5_accuracy: 0.9985 - head_5_loss: 0.0042 - head_6_accuracy: 0.9986 - head_6_loss: 0.0041 - head_7_accuracy: 0.9985 - head_7_loss: 0.0042 - loss: 0.0287 - val_head_1_accuracy: 0.9996 - val_head_1_loss: 9.9304e-04 - val_head_2_accuracy: 0.9995 - val_head_2_loss: 0.0010 - val_head_3_accuracy: 0.9995 - val_head_3_loss: 0.0011 - val_head_4_accuracy: 0.9996 - val_head_4_loss: 0.0010 - val_head_5_accuracy: 0.9996 - val_head_5_loss: 0.0010 - val_head_6_accuracy: 0.9995 - val_head_6_loss: 0.0010 - val_head_7_accuracy: 0.9996 - val_head_7_loss: 9.9779e-04 - val_loss: 0.0072 - learning_rate: 2.2012e-04\n","\n","Epoch 60: LearningRateScheduler setting learning rate to 0.00021098873342028398.\n","Epoch 60/100\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 89ms/step - head_1_accuracy: 0.9929 - head_1_loss: 0.0192 - head_2_accuracy: 0.9928 - head_2_loss: 0.0190 - head_3_accuracy: 0.9930 - head_3_loss: 0.0190 - head_4_accuracy: 0.9928 - head_4_loss: 0.0190 - head_5_accuracy: 0.9927 - head_5_loss: 0.0191 - head_6_accuracy: 0.9930 - head_6_loss: 0.0190 - head_7_accuracy: 0.9928 - head_7_loss: 0.0191 - loss: 0.1334\n","Epoch 60: val_head_1_accuracy did not improve from 1.00000\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m30s\u001b[0m 95ms/step - head_1_accuracy: 0.9963 - head_1_loss: 0.0103 - head_2_accuracy: 0.9962 - head_2_loss: 0.0103 - head_3_accuracy: 0.9964 - head_3_loss: 0.0102 - head_4_accuracy: 0.9963 - head_4_loss: 0.0102 - head_5_accuracy: 0.9962 - head_5_loss: 0.0103 - head_6_accuracy: 0.9964 - head_6_loss: 0.0102 - head_7_accuracy: 0.9963 - head_7_loss: 0.0103 - loss: 0.0718 - val_head_1_accuracy: 0.9999 - val_head_1_loss: 2.8481e-04 - val_head_2_accuracy: 0.9999 - val_head_2_loss: 3.0157e-04 - val_head_3_accuracy: 0.9999 - val_head_3_loss: 2.8680e-04 - val_head_4_accuracy: 0.9999 - val_head_4_loss: 2.9380e-04 - val_head_5_accuracy: 0.9999 - val_head_5_loss: 2.8907e-04 - val_head_6_accuracy: 0.9999 - val_head_6_loss: 3.0273e-04 - val_head_7_accuracy: 0.9999 - val_head_7_loss: 2.9364e-04 - val_loss: 0.0021 - learning_rate: 2.1099e-04\n","\n","Epoch 61: LearningRateScheduler setting learning rate to 0.00020194596440293946.\n","Epoch 61/100\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 89ms/step - head_1_accuracy: 0.9991 - head_1_loss: 0.0025 - head_2_accuracy: 0.9991 - head_2_loss: 0.0025 - head_3_accuracy: 0.9992 - head_3_loss: 0.0026 - head_4_accuracy: 0.9991 - head_4_loss: 0.0026 - head_5_accuracy: 0.9991 - head_5_loss: 0.0025 - head_6_accuracy: 0.9991 - head_6_loss: 0.0026 - head_7_accuracy: 0.9991 - head_7_loss: 0.0025 - loss: 0.0178\n","Epoch 61: val_head_1_accuracy did not improve from 1.00000\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m30s\u001b[0m 95ms/step - head_1_accuracy: 0.9992 - head_1_loss: 0.0023 - head_2_accuracy: 0.9992 - head_2_loss: 0.0023 - head_3_accuracy: 0.9992 - head_3_loss: 0.0023 - head_4_accuracy: 0.9992 - head_4_loss: 0.0023 - head_5_accuracy: 0.9993 - head_5_loss: 0.0023 - head_6_accuracy: 0.9992 - head_6_loss: 0.0023 - head_7_accuracy: 0.9992 - head_7_loss: 0.0023 - loss: 0.0162 - val_head_1_accuracy: 1.0000 - val_head_1_loss: 1.3465e-04 - val_head_2_accuracy: 1.0000 - val_head_2_loss: 1.3716e-04 - val_head_3_accuracy: 1.0000 - val_head_3_loss: 1.3639e-04 - val_head_4_accuracy: 1.0000 - val_head_4_loss: 1.3087e-04 - val_head_5_accuracy: 1.0000 - val_head_5_loss: 1.3222e-04 - val_head_6_accuracy: 1.0000 - val_head_6_loss: 1.3553e-04 - val_head_7_accuracy: 1.0000 - val_head_7_loss: 1.3090e-04 - val_loss: 9.4427e-04 - learning_rate: 2.0195e-04\n","\n","Epoch 62: LearningRateScheduler setting learning rate to 0.00019300197783351806.\n","Epoch 62/100\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 89ms/step - head_1_accuracy: 0.9990 - head_1_loss: 0.0031 - head_2_accuracy: 0.9990 - head_2_loss: 0.0031 - head_3_accuracy: 0.9990 - head_3_loss: 0.0031 - head_4_accuracy: 0.9990 - head_4_loss: 0.0031 - head_5_accuracy: 0.9990 - head_5_loss: 0.0031 - head_6_accuracy: 0.9990 - head_6_loss: 0.0031 - head_7_accuracy: 0.9990 - head_7_loss: 0.0030 - loss: 0.0216\n","Epoch 62: val_head_1_accuracy did not improve from 1.00000\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m30s\u001b[0m 95ms/step - head_1_accuracy: 0.9983 - head_1_loss: 0.0049 - head_2_accuracy: 0.9983 - head_2_loss: 0.0049 - head_3_accuracy: 0.9983 - head_3_loss: 0.0049 - head_4_accuracy: 0.9983 - head_4_loss: 0.0050 - head_5_accuracy: 0.9983 - head_5_loss: 0.0049 - head_6_accuracy: 0.9984 - head_6_loss: 0.0049 - head_7_accuracy: 0.9983 - head_7_loss: 0.0049 - loss: 0.0345 - val_head_1_accuracy: 0.9999 - val_head_1_loss: 4.2270e-04 - val_head_2_accuracy: 0.9999 - val_head_2_loss: 4.4156e-04 - val_head_3_accuracy: 0.9999 - val_head_3_loss: 4.2672e-04 - val_head_4_accuracy: 0.9999 - val_head_4_loss: 4.2134e-04 - val_head_5_accuracy: 0.9999 - val_head_5_loss: 4.1411e-04 - val_head_6_accuracy: 0.9999 - val_head_6_loss: 4.2960e-04 - val_head_7_accuracy: 0.9999 - val_head_7_loss: 4.1972e-04 - val_loss: 0.0030 - learning_rate: 1.9300e-04\n","\n","Epoch 63: LearningRateScheduler setting learning rate to 0.00018416577954834612.\n","Epoch 63/100\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 89ms/step - head_1_accuracy: 0.9987 - head_1_loss: 0.0040 - head_2_accuracy: 0.9987 - head_2_loss: 0.0039 - head_3_accuracy: 0.9987 - head_3_loss: 0.0039 - head_4_accuracy: 0.9987 - head_4_loss: 0.0040 - head_5_accuracy: 0.9988 - head_5_loss: 0.0040 - head_6_accuracy: 0.9987 - head_6_loss: 0.0040 - head_7_accuracy: 0.9986 - head_7_loss: 0.0040 - loss: 0.0279\n","Epoch 63: val_head_1_accuracy did not improve from 1.00000\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m30s\u001b[0m 95ms/step - head_1_accuracy: 0.9987 - head_1_loss: 0.0040 - head_2_accuracy: 0.9986 - head_2_loss: 0.0040 - head_3_accuracy: 0.9987 - head_3_loss: 0.0040 - head_4_accuracy: 0.9986 - head_4_loss: 0.0040 - head_5_accuracy: 0.9987 - head_5_loss: 0.0040 - head_6_accuracy: 0.9987 - head_6_loss: 0.0040 - head_7_accuracy: 0.9986 - head_7_loss: 0.0040 - loss: 0.0280 - val_head_1_accuracy: 0.9997 - val_head_1_loss: 0.0012 - val_head_2_accuracy: 0.9997 - val_head_2_loss: 0.0011 - val_head_3_accuracy: 0.9996 - val_head_3_loss: 0.0012 - val_head_4_accuracy: 0.9997 - val_head_4_loss: 0.0012 - val_head_5_accuracy: 0.9997 - val_head_5_loss: 0.0012 - val_head_6_accuracy: 0.9996 - val_head_6_loss: 0.0012 - val_head_7_accuracy: 0.9996 - val_head_7_loss: 0.0012 - val_loss: 0.0083 - learning_rate: 1.8417e-04\n","\n","Epoch 64: LearningRateScheduler setting learning rate to 0.00017544626685008413.\n","Epoch 64/100\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 89ms/step - head_1_accuracy: 0.9972 - head_1_loss: 0.0080 - head_2_accuracy: 0.9973 - head_2_loss: 0.0079 - head_3_accuracy: 0.9972 - head_3_loss: 0.0079 - head_4_accuracy: 0.9973 - head_4_loss: 0.0079 - head_5_accuracy: 0.9973 - head_5_loss: 0.0078 - head_6_accuracy: 0.9972 - head_6_loss: 0.0079 - head_7_accuracy: 0.9973 - head_7_loss: 0.0078 - loss: 0.0551\n","Epoch 64: val_head_1_accuracy did not improve from 1.00000\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m30s\u001b[0m 95ms/step - head_1_accuracy: 0.9965 - head_1_loss: 0.0099 - head_2_accuracy: 0.9966 - head_2_loss: 0.0097 - head_3_accuracy: 0.9966 - head_3_loss: 0.0098 - head_4_accuracy: 0.9966 - head_4_loss: 0.0098 - head_5_accuracy: 0.9967 - head_5_loss: 0.0097 - head_6_accuracy: 0.9966 - head_6_loss: 0.0098 - head_7_accuracy: 0.9966 - head_7_loss: 0.0097 - loss: 0.0685 - val_head_1_accuracy: 0.9999 - val_head_1_loss: 3.0881e-04 - val_head_2_accuracy: 0.9999 - val_head_2_loss: 3.0207e-04 - val_head_3_accuracy: 0.9999 - val_head_3_loss: 2.9677e-04 - val_head_4_accuracy: 0.9999 - val_head_4_loss: 3.1140e-04 - val_head_5_accuracy: 0.9999 - val_head_5_loss: 3.0019e-04 - val_head_6_accuracy: 0.9999 - val_head_6_loss: 2.9896e-04 - val_head_7_accuracy: 0.9999 - val_head_7_loss: 2.9084e-04 - val_loss: 0.0021 - learning_rate: 1.7545e-04\n","\n","Epoch 65: LearningRateScheduler setting learning rate to 0.00016685221954889804.\n","Epoch 65/100\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 89ms/step - head_1_accuracy: 0.9996 - head_1_loss: 0.0015 - head_2_accuracy: 0.9996 - head_2_loss: 0.0016 - head_3_accuracy: 0.9994 - head_3_loss: 0.0016 - head_4_accuracy: 0.9994 - head_4_loss: 0.0016 - head_5_accuracy: 0.9996 - head_5_loss: 0.0016 - head_6_accuracy: 0.9996 - head_6_loss: 0.0016 - head_7_accuracy: 0.9995 - head_7_loss: 0.0016 - loss: 0.0110\n","Epoch 65: val_head_1_accuracy did not improve from 1.00000\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m30s\u001b[0m 95ms/step - head_1_accuracy: 0.9996 - head_1_loss: 0.0014 - head_2_accuracy: 0.9996 - head_2_loss: 0.0014 - head_3_accuracy: 0.9995 - head_3_loss: 0.0014 - head_4_accuracy: 0.9995 - head_4_loss: 0.0014 - head_5_accuracy: 0.9996 - head_5_loss: 0.0014 - head_6_accuracy: 0.9996 - head_6_loss: 0.0014 - head_7_accuracy: 0.9996 - head_7_loss: 0.0014 - loss: 0.0099 - val_head_1_accuracy: 0.9999 - val_head_1_loss: 2.8063e-04 - val_head_2_accuracy: 0.9999 - val_head_2_loss: 2.9404e-04 - val_head_3_accuracy: 0.9999 - val_head_3_loss: 2.8335e-04 - val_head_4_accuracy: 0.9999 - val_head_4_loss: 2.9537e-04 - val_head_5_accuracy: 0.9999 - val_head_5_loss: 2.8202e-04 - val_head_6_accuracy: 0.9999 - val_head_6_loss: 2.8710e-04 - val_head_7_accuracy: 0.9999 - val_head_7_loss: 2.8274e-04 - val_loss: 0.0020 - learning_rate: 1.6685e-04\n","\n","Epoch 66: LearningRateScheduler setting learning rate to 0.00015839229112193386.\n","Epoch 66/100\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 89ms/step - head_1_accuracy: 0.9997 - head_1_loss: 0.0016 - head_2_accuracy: 0.9996 - head_2_loss: 0.0016 - head_3_accuracy: 0.9996 - head_3_loss: 0.0016 - head_4_accuracy: 0.9996 - head_4_loss: 0.0016 - head_5_accuracy: 0.9997 - head_5_loss: 0.0016 - head_6_accuracy: 0.9996 - head_6_loss: 0.0016 - head_7_accuracy: 0.9997 - head_7_loss: 0.0016 - loss: 0.0113\n","Epoch 66: val_head_1_accuracy did not improve from 1.00000\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m30s\u001b[0m 95ms/step - head_1_accuracy: 0.9998 - head_1_loss: 0.0010 - head_2_accuracy: 0.9997 - head_2_loss: 0.0011 - head_3_accuracy: 0.9997 - head_3_loss: 0.0011 - head_4_accuracy: 0.9997 - head_4_loss: 0.0010 - head_5_accuracy: 0.9997 - head_5_loss: 0.0011 - head_6_accuracy: 0.9997 - head_6_loss: 0.0011 - head_7_accuracy: 0.9997 - head_7_loss: 0.0011 - loss: 0.0074 - val_head_1_accuracy: 1.0000 - val_head_1_loss: 2.6853e-05 - val_head_2_accuracy: 1.0000 - val_head_2_loss: 2.8857e-05 - val_head_3_accuracy: 1.0000 - val_head_3_loss: 2.6910e-05 - val_head_4_accuracy: 1.0000 - val_head_4_loss: 2.5939e-05 - val_head_5_accuracy: 1.0000 - val_head_5_loss: 2.8919e-05 - val_head_6_accuracy: 1.0000 - val_head_6_loss: 2.9660e-05 - val_head_7_accuracy: 1.0000 - val_head_7_loss: 2.7831e-05 - val_loss: 1.9631e-04 - learning_rate: 1.5839e-04\n","\n","Epoch 67: LearningRateScheduler setting learning rate to 0.00015007500000000006.\n","Epoch 67/100\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 89ms/step - head_1_accuracy: 0.9990 - head_1_loss: 0.0031 - head_2_accuracy: 0.9990 - head_2_loss: 0.0031 - head_3_accuracy: 0.9989 - head_3_loss: 0.0032 - head_4_accuracy: 0.9991 - head_4_loss: 0.0031 - head_5_accuracy: 0.9989 - head_5_loss: 0.0031 - head_6_accuracy: 0.9990 - head_6_loss: 0.0031 - head_7_accuracy: 0.9990 - head_7_loss: 0.0032 - loss: 0.0218\n","Epoch 67: val_head_1_accuracy did not improve from 1.00000\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m30s\u001b[0m 95ms/step - head_1_accuracy: 0.9985 - head_1_loss: 0.0043 - head_2_accuracy: 0.9985 - head_2_loss: 0.0043 - head_3_accuracy: 0.9985 - head_3_loss: 0.0043 - head_4_accuracy: 0.9985 - head_4_loss: 0.0043 - head_5_accuracy: 0.9984 - head_5_loss: 0.0043 - head_6_accuracy: 0.9985 - head_6_loss: 0.0043 - head_7_accuracy: 0.9986 - head_7_loss: 0.0043 - loss: 0.0302 - val_head_1_accuracy: 1.0000 - val_head_1_loss: 2.3327e-04 - val_head_2_accuracy: 1.0000 - val_head_2_loss: 2.3870e-04 - val_head_3_accuracy: 1.0000 - val_head_3_loss: 2.4139e-04 - val_head_4_accuracy: 1.0000 - val_head_4_loss: 2.3167e-04 - val_head_5_accuracy: 1.0000 - val_head_5_loss: 2.3926e-04 - val_head_6_accuracy: 1.0000 - val_head_6_loss: 2.3918e-04 - val_head_7_accuracy: 1.0000 - val_head_7_loss: 2.4779e-04 - val_loss: 0.0017 - learning_rate: 1.5007e-04\n","\n","Epoch 68: LearningRateScheduler setting learning rate to 0.00014190872099022978.\n","Epoch 68/100\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 89ms/step - head_1_accuracy: 0.9993 - head_1_loss: 0.0028 - head_2_accuracy: 0.9992 - head_2_loss: 0.0028 - head_3_accuracy: 0.9993 - head_3_loss: 0.0028 - head_4_accuracy: 0.9994 - head_4_loss: 0.0028 - head_5_accuracy: 0.9992 - head_5_loss: 0.0029 - head_6_accuracy: 0.9992 - head_6_loss: 0.0028 - head_7_accuracy: 0.9993 - head_7_loss: 0.0028 - loss: 0.0198\n","Epoch 68: val_head_1_accuracy did not improve from 1.00000\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m30s\u001b[0m 95ms/step - head_1_accuracy: 0.9995 - head_1_loss: 0.0016 - head_2_accuracy: 0.9995 - head_2_loss: 0.0017 - head_3_accuracy: 0.9996 - head_3_loss: 0.0016 - head_4_accuracy: 0.9996 - head_4_loss: 0.0016 - head_5_accuracy: 0.9995 - head_5_loss: 0.0016 - head_6_accuracy: 0.9995 - head_6_loss: 0.0016 - head_7_accuracy: 0.9995 - head_7_loss: 0.0016 - loss: 0.0114 - val_head_1_accuracy: 1.0000 - val_head_1_loss: 4.4654e-05 - val_head_2_accuracy: 1.0000 - val_head_2_loss: 4.4022e-05 - val_head_3_accuracy: 1.0000 - val_head_3_loss: 4.3183e-05 - val_head_4_accuracy: 1.0000 - val_head_4_loss: 4.4791e-05 - val_head_5_accuracy: 1.0000 - val_head_5_loss: 4.5358e-05 - val_head_6_accuracy: 1.0000 - val_head_6_loss: 4.5882e-05 - val_head_7_accuracy: 1.0000 - val_head_7_loss: 4.5253e-05 - val_loss: 3.1533e-04 - learning_rate: 1.4191e-04\n","\n","Epoch 69: LearningRateScheduler setting learning rate to 0.00013390167684336024.\n","Epoch 69/100\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 89ms/step - head_1_accuracy: 0.9991 - head_1_loss: 0.0024 - head_2_accuracy: 0.9991 - head_2_loss: 0.0025 - head_3_accuracy: 0.9991 - head_3_loss: 0.0024 - head_4_accuracy: 0.9991 - head_4_loss: 0.0024 - head_5_accuracy: 0.9991 - head_5_loss: 0.0024 - head_6_accuracy: 0.9991 - head_6_loss: 0.0024 - head_7_accuracy: 0.9992 - head_7_loss: 0.0024 - loss: 0.0169\n","Epoch 69: val_head_1_accuracy did not improve from 1.00000\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m30s\u001b[0m 95ms/step - head_1_accuracy: 0.9993 - head_1_loss: 0.0022 - head_2_accuracy: 0.9992 - head_2_loss: 0.0022 - head_3_accuracy: 0.9993 - head_3_loss: 0.0022 - head_4_accuracy: 0.9992 - head_4_loss: 0.0021 - head_5_accuracy: 0.9992 - head_5_loss: 0.0022 - head_6_accuracy: 0.9992 - head_6_loss: 0.0021 - head_7_accuracy: 0.9993 - head_7_loss: 0.0021 - loss: 0.0151 - val_head_1_accuracy: 0.9996 - val_head_1_loss: 9.4016e-04 - val_head_2_accuracy: 0.9996 - val_head_2_loss: 9.1932e-04 - val_head_3_accuracy: 0.9996 - val_head_3_loss: 9.4494e-04 - val_head_4_accuracy: 0.9996 - val_head_4_loss: 9.3127e-04 - val_head_5_accuracy: 0.9996 - val_head_5_loss: 9.8071e-04 - val_head_6_accuracy: 0.9996 - val_head_6_loss: 9.3222e-04 - val_head_7_accuracy: 0.9996 - val_head_7_loss: 9.1383e-04 - val_loss: 0.0066 - learning_rate: 1.3390e-04\n","\n","Epoch 70: LearningRateScheduler setting learning rate to 0.00012606192997411915.\n","Epoch 70/100\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 89ms/step - head_1_accuracy: 0.9992 - head_1_loss: 0.0024 - head_2_accuracy: 0.9994 - head_2_loss: 0.0025 - head_3_accuracy: 0.9992 - head_3_loss: 0.0025 - head_4_accuracy: 0.9992 - head_4_loss: 0.0025 - head_5_accuracy: 0.9992 - head_5_loss: 0.0024 - head_6_accuracy: 0.9992 - head_6_loss: 0.0025 - head_7_accuracy: 0.9994 - head_7_loss: 0.0024 - loss: 0.0172\n","Epoch 70: val_head_1_accuracy did not improve from 1.00000\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m30s\u001b[0m 95ms/step - head_1_accuracy: 0.9990 - head_1_loss: 0.0035 - head_2_accuracy: 0.9990 - head_2_loss: 0.0035 - head_3_accuracy: 0.9989 - head_3_loss: 0.0035 - head_4_accuracy: 0.9990 - head_4_loss: 0.0035 - head_5_accuracy: 0.9990 - head_5_loss: 0.0035 - head_6_accuracy: 0.9990 - head_6_loss: 0.0036 - head_7_accuracy: 0.9989 - head_7_loss: 0.0035 - loss: 0.0245 - val_head_1_accuracy: 1.0000 - val_head_1_loss: 1.3542e-04 - val_head_2_accuracy: 1.0000 - val_head_2_loss: 1.3464e-04 - val_head_3_accuracy: 1.0000 - val_head_3_loss: 1.3111e-04 - val_head_4_accuracy: 1.0000 - val_head_4_loss: 1.3306e-04 - val_head_5_accuracy: 0.9999 - val_head_5_loss: 1.3449e-04 - val_head_6_accuracy: 1.0000 - val_head_6_loss: 1.3199e-04 - val_head_7_accuracy: 1.0000 - val_head_7_loss: 1.2605e-04 - val_loss: 9.3321e-04 - learning_rate: 1.2606e-04\n","\n","Epoch 71: LearningRateScheduler setting learning rate to 0.00011839737434305693.\n","Epoch 71/100\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 89ms/step - head_1_accuracy: 0.9993 - head_1_loss: 0.0020 - head_2_accuracy: 0.9995 - head_2_loss: 0.0020 - head_3_accuracy: 0.9993 - head_3_loss: 0.0021 - head_4_accuracy: 0.9992 - head_4_loss: 0.0020 - head_5_accuracy: 0.9994 - head_5_loss: 0.0020 - head_6_accuracy: 0.9992 - head_6_loss: 0.0020 - head_7_accuracy: 0.9992 - head_7_loss: 0.0020 - loss: 0.0141\n","Epoch 71: val_head_1_accuracy did not improve from 1.00000\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m30s\u001b[0m 95ms/step - head_1_accuracy: 0.9996 - head_1_loss: 0.0014 - head_2_accuracy: 0.9997 - head_2_loss: 0.0014 - head_3_accuracy: 0.9996 - head_3_loss: 0.0014 - head_4_accuracy: 0.9996 - head_4_loss: 0.0014 - head_5_accuracy: 0.9996 - head_5_loss: 0.0014 - head_6_accuracy: 0.9995 - head_6_loss: 0.0014 - head_7_accuracy: 0.9996 - head_7_loss: 0.0014 - loss: 0.0098 - val_head_1_accuracy: 0.9999 - val_head_1_loss: 4.3220e-04 - val_head_2_accuracy: 0.9999 - val_head_2_loss: 4.7174e-04 - val_head_3_accuracy: 0.9999 - val_head_3_loss: 4.5067e-04 - val_head_4_accuracy: 0.9999 - val_head_4_loss: 4.6863e-04 - val_head_5_accuracy: 0.9999 - val_head_5_loss: 4.2757e-04 - val_head_6_accuracy: 0.9999 - val_head_6_loss: 4.5198e-04 - val_head_7_accuracy: 0.9999 - val_head_7_loss: 4.4567e-04 - val_loss: 0.0032 - learning_rate: 1.1840e-04\n","\n","Epoch 72: LearningRateScheduler setting learning rate to 0.00011091572750799748.\n","Epoch 72/100\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 89ms/step - head_1_accuracy: 0.9991 - head_1_loss: 0.0027 - head_2_accuracy: 0.9992 - head_2_loss: 0.0026 - head_3_accuracy: 0.9991 - head_3_loss: 0.0026 - head_4_accuracy: 0.9992 - head_4_loss: 0.0026 - head_5_accuracy: 0.9991 - head_5_loss: 0.0026 - head_6_accuracy: 0.9992 - head_6_loss: 0.0025 - head_7_accuracy: 0.9991 - head_7_loss: 0.0026 - loss: 0.0183\n","Epoch 72: val_head_1_accuracy did not improve from 1.00000\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m30s\u001b[0m 95ms/step - head_1_accuracy: 0.9993 - head_1_loss: 0.0023 - head_2_accuracy: 0.9993 - head_2_loss: 0.0023 - head_3_accuracy: 0.9993 - head_3_loss: 0.0023 - head_4_accuracy: 0.9993 - head_4_loss: 0.0023 - head_5_accuracy: 0.9993 - head_5_loss: 0.0023 - head_6_accuracy: 0.9993 - head_6_loss: 0.0023 - head_7_accuracy: 0.9993 - head_7_loss: 0.0023 - loss: 0.0163 - val_head_1_accuracy: 0.9999 - val_head_1_loss: 1.6518e-04 - val_head_2_accuracy: 0.9999 - val_head_2_loss: 1.7783e-04 - val_head_3_accuracy: 0.9999 - val_head_3_loss: 1.6543e-04 - val_head_4_accuracy: 0.9999 - val_head_4_loss: 1.7282e-04 - val_head_5_accuracy: 0.9999 - val_head_5_loss: 1.6192e-04 - val_head_6_accuracy: 0.9999 - val_head_6_loss: 1.7459e-04 - val_head_7_accuracy: 0.9999 - val_head_7_loss: 1.6490e-04 - val_loss: 0.0012 - learning_rate: 1.1092e-04\n","\n","Epoch 73: LearningRateScheduler setting learning rate to 0.00010362452285311177.\n","Epoch 73/100\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 89ms/step - head_1_accuracy: 0.9996 - head_1_loss: 0.0016 - head_2_accuracy: 0.9996 - head_2_loss: 0.0016 - head_3_accuracy: 0.9996 - head_3_loss: 0.0016 - head_4_accuracy: 0.9995 - head_4_loss: 0.0016 - head_5_accuracy: 0.9996 - head_5_loss: 0.0016 - head_6_accuracy: 0.9995 - head_6_loss: 0.0016 - head_7_accuracy: 0.9996 - head_7_loss: 0.0016 - loss: 0.0113\n","Epoch 73: val_head_1_accuracy did not improve from 1.00000\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m30s\u001b[0m 95ms/step - head_1_accuracy: 0.9997 - head_1_loss: 0.0013 - head_2_accuracy: 0.9997 - head_2_loss: 0.0013 - head_3_accuracy: 0.9997 - head_3_loss: 0.0013 - head_4_accuracy: 0.9997 - head_4_loss: 0.0013 - head_5_accuracy: 0.9997 - head_5_loss: 0.0013 - head_6_accuracy: 0.9997 - head_6_loss: 0.0013 - head_7_accuracy: 0.9997 - head_7_loss: 0.0013 - loss: 0.0092 - val_head_1_accuracy: 1.0000 - val_head_1_loss: 7.1246e-05 - val_head_2_accuracy: 0.9999 - val_head_2_loss: 7.9317e-05 - val_head_3_accuracy: 0.9999 - val_head_3_loss: 7.4067e-05 - val_head_4_accuracy: 0.9999 - val_head_4_loss: 8.2463e-05 - val_head_5_accuracy: 1.0000 - val_head_5_loss: 6.6600e-05 - val_head_6_accuracy: 0.9999 - val_head_6_loss: 7.6179e-05 - val_head_7_accuracy: 0.9999 - val_head_7_loss: 7.5369e-05 - val_loss: 5.2891e-04 - learning_rate: 1.0362e-04\n","\n","Epoch 74: LearningRateScheduler setting learning rate to 9.653110200343824e-05.\n","Epoch 74/100\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 89ms/step - head_1_accuracy: 0.9995 - head_1_loss: 0.0018 - head_2_accuracy: 0.9995 - head_2_loss: 0.0019 - head_3_accuracy: 0.9995 - head_3_loss: 0.0018 - head_4_accuracy: 0.9995 - head_4_loss: 0.0018 - head_5_accuracy: 0.9995 - head_5_loss: 0.0018 - head_6_accuracy: 0.9995 - head_6_loss: 0.0018 - head_7_accuracy: 0.9995 - head_7_loss: 0.0018 - loss: 0.0127\n","Epoch 74: val_head_1_accuracy did not improve from 1.00000\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m30s\u001b[0m 95ms/step - head_1_accuracy: 0.9996 - head_1_loss: 0.0014 - head_2_accuracy: 0.9996 - head_2_loss: 0.0014 - head_3_accuracy: 0.9996 - head_3_loss: 0.0014 - head_4_accuracy: 0.9997 - head_4_loss: 0.0014 - head_5_accuracy: 0.9996 - head_5_loss: 0.0014 - head_6_accuracy: 0.9996 - head_6_loss: 0.0014 - head_7_accuracy: 0.9996 - head_7_loss: 0.0014 - loss: 0.0097 - val_head_1_accuracy: 0.9988 - val_head_1_loss: 0.0057 - val_head_2_accuracy: 0.9988 - val_head_2_loss: 0.0057 - val_head_3_accuracy: 0.9989 - val_head_3_loss: 0.0057 - val_head_4_accuracy: 0.9989 - val_head_4_loss: 0.0058 - val_head_5_accuracy: 0.9989 - val_head_5_loss: 0.0056 - val_head_6_accuracy: 0.9989 - val_head_6_loss: 0.0057 - val_head_7_accuracy: 0.9989 - val_head_7_loss: 0.0056 - val_loss: 0.0401 - learning_rate: 9.6531e-05\n","\n","Epoch 75: LearningRateScheduler setting learning rate to 8.96426074324889e-05.\n","Epoch 75/100\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 89ms/step - head_1_accuracy: 0.9995 - head_1_loss: 0.0015 - head_2_accuracy: 0.9994 - head_2_loss: 0.0016 - head_3_accuracy: 0.9994 - head_3_loss: 0.0015 - head_4_accuracy: 0.9994 - head_4_loss: 0.0015 - head_5_accuracy: 0.9995 - head_5_loss: 0.0015 - head_6_accuracy: 0.9994 - head_6_loss: 0.0016 - head_7_accuracy: 0.9995 - head_7_loss: 0.0015 - loss: 0.0107\n","Epoch 75: val_head_1_accuracy did not improve from 1.00000\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m30s\u001b[0m 95ms/step - head_1_accuracy: 0.9997 - head_1_loss: 0.0011 - head_2_accuracy: 0.9996 - head_2_loss: 0.0011 - head_3_accuracy: 0.9997 - head_3_loss: 0.0011 - head_4_accuracy: 0.9996 - head_4_loss: 0.0011 - head_5_accuracy: 0.9997 - head_5_loss: 0.0011 - head_6_accuracy: 0.9997 - head_6_loss: 0.0011 - head_7_accuracy: 0.9997 - head_7_loss: 0.0011 - loss: 0.0075 - val_head_1_accuracy: 0.9999 - val_head_1_loss: 1.4751e-04 - val_head_2_accuracy: 0.9999 - val_head_2_loss: 1.5515e-04 - val_head_3_accuracy: 0.9999 - val_head_3_loss: 1.5027e-04 - val_head_4_accuracy: 0.9999 - val_head_4_loss: 1.5336e-04 - val_head_5_accuracy: 0.9999 - val_head_5_loss: 1.5199e-04 - val_head_6_accuracy: 0.9999 - val_head_6_loss: 1.5814e-04 - val_head_7_accuracy: 0.9999 - val_head_7_loss: 1.5268e-04 - val_loss: 0.0011 - learning_rate: 8.9643e-05\n","\n","Epoch 76: LearningRateScheduler setting learning rate to 8.29659752703842e-05.\n","Epoch 76/100\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 89ms/step - head_1_accuracy: 0.9995 - head_1_loss: 0.0012 - head_2_accuracy: 0.9995 - head_2_loss: 0.0012 - head_3_accuracy: 0.9995 - head_3_loss: 0.0012 - head_4_accuracy: 0.9996 - head_4_loss: 0.0011 - head_5_accuracy: 0.9995 - head_5_loss: 0.0011 - head_6_accuracy: 0.9995 - head_6_loss: 0.0012 - head_7_accuracy: 0.9995 - head_7_loss: 0.0011 - loss: 0.0081\n","Epoch 76: val_head_1_accuracy did not improve from 1.00000\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m30s\u001b[0m 95ms/step - head_1_accuracy: 0.9998 - head_1_loss: 7.2562e-04 - head_2_accuracy: 0.9998 - head_2_loss: 7.0393e-04 - head_3_accuracy: 0.9998 - head_3_loss: 7.2023e-04 - head_4_accuracy: 0.9998 - head_4_loss: 6.7451e-04 - head_5_accuracy: 0.9998 - head_5_loss: 6.7303e-04 - head_6_accuracy: 0.9998 - head_6_loss: 7.0709e-04 - head_7_accuracy: 0.9998 - head_7_loss: 7.0365e-04 - loss: 0.0049 - val_head_1_accuracy: 1.0000 - val_head_1_loss: 5.1532e-05 - val_head_2_accuracy: 1.0000 - val_head_2_loss: 4.9459e-05 - val_head_3_accuracy: 1.0000 - val_head_3_loss: 5.2481e-05 - val_head_4_accuracy: 1.0000 - val_head_4_loss: 5.1719e-05 - val_head_5_accuracy: 1.0000 - val_head_5_loss: 4.7487e-05 - val_head_6_accuracy: 1.0000 - val_head_6_loss: 4.8762e-05 - val_head_7_accuracy: 1.0000 - val_head_7_loss: 4.5578e-05 - val_loss: 3.4944e-04 - learning_rate: 8.2966e-05\n","\n","Epoch 77: LearningRateScheduler setting learning rate to 7.650792831975739e-05.\n","Epoch 77/100\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 89ms/step - head_1_accuracy: 0.9997 - head_1_loss: 0.0011 - head_2_accuracy: 0.9998 - head_2_loss: 0.0011 - head_3_accuracy: 0.9997 - head_3_loss: 0.0011 - head_4_accuracy: 0.9997 - head_4_loss: 0.0011 - head_5_accuracy: 0.9998 - head_5_loss: 0.0011 - head_6_accuracy: 0.9998 - head_6_loss: 0.0011 - head_7_accuracy: 0.9998 - head_7_loss: 0.0011 - loss: 0.0078\n","Epoch 77: val_head_1_accuracy did not improve from 1.00000\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m30s\u001b[0m 95ms/step - head_1_accuracy: 0.9992 - head_1_loss: 0.0030 - head_2_accuracy: 0.9992 - head_2_loss: 0.0030 - head_3_accuracy: 0.9992 - head_3_loss: 0.0030 - head_4_accuracy: 0.9992 - head_4_loss: 0.0030 - head_5_accuracy: 0.9993 - head_5_loss: 0.0030 - head_6_accuracy: 0.9992 - head_6_loss: 0.0030 - head_7_accuracy: 0.9993 - head_7_loss: 0.0030 - loss: 0.0209 - val_head_1_accuracy: 1.0000 - val_head_1_loss: 4.8601e-05 - val_head_2_accuracy: 1.0000 - val_head_2_loss: 4.8289e-05 - val_head_3_accuracy: 1.0000 - val_head_3_loss: 4.8854e-05 - val_head_4_accuracy: 1.0000 - val_head_4_loss: 4.7436e-05 - val_head_5_accuracy: 1.0000 - val_head_5_loss: 4.5307e-05 - val_head_6_accuracy: 1.0000 - val_head_6_loss: 4.7890e-05 - val_head_7_accuracy: 1.0000 - val_head_7_loss: 4.5236e-05 - val_loss: 3.3392e-04 - learning_rate: 7.6508e-05\n","\n","Epoch 78: LearningRateScheduler setting learning rate to 7.027496928646257e-05.\n","Epoch 78/100\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 89ms/step - head_1_accuracy: 0.9999 - head_1_loss: 8.3512e-04 - head_2_accuracy: 0.9998 - head_2_loss: 8.4918e-04 - head_3_accuracy: 0.9999 - head_3_loss: 8.0229e-04 - head_4_accuracy: 0.9998 - head_4_loss: 8.4191e-04 - head_5_accuracy: 0.9998 - head_5_loss: 8.4668e-04 - head_6_accuracy: 0.9998 - head_6_loss: 8.4039e-04 - head_7_accuracy: 0.9998 - head_7_loss: 8.5800e-04 - loss: 0.0059\n","Epoch 78: val_head_1_accuracy did not improve from 1.00000\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m30s\u001b[0m 95ms/step - head_1_accuracy: 0.9999 - head_1_loss: 5.8673e-04 - head_2_accuracy: 0.9999 - head_2_loss: 6.0381e-04 - head_3_accuracy: 0.9999 - head_3_loss: 5.7807e-04 - head_4_accuracy: 0.9999 - head_4_loss: 5.8848e-04 - head_5_accuracy: 0.9999 - head_5_loss: 5.9712e-04 - head_6_accuracy: 0.9998 - head_6_loss: 5.9147e-04 - head_7_accuracy: 0.9999 - head_7_loss: 6.0007e-04 - loss: 0.0041 - val_head_1_accuracy: 0.9999 - val_head_1_loss: 1.6805e-04 - val_head_2_accuracy: 0.9999 - val_head_2_loss: 1.7380e-04 - val_head_3_accuracy: 0.9999 - val_head_3_loss: 1.7572e-04 - val_head_4_accuracy: 0.9999 - val_head_4_loss: 1.7454e-04 - val_head_5_accuracy: 0.9999 - val_head_5_loss: 1.7173e-04 - val_head_6_accuracy: 0.9999 - val_head_6_loss: 1.7697e-04 - val_head_7_accuracy: 0.9999 - val_head_7_loss: 1.7377e-04 - val_loss: 0.0012 - learning_rate: 7.0275e-05\n","\n","Epoch 79: LearningRateScheduler setting learning rate to 6.427337423190096e-05.\n","Epoch 79/100\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 89ms/step - head_1_accuracy: 0.9998 - head_1_loss: 5.6513e-04 - head_2_accuracy: 0.9998 - head_2_loss: 5.3525e-04 - head_3_accuracy: 0.9998 - head_3_loss: 5.6253e-04 - head_4_accuracy: 0.9998 - head_4_loss: 5.3651e-04 - head_5_accuracy: 0.9998 - head_5_loss: 5.5982e-04 - head_6_accuracy: 0.9998 - head_6_loss: 5.1316e-04 - head_7_accuracy: 0.9998 - head_7_loss: 5.4792e-04 - loss: 0.0038\n","Epoch 79: val_head_1_accuracy did not improve from 1.00000\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m30s\u001b[0m 96ms/step - head_1_accuracy: 0.9999 - head_1_loss: 4.6586e-04 - head_2_accuracy: 0.9999 - head_2_loss: 4.4464e-04 - head_3_accuracy: 0.9999 - head_3_loss: 4.5710e-04 - head_4_accuracy: 0.9999 - head_4_loss: 4.4726e-04 - head_5_accuracy: 0.9999 - head_5_loss: 4.6613e-04 - head_6_accuracy: 0.9999 - head_6_loss: 4.3658e-04 - head_7_accuracy: 0.9999 - head_7_loss: 4.6393e-04 - loss: 0.0032 - val_head_1_accuracy: 1.0000 - val_head_1_loss: 5.6864e-05 - val_head_2_accuracy: 1.0000 - val_head_2_loss: 5.3938e-05 - val_head_3_accuracy: 1.0000 - val_head_3_loss: 5.5484e-05 - val_head_4_accuracy: 1.0000 - val_head_4_loss: 5.5302e-05 - val_head_5_accuracy: 1.0000 - val_head_5_loss: 5.1014e-05 - val_head_6_accuracy: 1.0000 - val_head_6_loss: 5.0242e-05 - val_head_7_accuracy: 1.0000 - val_head_7_loss: 5.0035e-05 - val_loss: 3.7548e-04 - learning_rate: 6.4273e-05\n","\n","Epoch 80: LearningRateScheduler setting learning rate to 5.850918625355897e-05.\n","Epoch 80/100\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 89ms/step - head_1_accuracy: 0.9998 - head_1_loss: 7.6611e-04 - head_2_accuracy: 0.9998 - head_2_loss: 7.5267e-04 - head_3_accuracy: 0.9997 - head_3_loss: 7.6456e-04 - head_4_accuracy: 0.9998 - head_4_loss: 7.4608e-04 - head_5_accuracy: 0.9998 - head_5_loss: 7.6162e-04 - head_6_accuracy: 0.9998 - head_6_loss: 7.4543e-04 - head_7_accuracy: 0.9997 - head_7_loss: 7.6203e-04 - loss: 0.0053\n","Epoch 80: val_head_1_accuracy did not improve from 1.00000\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m30s\u001b[0m 95ms/step - head_1_accuracy: 0.9998 - head_1_loss: 8.2296e-04 - head_2_accuracy: 0.9997 - head_2_loss: 8.2150e-04 - head_3_accuracy: 0.9997 - head_3_loss: 8.0530e-04 - head_4_accuracy: 0.9998 - head_4_loss: 7.9988e-04 - head_5_accuracy: 0.9997 - head_5_loss: 8.3401e-04 - head_6_accuracy: 0.9998 - head_6_loss: 8.1273e-04 - head_7_accuracy: 0.9997 - head_7_loss: 8.1664e-04 - loss: 0.0057 - val_head_1_accuracy: 1.0000 - val_head_1_loss: 5.6477e-05 - val_head_2_accuracy: 1.0000 - val_head_2_loss: 5.4542e-05 - val_head_3_accuracy: 1.0000 - val_head_3_loss: 5.6506e-05 - val_head_4_accuracy: 1.0000 - val_head_4_loss: 5.5977e-05 - val_head_5_accuracy: 1.0000 - val_head_5_loss: 5.2193e-05 - val_head_6_accuracy: 1.0000 - val_head_6_loss: 5.4756e-05 - val_head_7_accuracy: 1.0000 - val_head_7_loss: 5.5457e-05 - val_loss: 3.8860e-04 - learning_rate: 5.8509e-05\n","\n","Epoch 81: LearningRateScheduler setting learning rate to 5.298820940012169e-05.\n","Epoch 81/100\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 89ms/step - head_1_accuracy: 0.9999 - head_1_loss: 2.8501e-04 - head_2_accuracy: 0.9999 - head_2_loss: 2.7915e-04 - head_3_accuracy: 0.9999 - head_3_loss: 2.7280e-04 - head_4_accuracy: 0.9999 - head_4_loss: 2.7919e-04 - head_5_accuracy: 1.0000 - head_5_loss: 2.6439e-04 - head_6_accuracy: 0.9999 - head_6_loss: 2.8232e-04 - head_7_accuracy: 0.9999 - head_7_loss: 2.8337e-04 - loss: 0.0019\n","Epoch 81: val_head_1_accuracy did not improve from 1.00000\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m30s\u001b[0m 95ms/step - head_1_accuracy: 0.9999 - head_1_loss: 4.0046e-04 - head_2_accuracy: 0.9999 - head_2_loss: 3.7977e-04 - head_3_accuracy: 0.9999 - head_3_loss: 3.9442e-04 - head_4_accuracy: 0.9999 - head_4_loss: 3.8472e-04 - head_5_accuracy: 0.9999 - head_5_loss: 3.8813e-04 - head_6_accuracy: 0.9999 - head_6_loss: 3.8766e-04 - head_7_accuracy: 0.9999 - head_7_loss: 3.8965e-04 - loss: 0.0027 - val_head_1_accuracy: 1.0000 - val_head_1_loss: 6.2386e-05 - val_head_2_accuracy: 1.0000 - val_head_2_loss: 6.3572e-05 - val_head_3_accuracy: 1.0000 - val_head_3_loss: 6.4804e-05 - val_head_4_accuracy: 1.0000 - val_head_4_loss: 6.3669e-05 - val_head_5_accuracy: 1.0000 - val_head_5_loss: 6.2610e-05 - val_head_6_accuracy: 1.0000 - val_head_6_loss: 6.5055e-05 - val_head_7_accuracy: 1.0000 - val_head_7_loss: 6.0941e-05 - val_loss: 4.4613e-04 - learning_rate: 5.2988e-05\n","\n","Epoch 82: LearningRateScheduler setting learning rate to 4.7716002827287234e-05.\n","Epoch 82/100\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 89ms/step - head_1_accuracy: 1.0000 - head_1_loss: 2.2160e-04 - head_2_accuracy: 1.0000 - head_2_loss: 2.3051e-04 - head_3_accuracy: 1.0000 - head_3_loss: 2.2523e-04 - head_4_accuracy: 1.0000 - head_4_loss: 2.3036e-04 - head_5_accuracy: 1.0000 - head_5_loss: 2.2890e-04 - head_6_accuracy: 1.0000 - head_6_loss: 2.3410e-04 - head_7_accuracy: 1.0000 - head_7_loss: 2.2490e-04 - loss: 0.0016\n","Epoch 82: val_head_1_accuracy did not improve from 1.00000\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m30s\u001b[0m 95ms/step - head_1_accuracy: 0.9999 - head_1_loss: 3.9494e-04 - head_2_accuracy: 0.9999 - head_2_loss: 4.0683e-04 - head_3_accuracy: 0.9999 - head_3_loss: 4.0554e-04 - head_4_accuracy: 0.9999 - head_4_loss: 4.1211e-04 - head_5_accuracy: 0.9999 - head_5_loss: 4.2304e-04 - head_6_accuracy: 0.9999 - head_6_loss: 4.2407e-04 - head_7_accuracy: 0.9999 - head_7_loss: 4.0730e-04 - loss: 0.0029 - val_head_1_accuracy: 1.0000 - val_head_1_loss: 2.5394e-05 - val_head_2_accuracy: 1.0000 - val_head_2_loss: 2.7252e-05 - val_head_3_accuracy: 1.0000 - val_head_3_loss: 2.7814e-05 - val_head_4_accuracy: 1.0000 - val_head_4_loss: 2.5563e-05 - val_head_5_accuracy: 1.0000 - val_head_5_loss: 2.4278e-05 - val_head_6_accuracy: 1.0000 - val_head_6_loss: 2.6670e-05 - val_head_7_accuracy: 1.0000 - val_head_7_loss: 2.4510e-05 - val_loss: 1.8274e-04 - learning_rate: 4.7716e-05\n","\n","Epoch 83: LearningRateScheduler setting learning rate to 4.26978752001687e-05.\n","Epoch 83/100\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 89ms/step - head_1_accuracy: 0.9999 - head_1_loss: 3.5709e-04 - head_2_accuracy: 0.9999 - head_2_loss: 3.5464e-04 - head_3_accuracy: 0.9999 - head_3_loss: 3.6564e-04 - head_4_accuracy: 0.9999 - head_4_loss: 3.7222e-04 - head_5_accuracy: 0.9999 - head_5_loss: 3.6134e-04 - head_6_accuracy: 0.9999 - head_6_loss: 3.7280e-04 - head_7_accuracy: 0.9999 - head_7_loss: 3.6627e-04 - loss: 0.0026\n","Epoch 83: val_head_1_accuracy did not improve from 1.00000\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m30s\u001b[0m 95ms/step - head_1_accuracy: 0.9999 - head_1_loss: 3.5264e-04 - head_2_accuracy: 0.9999 - head_2_loss: 3.4693e-04 - head_3_accuracy: 0.9999 - head_3_loss: 3.5780e-04 - head_4_accuracy: 0.9999 - head_4_loss: 3.6050e-04 - head_5_accuracy: 0.9999 - head_5_loss: 3.5162e-04 - head_6_accuracy: 0.9999 - head_6_loss: 3.6530e-04 - head_7_accuracy: 0.9999 - head_7_loss: 3.5773e-04 - loss: 0.0025 - val_head_1_accuracy: 1.0000 - val_head_1_loss: 1.3172e-05 - val_head_2_accuracy: 1.0000 - val_head_2_loss: 1.3013e-05 - val_head_3_accuracy: 1.0000 - val_head_3_loss: 1.3067e-05 - val_head_4_accuracy: 1.0000 - val_head_4_loss: 1.3437e-05 - val_head_5_accuracy: 1.0000 - val_head_5_loss: 1.3277e-05 - val_head_6_accuracy: 1.0000 - val_head_6_loss: 1.3287e-05 - val_head_7_accuracy: 1.0000 - val_head_7_loss: 1.3586e-05 - val_loss: 9.3485e-05 - learning_rate: 4.2698e-05\n","\n","Epoch 84: LearningRateScheduler setting learning rate to 3.793887934791801e-05.\n","Epoch 84/100\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 89ms/step - head_1_accuracy: 1.0000 - head_1_loss: 1.7101e-04 - head_2_accuracy: 1.0000 - head_2_loss: 1.7265e-04 - head_3_accuracy: 1.0000 - head_3_loss: 1.7886e-04 - head_4_accuracy: 1.0000 - head_4_loss: 1.7530e-04 - head_5_accuracy: 1.0000 - head_5_loss: 1.6560e-04 - head_6_accuracy: 1.0000 - head_6_loss: 1.6838e-04 - head_7_accuracy: 1.0000 - head_7_loss: 1.7570e-04 - loss: 0.0012\n","Epoch 84: val_head_1_accuracy did not improve from 1.00000\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m30s\u001b[0m 95ms/step - head_1_accuracy: 1.0000 - head_1_loss: 1.5775e-04 - head_2_accuracy: 1.0000 - head_2_loss: 1.5823e-04 - head_3_accuracy: 1.0000 - head_3_loss: 1.6352e-04 - head_4_accuracy: 1.0000 - head_4_loss: 1.6149e-04 - head_5_accuracy: 1.0000 - head_5_loss: 1.5685e-04 - head_6_accuracy: 1.0000 - head_6_loss: 1.5743e-04 - head_7_accuracy: 1.0000 - head_7_loss: 1.6310e-04 - loss: 0.0011 - val_head_1_accuracy: 0.9999 - val_head_1_loss: 3.0656e-04 - val_head_2_accuracy: 0.9999 - val_head_2_loss: 2.9264e-04 - val_head_3_accuracy: 0.9999 - val_head_3_loss: 2.9940e-04 - val_head_4_accuracy: 0.9999 - val_head_4_loss: 3.0273e-04 - val_head_5_accuracy: 0.9999 - val_head_5_loss: 2.9481e-04 - val_head_6_accuracy: 0.9999 - val_head_6_loss: 2.8457e-04 - val_head_7_accuracy: 0.9999 - val_head_7_loss: 2.8605e-04 - val_loss: 0.0021 - learning_rate: 3.7939e-05\n","\n","Epoch 85: LearningRateScheduler setting learning rate to 3.344380717595574e-05.\n","Epoch 85/100\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 89ms/step - head_1_accuracy: 1.0000 - head_1_loss: 2.6496e-04 - head_2_accuracy: 0.9999 - head_2_loss: 2.5632e-04 - head_3_accuracy: 1.0000 - head_3_loss: 2.6068e-04 - head_4_accuracy: 1.0000 - head_4_loss: 2.6322e-04 - head_5_accuracy: 1.0000 - head_5_loss: 2.4849e-04 - head_6_accuracy: 1.0000 - head_6_loss: 2.5119e-04 - head_7_accuracy: 1.0000 - head_7_loss: 2.4150e-04 - loss: 0.0018\n","Epoch 85: val_head_1_accuracy did not improve from 1.00000\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m30s\u001b[0m 95ms/step - head_1_accuracy: 1.0000 - head_1_loss: 1.7626e-04 - head_2_accuracy: 1.0000 - head_2_loss: 1.7126e-04 - head_3_accuracy: 1.0000 - head_3_loss: 1.7330e-04 - head_4_accuracy: 1.0000 - head_4_loss: 1.7671e-04 - head_5_accuracy: 1.0000 - head_5_loss: 1.7252e-04 - head_6_accuracy: 1.0000 - head_6_loss: 1.7298e-04 - head_7_accuracy: 1.0000 - head_7_loss: 1.7400e-04 - loss: 0.0012 - val_head_1_accuracy: 1.0000 - val_head_1_loss: 9.6045e-06 - val_head_2_accuracy: 1.0000 - val_head_2_loss: 1.0111e-05 - val_head_3_accuracy: 1.0000 - val_head_3_loss: 9.4733e-06 - val_head_4_accuracy: 1.0000 - val_head_4_loss: 1.0498e-05 - val_head_5_accuracy: 1.0000 - val_head_5_loss: 9.3808e-06 - val_head_6_accuracy: 1.0000 - val_head_6_loss: 9.9170e-06 - val_head_7_accuracy: 1.0000 - val_head_7_loss: 9.9518e-06 - val_loss: 6.9417e-05 - learning_rate: 3.3444e-05\n","\n","Epoch 86: LearningRateScheduler setting learning rate to 2.9217184840928013e-05.\n","Epoch 86/100\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 89ms/step - head_1_accuracy: 0.9999 - head_1_loss: 2.2326e-04 - head_2_accuracy: 0.9999 - head_2_loss: 2.1136e-04 - head_3_accuracy: 0.9999 - head_3_loss: 2.0592e-04 - head_4_accuracy: 1.0000 - head_4_loss: 2.0056e-04 - head_5_accuracy: 0.9999 - head_5_loss: 2.1665e-04 - head_6_accuracy: 0.9999 - head_6_loss: 2.2289e-04 - head_7_accuracy: 1.0000 - head_7_loss: 1.9896e-04 - loss: 0.0015\n","Epoch 86: val_head_1_accuracy did not improve from 1.00000\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m30s\u001b[0m 95ms/step - head_1_accuracy: 1.0000 - head_1_loss: 1.6984e-04 - head_2_accuracy: 1.0000 - head_2_loss: 1.6066e-04 - head_3_accuracy: 1.0000 - head_3_loss: 1.5818e-04 - head_4_accuracy: 1.0000 - head_4_loss: 1.5253e-04 - head_5_accuracy: 0.9999 - head_5_loss: 1.6503e-04 - head_6_accuracy: 1.0000 - head_6_loss: 1.6654e-04 - head_7_accuracy: 1.0000 - head_7_loss: 1.5172e-04 - loss: 0.0011 - val_head_1_accuracy: 1.0000 - val_head_1_loss: 9.1678e-06 - val_head_2_accuracy: 1.0000 - val_head_2_loss: 8.8066e-06 - val_head_3_accuracy: 1.0000 - val_head_3_loss: 8.8309e-06 - val_head_4_accuracy: 1.0000 - val_head_4_loss: 8.8174e-06 - val_head_5_accuracy: 1.0000 - val_head_5_loss: 8.7825e-06 - val_head_6_accuracy: 1.0000 - val_head_6_loss: 9.0212e-06 - val_head_7_accuracy: 1.0000 - val_head_7_loss: 8.9285e-06 - val_loss: 6.2789e-05 - learning_rate: 2.9217e-05\n","\n","Epoch 87: LearningRateScheduler setting learning rate to 2.526326819325076e-05.\n","Epoch 87/100\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 89ms/step - head_1_accuracy: 1.0000 - head_1_loss: 1.5144e-04 - head_2_accuracy: 1.0000 - head_2_loss: 1.5157e-04 - head_3_accuracy: 1.0000 - head_3_loss: 1.5541e-04 - head_4_accuracy: 1.0000 - head_4_loss: 1.5121e-04 - head_5_accuracy: 1.0000 - head_5_loss: 1.5460e-04 - head_6_accuracy: 1.0000 - head_6_loss: 1.4563e-04 - head_7_accuracy: 1.0000 - head_7_loss: 1.4581e-04 - loss: 0.0011\n","Epoch 87: val_head_1_accuracy did not improve from 1.00000\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m30s\u001b[0m 95ms/step - head_1_accuracy: 1.0000 - head_1_loss: 1.2762e-04 - head_2_accuracy: 1.0000 - head_2_loss: 1.2592e-04 - head_3_accuracy: 1.0000 - head_3_loss: 1.3204e-04 - head_4_accuracy: 1.0000 - head_4_loss: 1.2748e-04 - head_5_accuracy: 1.0000 - head_5_loss: 1.3330e-04 - head_6_accuracy: 1.0000 - head_6_loss: 1.2290e-04 - head_7_accuracy: 1.0000 - head_7_loss: 1.2440e-04 - loss: 8.9161e-04 - val_head_1_accuracy: 0.9999 - val_head_1_loss: 7.4092e-05 - val_head_2_accuracy: 0.9999 - val_head_2_loss: 7.9275e-05 - val_head_3_accuracy: 0.9999 - val_head_3_loss: 8.1123e-05 - val_head_4_accuracy: 0.9999 - val_head_4_loss: 8.0718e-05 - val_head_5_accuracy: 0.9999 - val_head_5_loss: 8.3517e-05 - val_head_6_accuracy: 0.9999 - val_head_6_loss: 8.4519e-05 - val_head_7_accuracy: 0.9999 - val_head_7_loss: 8.0592e-05 - val_loss: 5.6776e-04 - learning_rate: 2.5263e-05\n","\n","Epoch 88: LearningRateScheduler setting learning rate to 2.1586038491829064e-05.\n","Epoch 88/100\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 89ms/step - head_1_accuracy: 1.0000 - head_1_loss: 1.3718e-04 - head_2_accuracy: 1.0000 - head_2_loss: 1.3472e-04 - head_3_accuracy: 1.0000 - head_3_loss: 1.2878e-04 - head_4_accuracy: 1.0000 - head_4_loss: 1.3393e-04 - head_5_accuracy: 1.0000 - head_5_loss: 1.4324e-04 - head_6_accuracy: 1.0000 - head_6_loss: 1.3084e-04 - head_7_accuracy: 1.0000 - head_7_loss: 1.3156e-04 - loss: 9.4026e-04\n","Epoch 88: val_head_1_accuracy did not improve from 1.00000\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m30s\u001b[0m 95ms/step - head_1_accuracy: 1.0000 - head_1_loss: 1.3968e-04 - head_2_accuracy: 1.0000 - head_2_loss: 1.4338e-04 - head_3_accuracy: 1.0000 - head_3_loss: 1.2835e-04 - head_4_accuracy: 1.0000 - head_4_loss: 1.3780e-04 - head_5_accuracy: 1.0000 - head_5_loss: 1.4493e-04 - head_6_accuracy: 1.0000 - head_6_loss: 1.3851e-04 - head_7_accuracy: 1.0000 - head_7_loss: 1.4158e-04 - loss: 9.7479e-04 - val_head_1_accuracy: 1.0000 - val_head_1_loss: 2.3098e-05 - val_head_2_accuracy: 1.0000 - val_head_2_loss: 2.3155e-05 - val_head_3_accuracy: 1.0000 - val_head_3_loss: 2.4052e-05 - val_head_4_accuracy: 1.0000 - val_head_4_loss: 2.4080e-05 - val_head_5_accuracy: 1.0000 - val_head_5_loss: 2.3612e-05 - val_head_6_accuracy: 1.0000 - val_head_6_loss: 2.3385e-05 - val_head_7_accuracy: 1.0000 - val_head_7_loss: 2.2273e-05 - val_loss: 1.6480e-04 - learning_rate: 2.1586e-05\n","\n","Epoch 89: LearningRateScheduler setting learning rate to 1.8189198395266834e-05.\n","Epoch 89/100\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 89ms/step - head_1_accuracy: 0.9998 - head_1_loss: 5.4356e-04 - head_2_accuracy: 0.9998 - head_2_loss: 5.5722e-04 - head_3_accuracy: 0.9998 - head_3_loss: 5.2477e-04 - head_4_accuracy: 0.9998 - head_4_loss: 5.4821e-04 - head_5_accuracy: 0.9998 - head_5_loss: 5.6561e-04 - head_6_accuracy: 0.9998 - head_6_loss: 5.2708e-04 - head_7_accuracy: 0.9998 - head_7_loss: 5.2942e-04 - loss: 0.0038\n","Epoch 89: val_head_1_accuracy did not improve from 1.00000\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m30s\u001b[0m 96ms/step - head_1_accuracy: 0.9999 - head_1_loss: 2.9616e-04 - head_2_accuracy: 0.9999 - head_2_loss: 3.0247e-04 - head_3_accuracy: 0.9999 - head_3_loss: 2.8934e-04 - head_4_accuracy: 0.9999 - head_4_loss: 3.0289e-04 - head_5_accuracy: 0.9999 - head_5_loss: 3.0610e-04 - head_6_accuracy: 0.9999 - head_6_loss: 2.8605e-04 - head_7_accuracy: 0.9999 - head_7_loss: 2.8671e-04 - loss: 0.0021 - val_head_1_accuracy: 1.0000 - val_head_1_loss: 3.2178e-05 - val_head_2_accuracy: 1.0000 - val_head_2_loss: 3.3733e-05 - val_head_3_accuracy: 1.0000 - val_head_3_loss: 3.4002e-05 - val_head_4_accuracy: 1.0000 - val_head_4_loss: 3.5576e-05 - val_head_5_accuracy: 1.0000 - val_head_5_loss: 3.3553e-05 - val_head_6_accuracy: 1.0000 - val_head_6_loss: 3.4017e-05 - val_head_7_accuracy: 1.0000 - val_head_7_loss: 3.3137e-05 - val_loss: 2.3784e-04 - learning_rate: 1.8189e-05\n","\n","Epoch 90: LearningRateScheduler setting learning rate to 1.5076168233603417e-05.\n","Epoch 90/100\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 89ms/step - head_1_accuracy: 1.0000 - head_1_loss: 1.0339e-04 - head_2_accuracy: 1.0000 - head_2_loss: 1.0317e-04 - head_3_accuracy: 1.0000 - head_3_loss: 9.9619e-05 - head_4_accuracy: 1.0000 - head_4_loss: 1.1206e-04 - head_5_accuracy: 1.0000 - head_5_loss: 1.0688e-04 - head_6_accuracy: 1.0000 - head_6_loss: 1.0489e-04 - head_7_accuracy: 1.0000 - head_7_loss: 1.0702e-04 - loss: 7.3704e-04\n","Epoch 90: val_head_1_accuracy did not improve from 1.00000\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m30s\u001b[0m 95ms/step - head_1_accuracy: 1.0000 - head_1_loss: 8.4501e-05 - head_2_accuracy: 1.0000 - head_2_loss: 8.5928e-05 - head_3_accuracy: 1.0000 - head_3_loss: 8.0043e-05 - head_4_accuracy: 1.0000 - head_4_loss: 8.8200e-05 - head_5_accuracy: 1.0000 - head_5_loss: 8.6679e-05 - head_6_accuracy: 1.0000 - head_6_loss: 8.4479e-05 - head_7_accuracy: 1.0000 - head_7_loss: 8.9117e-05 - loss: 5.9930e-04 - val_head_1_accuracy: 1.0000 - val_head_1_loss: 7.2088e-06 - val_head_2_accuracy: 1.0000 - val_head_2_loss: 7.6918e-06 - val_head_3_accuracy: 1.0000 - val_head_3_loss: 6.9244e-06 - val_head_4_accuracy: 1.0000 - val_head_4_loss: 7.6236e-06 - val_head_5_accuracy: 1.0000 - val_head_5_loss: 7.3770e-06 - val_head_6_accuracy: 1.0000 - val_head_6_loss: 7.7198e-06 - val_head_7_accuracy: 1.0000 - val_head_7_loss: 7.4289e-06 - val_loss: 5.2336e-05 - learning_rate: 1.5076e-05\n","\n","Epoch 91: LearningRateScheduler setting learning rate to 1.2250082564331516e-05.\n","Epoch 91/100\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 89ms/step - head_1_accuracy: 1.0000 - head_1_loss: 1.1057e-04 - head_2_accuracy: 1.0000 - head_2_loss: 1.1143e-04 - head_3_accuracy: 1.0000 - head_3_loss: 1.1142e-04 - head_4_accuracy: 1.0000 - head_4_loss: 1.1576e-04 - head_5_accuracy: 1.0000 - head_5_loss: 1.1539e-04 - head_6_accuracy: 1.0000 - head_6_loss: 1.0933e-04 - head_7_accuracy: 1.0000 - head_7_loss: 1.1193e-04 - loss: 7.8583e-04\n","Epoch 91: val_head_1_accuracy did not improve from 1.00000\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m30s\u001b[0m 95ms/step - head_1_accuracy: 1.0000 - head_1_loss: 1.0465e-04 - head_2_accuracy: 1.0000 - head_2_loss: 1.0763e-04 - head_3_accuracy: 1.0000 - head_3_loss: 1.0630e-04 - head_4_accuracy: 1.0000 - head_4_loss: 1.1195e-04 - head_5_accuracy: 1.0000 - head_5_loss: 1.1242e-04 - head_6_accuracy: 1.0000 - head_6_loss: 1.0453e-04 - head_7_accuracy: 1.0000 - head_7_loss: 1.0769e-04 - loss: 7.5446e-04 - val_head_1_accuracy: 0.9999 - val_head_1_loss: 1.0414e-04 - val_head_2_accuracy: 0.9999 - val_head_2_loss: 9.5898e-05 - val_head_3_accuracy: 0.9999 - val_head_3_loss: 1.0146e-04 - val_head_4_accuracy: 0.9999 - val_head_4_loss: 1.0190e-04 - val_head_5_accuracy: 0.9999 - val_head_5_loss: 9.5517e-05 - val_head_6_accuracy: 0.9999 - val_head_6_loss: 8.9359e-05 - val_head_7_accuracy: 0.9999 - val_head_7_loss: 8.9351e-05 - val_loss: 6.8235e-04 - learning_rate: 1.2250e-05\n","\n","Epoch 92: LearningRateScheduler setting learning rate to 9.713787016162943e-06.\n","Epoch 92/100\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 89ms/step - head_1_accuracy: 1.0000 - head_1_loss: 7.9786e-05 - head_2_accuracy: 1.0000 - head_2_loss: 8.9176e-05 - head_3_accuracy: 1.0000 - head_3_loss: 8.1423e-05 - head_4_accuracy: 1.0000 - head_4_loss: 8.4485e-05 - head_5_accuracy: 1.0000 - head_5_loss: 7.8812e-05 - head_6_accuracy: 1.0000 - head_6_loss: 8.0109e-05 - head_7_accuracy: 1.0000 - head_7_loss: 7.9729e-05 - loss: 5.7352e-04\n","Epoch 92: val_head_1_accuracy did not improve from 1.00000\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m30s\u001b[0m 95ms/step - head_1_accuracy: 1.0000 - head_1_loss: 1.2792e-04 - head_2_accuracy: 1.0000 - head_2_loss: 1.4155e-04 - head_3_accuracy: 1.0000 - head_3_loss: 1.3132e-04 - head_4_accuracy: 1.0000 - head_4_loss: 1.2826e-04 - head_5_accuracy: 1.0000 - head_5_loss: 1.2439e-04 - head_6_accuracy: 1.0000 - head_6_loss: 1.2077e-04 - head_7_accuracy: 1.0000 - head_7_loss: 1.1539e-04 - loss: 8.9021e-04 - val_head_1_accuracy: 1.0000 - val_head_1_loss: 1.5237e-05 - val_head_2_accuracy: 1.0000 - val_head_2_loss: 1.6462e-05 - val_head_3_accuracy: 1.0000 - val_head_3_loss: 1.5359e-05 - val_head_4_accuracy: 1.0000 - val_head_4_loss: 1.7595e-05 - val_head_5_accuracy: 1.0000 - val_head_5_loss: 1.5357e-05 - val_head_6_accuracy: 1.0000 - val_head_6_loss: 1.6077e-05 - val_head_7_accuracy: 1.0000 - val_head_7_loss: 1.6105e-05 - val_loss: 1.1297e-04 - learning_rate: 9.7138e-06\n","\n","Epoch 93: LearningRateScheduler setting learning rate to 7.469835423722156e-06.\n","Epoch 93/100\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 89ms/step - head_1_accuracy: 1.0000 - head_1_loss: 7.9534e-05 - head_2_accuracy: 1.0000 - head_2_loss: 7.6264e-05 - head_3_accuracy: 1.0000 - head_3_loss: 7.8095e-05 - head_4_accuracy: 1.0000 - head_4_loss: 7.0508e-05 - head_5_accuracy: 1.0000 - head_5_loss: 7.1797e-05 - head_6_accuracy: 1.0000 - head_6_loss: 7.6241e-05 - head_7_accuracy: 1.0000 - head_7_loss: 7.4089e-05 - loss: 5.2653e-04\n","Epoch 93: val_head_1_accuracy did not improve from 1.00000\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m30s\u001b[0m 95ms/step - head_1_accuracy: 1.0000 - head_1_loss: 6.9385e-05 - head_2_accuracy: 1.0000 - head_2_loss: 6.8590e-05 - head_3_accuracy: 1.0000 - head_3_loss: 6.8017e-05 - head_4_accuracy: 1.0000 - head_4_loss: 6.6017e-05 - head_5_accuracy: 1.0000 - head_5_loss: 6.5178e-05 - head_6_accuracy: 1.0000 - head_6_loss: 6.7122e-05 - head_7_accuracy: 1.0000 - head_7_loss: 6.5536e-05 - loss: 4.7003e-04 - val_head_1_accuracy: 1.0000 - val_head_1_loss: 4.0117e-06 - val_head_2_accuracy: 1.0000 - val_head_2_loss: 4.3659e-06 - val_head_3_accuracy: 1.0000 - val_head_3_loss: 4.0575e-06 - val_head_4_accuracy: 1.0000 - val_head_4_loss: 4.1283e-06 - val_head_5_accuracy: 1.0000 - val_head_5_loss: 4.1111e-06 - val_head_6_accuracy: 1.0000 - val_head_6_loss: 4.3546e-06 - val_head_7_accuracy: 1.0000 - val_head_7_loss: 4.1192e-06 - val_loss: 2.9350e-05 - learning_rate: 7.4698e-06\n","\n","Epoch 94: LearningRateScheduler setting learning rate to 5.520487256051157e-06.\n","Epoch 94/100\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 89ms/step - head_1_accuracy: 1.0000 - head_1_loss: 9.2994e-05 - head_2_accuracy: 1.0000 - head_2_loss: 9.6842e-05 - head_3_accuracy: 1.0000 - head_3_loss: 1.0028e-04 - head_4_accuracy: 1.0000 - head_4_loss: 1.0020e-04 - head_5_accuracy: 1.0000 - head_5_loss: 8.9154e-05 - head_6_accuracy: 1.0000 - head_6_loss: 1.0020e-04 - head_7_accuracy: 1.0000 - head_7_loss: 9.5727e-05 - loss: 6.7540e-04\n","Epoch 94: val_head_1_accuracy did not improve from 1.00000\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m30s\u001b[0m 95ms/step - head_1_accuracy: 1.0000 - head_1_loss: 1.1442e-04 - head_2_accuracy: 1.0000 - head_2_loss: 1.1768e-04 - head_3_accuracy: 1.0000 - head_3_loss: 1.1067e-04 - head_4_accuracy: 1.0000 - head_4_loss: 1.1836e-04 - head_5_accuracy: 1.0000 - head_5_loss: 1.1018e-04 - head_6_accuracy: 1.0000 - head_6_loss: 1.1595e-04 - head_7_accuracy: 1.0000 - head_7_loss: 1.1438e-04 - loss: 8.0219e-04 - val_head_1_accuracy: 1.0000 - val_head_1_loss: 1.4466e-05 - val_head_2_accuracy: 1.0000 - val_head_2_loss: 1.6008e-05 - val_head_3_accuracy: 1.0000 - val_head_3_loss: 1.5609e-05 - val_head_4_accuracy: 1.0000 - val_head_4_loss: 1.6348e-05 - val_head_5_accuracy: 1.0000 - val_head_5_loss: 1.5543e-05 - val_head_6_accuracy: 1.0000 - val_head_6_loss: 1.6061e-05 - val_head_7_accuracy: 1.0000 - val_head_7_loss: 1.5722e-05 - val_loss: 1.1051e-04 - learning_rate: 5.5205e-06\n","\n","Epoch 95: LearningRateScheduler setting learning rate to 3.867705341515533e-06.\n","Epoch 95/100\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 89ms/step - head_1_accuracy: 1.0000 - head_1_loss: 9.0973e-05 - head_2_accuracy: 1.0000 - head_2_loss: 8.6011e-05 - head_3_accuracy: 1.0000 - head_3_loss: 9.2422e-05 - head_4_accuracy: 1.0000 - head_4_loss: 9.1574e-05 - head_5_accuracy: 1.0000 - head_5_loss: 9.5181e-05 - head_6_accuracy: 1.0000 - head_6_loss: 9.4777e-05 - head_7_accuracy: 1.0000 - head_7_loss: 9.0205e-05 - loss: 6.4114e-04\n","Epoch 95: val_head_1_accuracy did not improve from 1.00000\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m30s\u001b[0m 95ms/step - head_1_accuracy: 1.0000 - head_1_loss: 6.8016e-05 - head_2_accuracy: 1.0000 - head_2_loss: 6.6420e-05 - head_3_accuracy: 1.0000 - head_3_loss: 7.0472e-05 - head_4_accuracy: 1.0000 - head_4_loss: 7.0489e-05 - head_5_accuracy: 1.0000 - head_5_loss: 7.3146e-05 - head_6_accuracy: 1.0000 - head_6_loss: 6.8495e-05 - head_7_accuracy: 1.0000 - head_7_loss: 6.8110e-05 - loss: 4.8543e-04 - val_head_1_accuracy: 1.0000 - val_head_1_loss: 3.9671e-06 - val_head_2_accuracy: 1.0000 - val_head_2_loss: 4.2125e-06 - val_head_3_accuracy: 1.0000 - val_head_3_loss: 3.9887e-06 - val_head_4_accuracy: 1.0000 - val_head_4_loss: 4.0388e-06 - val_head_5_accuracy: 1.0000 - val_head_5_loss: 3.9348e-06 - val_head_6_accuracy: 1.0000 - val_head_6_loss: 4.1881e-06 - val_head_7_accuracy: 1.0000 - val_head_7_loss: 3.9927e-06 - val_loss: 2.8520e-05 - learning_rate: 3.8677e-06\n","\n","Epoch 96: LearningRateScheduler setting learning rate to 2.5131538914029374e-06.\n","Epoch 96/100\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 89ms/step - head_1_accuracy: 0.9999 - head_1_loss: 2.1275e-04 - head_2_accuracy: 0.9999 - head_2_loss: 1.7683e-04 - head_3_accuracy: 0.9999 - head_3_loss: 1.7884e-04 - head_4_accuracy: 0.9999 - head_4_loss: 1.8409e-04 - head_5_accuracy: 0.9999 - head_5_loss: 1.7007e-04 - head_6_accuracy: 0.9999 - head_6_loss: 1.7828e-04 - head_7_accuracy: 0.9999 - head_7_loss: 2.0879e-04 - loss: 0.0013\n","Epoch 96: val_head_1_accuracy did not improve from 1.00000\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m30s\u001b[0m 95ms/step - head_1_accuracy: 1.0000 - head_1_loss: 9.5276e-05 - head_2_accuracy: 1.0000 - head_2_loss: 8.5187e-05 - head_3_accuracy: 1.0000 - head_3_loss: 8.5227e-05 - head_4_accuracy: 1.0000 - head_4_loss: 8.6800e-05 - head_5_accuracy: 1.0000 - head_5_loss: 8.2501e-05 - head_6_accuracy: 1.0000 - head_6_loss: 8.4307e-05 - head_7_accuracy: 1.0000 - head_7_loss: 9.0872e-05 - loss: 6.1048e-04 - val_head_1_accuracy: 1.0000 - val_head_1_loss: 5.3944e-06 - val_head_2_accuracy: 1.0000 - val_head_2_loss: 5.8772e-06 - val_head_3_accuracy: 1.0000 - val_head_3_loss: 5.5026e-06 - val_head_4_accuracy: 1.0000 - val_head_4_loss: 5.9825e-06 - val_head_5_accuracy: 1.0000 - val_head_5_loss: 5.3319e-06 - val_head_6_accuracy: 1.0000 - val_head_6_loss: 5.7356e-06 - val_head_7_accuracy: 1.0000 - val_head_7_loss: 5.6150e-06 - val_loss: 3.9714e-05 - learning_rate: 2.5132e-06\n","\n","Epoch 97: LearningRateScheduler setting learning rate to 1.4581968242032775e-06.\n","Epoch 97/100\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 89ms/step - head_1_accuracy: 1.0000 - head_1_loss: 1.5106e-04 - head_2_accuracy: 1.0000 - head_2_loss: 1.5534e-04 - head_3_accuracy: 1.0000 - head_3_loss: 1.5356e-04 - head_4_accuracy: 1.0000 - head_4_loss: 1.5448e-04 - head_5_accuracy: 1.0000 - head_5_loss: 1.6921e-04 - head_6_accuracy: 1.0000 - head_6_loss: 1.6263e-04 - head_7_accuracy: 1.0000 - head_7_loss: 1.5363e-04 - loss: 0.0011\n","Epoch 97: val_head_1_accuracy did not improve from 1.00000\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m30s\u001b[0m 95ms/step - head_1_accuracy: 1.0000 - head_1_loss: 2.1782e-04 - head_2_accuracy: 1.0000 - head_2_loss: 2.3053e-04 - head_3_accuracy: 1.0000 - head_3_loss: 2.3242e-04 - head_4_accuracy: 1.0000 - head_4_loss: 2.3407e-04 - head_5_accuracy: 1.0000 - head_5_loss: 2.3814e-04 - head_6_accuracy: 1.0000 - head_6_loss: 2.4209e-04 - head_7_accuracy: 1.0000 - head_7_loss: 2.3114e-04 - loss: 0.0016 - val_head_1_accuracy: 1.0000 - val_head_1_loss: 4.2882e-06 - val_head_2_accuracy: 1.0000 - val_head_2_loss: 4.3941e-06 - val_head_3_accuracy: 1.0000 - val_head_3_loss: 4.3206e-06 - val_head_4_accuracy: 1.0000 - val_head_4_loss: 4.1495e-06 - val_head_5_accuracy: 1.0000 - val_head_5_loss: 4.2758e-06 - val_head_6_accuracy: 1.0000 - val_head_6_loss: 4.4962e-06 - val_head_7_accuracy: 1.0000 - val_head_7_loss: 4.3183e-06 - val_loss: 3.0453e-05 - learning_rate: 1.4582e-06\n","\n","Epoch 98: LearningRateScheduler setting learning rate to 7.038963922582963e-07.\n","Epoch 98/100\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 89ms/step - head_1_accuracy: 1.0000 - head_1_loss: 1.6225e-04 - head_2_accuracy: 1.0000 - head_2_loss: 1.7097e-04 - head_3_accuracy: 1.0000 - head_3_loss: 1.5932e-04 - head_4_accuracy: 1.0000 - head_4_loss: 1.6166e-04 - head_5_accuracy: 1.0000 - head_5_loss: 1.6820e-04 - head_6_accuracy: 1.0000 - head_6_loss: 1.7110e-04 - head_7_accuracy: 1.0000 - head_7_loss: 1.6266e-04 - loss: 0.0012\n","Epoch 98: val_head_1_accuracy did not improve from 1.00000\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m30s\u001b[0m 95ms/step - head_1_accuracy: 1.0000 - head_1_loss: 1.1367e-04 - head_2_accuracy: 1.0000 - head_2_loss: 1.1608e-04 - head_3_accuracy: 1.0000 - head_3_loss: 1.1067e-04 - head_4_accuracy: 1.0000 - head_4_loss: 1.1259e-04 - head_5_accuracy: 1.0000 - head_5_loss: 1.1501e-04 - head_6_accuracy: 1.0000 - head_6_loss: 1.1676e-04 - head_7_accuracy: 1.0000 - head_7_loss: 1.1297e-04 - loss: 7.9826e-04 - val_head_1_accuracy: 1.0000 - val_head_1_loss: 4.4100e-06 - val_head_2_accuracy: 1.0000 - val_head_2_loss: 4.5343e-06 - val_head_3_accuracy: 1.0000 - val_head_3_loss: 4.4625e-06 - val_head_4_accuracy: 1.0000 - val_head_4_loss: 4.3040e-06 - val_head_5_accuracy: 1.0000 - val_head_5_loss: 4.3709e-06 - val_head_6_accuracy: 1.0000 - val_head_6_loss: 4.6223e-06 - val_head_7_accuracy: 1.0000 - val_head_7_loss: 4.4257e-06 - val_loss: 3.1346e-05 - learning_rate: 7.0390e-07\n","\n","Epoch 99: LearningRateScheduler setting learning rate to 2.510121121636345e-07.\n","Epoch 99/100\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 89ms/step - head_1_accuracy: 1.0000 - head_1_loss: 1.1616e-04 - head_2_accuracy: 1.0000 - head_2_loss: 1.2448e-04 - head_3_accuracy: 1.0000 - head_3_loss: 1.2043e-04 - head_4_accuracy: 1.0000 - head_4_loss: 1.2329e-04 - head_5_accuracy: 1.0000 - head_5_loss: 1.2046e-04 - head_6_accuracy: 1.0000 - head_6_loss: 1.2430e-04 - head_7_accuracy: 1.0000 - head_7_loss: 1.2231e-04 - loss: 8.5144e-04\n","Epoch 99: val_head_1_accuracy did not improve from 1.00000\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m30s\u001b[0m 95ms/step - head_1_accuracy: 1.0000 - head_1_loss: 1.9350e-04 - head_2_accuracy: 1.0000 - head_2_loss: 1.9564e-04 - head_3_accuracy: 1.0000 - head_3_loss: 1.9164e-04 - head_4_accuracy: 1.0000 - head_4_loss: 1.8852e-04 - head_5_accuracy: 1.0000 - head_5_loss: 1.9402e-04 - head_6_accuracy: 1.0000 - head_6_loss: 1.9526e-04 - head_7_accuracy: 0.9999 - head_7_loss: 1.9468e-04 - loss: 0.0014 - val_head_1_accuracy: 1.0000 - val_head_1_loss: 4.3157e-06 - val_head_2_accuracy: 1.0000 - val_head_2_loss: 4.4493e-06 - val_head_3_accuracy: 1.0000 - val_head_3_loss: 4.3851e-06 - val_head_4_accuracy: 1.0000 - val_head_4_loss: 4.2352e-06 - val_head_5_accuracy: 1.0000 - val_head_5_loss: 4.2856e-06 - val_head_6_accuracy: 1.0000 - val_head_6_loss: 4.5413e-06 - val_head_7_accuracy: 1.0000 - val_head_7_loss: 4.3492e-06 - val_loss: 3.0774e-05 - learning_rate: 2.5101e-07\n","\n","Epoch 100: LearningRateScheduler setting learning rate to 1e-07.\n","Epoch 100/100\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 89ms/step - head_1_accuracy: 1.0000 - head_1_loss: 4.8142e-05 - head_2_accuracy: 1.0000 - head_2_loss: 4.9691e-05 - head_3_accuracy: 1.0000 - head_3_loss: 4.6665e-05 - head_4_accuracy: 1.0000 - head_4_loss: 4.8337e-05 - head_5_accuracy: 1.0000 - head_5_loss: 4.8502e-05 - head_6_accuracy: 1.0000 - head_6_loss: 5.0786e-05 - head_7_accuracy: 1.0000 - head_7_loss: 4.7995e-05 - loss: 3.4011e-04\n","Epoch 100: val_head_1_accuracy did not improve from 1.00000\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m30s\u001b[0m 95ms/step - head_1_accuracy: 1.0000 - head_1_loss: 4.3042e-05 - head_2_accuracy: 1.0000 - head_2_loss: 4.4330e-05 - head_3_accuracy: 1.0000 - head_3_loss: 4.3335e-05 - head_4_accuracy: 1.0000 - head_4_loss: 4.2956e-05 - head_5_accuracy: 1.0000 - head_5_loss: 4.3394e-05 - head_6_accuracy: 1.0000 - head_6_loss: 4.6109e-05 - head_7_accuracy: 1.0000 - head_7_loss: 4.3292e-05 - loss: 3.0545e-04 - val_head_1_accuracy: 1.0000 - val_head_1_loss: 4.3265e-06 - val_head_2_accuracy: 1.0000 - val_head_2_loss: 4.4602e-06 - val_head_3_accuracy: 1.0000 - val_head_3_loss: 4.3950e-06 - val_head_4_accuracy: 1.0000 - val_head_4_loss: 4.2466e-06 - val_head_5_accuracy: 1.0000 - val_head_5_loss: 4.2946e-06 - val_head_6_accuracy: 1.0000 - val_head_6_loss: 4.5523e-06 - val_head_7_accuracy: 1.0000 - val_head_7_loss: 4.3586e-06 - val_loss: 3.0847e-05 - learning_rate: 1.0000e-07\n"]},{"name":"stderr","output_type":"stream","text":["/usr/local/lib/python3.12/dist-packages/keras/src/layers/layer.py:424: UserWarning: `build()` was called on layer 'gf__patch_encoder', however the layer does not have a `build()` method implemented and it looks like it has unbuilt state. This will cause the layer to be marked as built, despite not being actually built, which may cause failures down the line. Make sure to implement a proper `build()` method.\n"," warnings.warn(\n","/usr/local/lib/python3.12/dist-packages/keras/src/layers/layer.py:424: UserWarning: `build()` was called on layer 'gf__block', however the layer does not have a `build()` method implemented and it looks like it has unbuilt state. This will cause the layer to be marked as built, despite not being actually built, which may cause failures down the line. Make sure to implement a proper `build()` method.\n"," warnings.warn(\n","/usr/local/lib/python3.12/dist-packages/keras/src/layers/layer.py:424: UserWarning: `build()` was called on layer 'gf__block_1', however the layer does not have a `build()` method implemented and it looks like it has unbuilt state. This will cause the layer to be marked as built, despite not being actually built, which may cause failures down the line. Make sure to implement a proper `build()` method.\n"," warnings.warn(\n","/usr/local/lib/python3.12/dist-packages/keras/src/layers/layer.py:424: UserWarning: `build()` was called on layer 'gf__block_2', however the layer does not have a `build()` method implemented and it looks like it has unbuilt state. This will cause the layer to be marked as built, despite not being actually built, which may cause failures down the line. Make sure to implement a proper `build()` method.\n"," warnings.warn(\n","/usr/local/lib/python3.12/dist-packages/keras/src/layers/layer.py:424: UserWarning: `build()` was called on layer 'gf__block_3', however the layer does not have a `build()` method implemented and it looks like it has unbuilt state. This will cause the layer to be marked as built, despite not being actually built, which may cause failures down the line. Make sure to implement a proper `build()` method.\n"," warnings.warn(\n","/usr/local/lib/python3.12/dist-packages/keras/src/layers/layer.py:424: UserWarning: `build()` was called on layer 'gf__block_4', however the layer does not have a `build()` method implemented and it looks like it has unbuilt state. This will cause the layer to be marked as built, despite not being actually built, which may cause failures down the line. Make sure to implement a proper `build()` method.\n"," warnings.warn(\n","/usr/local/lib/python3.12/dist-packages/keras/src/layers/layer.py:424: UserWarning: `build()` was called on layer 'gf__block_5', however the layer does not have a `build()` method implemented and it looks like it has unbuilt state. This will cause the layer to be marked as built, despite not being actually built, which may cause failures down the line. Make sure to implement a proper `build()` method.\n"," warnings.warn(\n","/usr/local/lib/python3.12/dist-packages/keras/src/layers/layer.py:424: UserWarning: `build()` was called on layer 'gf__block_6', however the layer does not have a `build()` method implemented and it looks like it has unbuilt state. This will cause the layer to be marked as built, despite not being actually built, which may cause failures down the line. Make sure to implement a proper `build()` method.\n"," warnings.warn(\n","/usr/local/lib/python3.12/dist-packages/keras/src/layers/layer.py:424: UserWarning: `build()` was called on layer 'gf__block_7', however the layer does not have a `build()` method implemented and it looks like it has unbuilt state. This will cause the layer to be marked as built, despite not being actually built, which may cause failures down the line. Make sure to implement a proper `build()` method.\n"," warnings.warn(\n","/usr/local/lib/python3.12/dist-packages/keras/src/layers/layer.py:424: UserWarning: `build()` was called on layer 'gf__block_8', however the layer does not have a `build()` method implemented and it looks like it has unbuilt state. This will cause the layer to be marked as built, despite not being actually built, which may cause failures down the line. Make sure to implement a proper `build()` method.\n"," warnings.warn(\n","/usr/local/lib/python3.12/dist-packages/keras/src/layers/layer.py:424: UserWarning: `build()` was called on layer 'gf__block_9', however the layer does not have a `build()` method implemented and it looks like it has unbuilt state. This will cause the layer to be marked as built, despite not being actually built, which may cause failures down the line. Make sure to implement a proper `build()` method.\n"," warnings.warn(\n","/usr/local/lib/python3.12/dist-packages/keras/src/layers/layer.py:424: UserWarning: `build()` was called on layer 'gf__block_10', however the layer does not have a `build()` method implemented and it looks like it has unbuilt state. This will cause the layer to be marked as built, despite not being actually built, which may cause failures down the line. Make sure to implement a proper `build()` method.\n"," warnings.warn(\n","/usr/local/lib/python3.12/dist-packages/keras/src/layers/layer.py:424: UserWarning: `build()` was called on layer 'gf__block_11', however the layer does not have a `build()` method implemented and it looks like it has unbuilt state. This will cause the layer to be marked as built, despite not being actually built, which may cause failures down the line. Make sure to implement a proper `build()` method.\n"," warnings.warn(\n","WARNING:absl:You are saving your model as an HDF5 file via `model.save()` or `keras.saving.save_model(model)`. This file format is considered legacy. We recommend using instead the native Keras format, e.g. `model.save('my_model.keras')` or `keras.saving.save_model(model, 'my_model.keras')`. \n"]},{"name":"stdout","output_type":"stream","text":["Saved .h5 models: GFNet_MultiHead_best.h5, GFNet_MultiHead_final.h5\n"]},{"data":{"image/png":"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\n","text/plain":["
"]},"metadata":{},"output_type":"display_data"},{"name":"stdout","output_type":"stream","text":["Saved training plots directly to /kaggle/working/results\n","\n","============================== Training ViT_UNet_MultiHead ==============================\n"]},{"data":{"image/png":"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\n","text/plain":["
"]},"metadata":{},"output_type":"display_data"},{"data":{"image/png":"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\n","text/plain":["
"]},"metadata":{},"output_type":"display_data"},{"name":"stdout","output_type":"stream","text":["\n","Epoch 1: LearningRateScheduler setting learning rate to 0.0006.\n","Epoch 1/100\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 220ms/step - head_1_accuracy: 0.4676 - head_1_loss: 1.3583 - head_2_accuracy: 0.5018 - head_2_loss: 1.3376 - head_3_accuracy: 0.4799 - head_3_loss: 1.3355 - head_4_accuracy: 0.4797 - head_4_loss: 1.3542 - head_5_accuracy: 0.4885 - head_5_loss: 1.3253 - head_6_accuracy: 0.5056 - head_6_loss: 1.3341 - head_7_accuracy: 0.4979 - head_7_loss: 1.3756 - loss: 9.4205\n","Epoch 1: val_head_1_accuracy improved from None to 0.80273, saving model to /kaggle/working/results/models/ViT_UNet_MultiHead_best.keras\n","\n","Epoch 1: finished saving model to /kaggle/working/results/models/ViT_UNet_MultiHead_best.keras\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m177s\u001b[0m 272ms/step - head_1_accuracy: 0.5937 - head_1_loss: 1.0700 - head_2_accuracy: 0.6073 - head_2_loss: 1.0668 - head_3_accuracy: 0.5973 - head_3_loss: 1.0616 - head_4_accuracy: 0.5985 - head_4_loss: 1.0736 - head_5_accuracy: 0.5980 - head_5_loss: 1.0569 - head_6_accuracy: 0.6123 - head_6_loss: 1.0574 - head_7_accuracy: 0.6045 - head_7_loss: 1.0834 - loss: 7.4724 - val_head_1_accuracy: 0.8027 - val_head_1_loss: 0.5733 - val_head_2_accuracy: 0.8078 - val_head_2_loss: 0.5861 - val_head_3_accuracy: 0.8056 - val_head_3_loss: 0.5694 - val_head_4_accuracy: 0.8038 - val_head_4_loss: 0.5762 - val_head_5_accuracy: 0.8061 - val_head_5_loss: 0.5765 - val_head_6_accuracy: 0.8089 - val_head_6_loss: 0.5682 - val_head_7_accuracy: 0.8038 - val_head_7_loss: 0.5754 - val_loss: 4.0378 - learning_rate: 6.0000e-04\n","\n","Epoch 2: LearningRateScheduler setting learning rate to 0.0005998489878878364.\n","Epoch 2/100\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 79ms/step - head_1_accuracy: 0.7551 - head_1_loss: 0.6997 - head_2_accuracy: 0.7556 - head_2_loss: 0.7030 - head_3_accuracy: 0.7544 - head_3_loss: 0.6967 - head_4_accuracy: 0.7528 - head_4_loss: 0.7057 - head_5_accuracy: 0.7532 - head_5_loss: 0.7008 - head_6_accuracy: 0.7551 - head_6_loss: 0.6953 - head_7_accuracy: 0.7527 - head_7_loss: 0.7065 - loss: 4.9076\n","Epoch 2: val_head_1_accuracy improved from 0.80273 to 0.80993, saving model to /kaggle/working/results/models/ViT_UNet_MultiHead_best.keras\n","\n","Epoch 2: finished saving model to /kaggle/working/results/models/ViT_UNet_MultiHead_best.keras\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m29s\u001b[0m 91ms/step - head_1_accuracy: 0.7388 - head_1_loss: 0.7328 - head_2_accuracy: 0.7391 - head_2_loss: 0.7351 - head_3_accuracy: 0.7400 - head_3_loss: 0.7299 - head_4_accuracy: 0.7396 - head_4_loss: 0.7339 - head_5_accuracy: 0.7400 - head_5_loss: 0.7292 - head_6_accuracy: 0.7395 - head_6_loss: 0.7291 - head_7_accuracy: 0.7381 - head_7_loss: 0.7370 - loss: 5.1278 - val_head_1_accuracy: 0.8099 - val_head_1_loss: 0.5492 - val_head_2_accuracy: 0.8130 - val_head_2_loss: 0.5435 - val_head_3_accuracy: 0.8084 - val_head_3_loss: 0.5413 - val_head_4_accuracy: 0.8109 - val_head_4_loss: 0.5388 - val_head_5_accuracy: 0.8145 - val_head_5_loss: 0.5385 - val_head_6_accuracy: 0.8081 - val_head_6_loss: 0.5425 - val_head_7_accuracy: 0.8108 - val_head_7_loss: 0.5523 - val_loss: 3.8093 - learning_rate: 5.9985e-04\n","\n","Epoch 3: LearningRateScheduler setting learning rate to 0.0005993961036077416.\n","Epoch 3/100\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 79ms/step - head_1_accuracy: 0.8041 - head_1_loss: 0.5630 - head_2_accuracy: 0.8034 - head_2_loss: 0.5603 - head_3_accuracy: 0.8018 - head_3_loss: 0.5597 - head_4_accuracy: 0.8006 - head_4_loss: 0.5622 - head_5_accuracy: 0.8030 - head_5_loss: 0.5606 - head_6_accuracy: 0.8009 - head_6_loss: 0.5606 - head_7_accuracy: 0.8015 - head_7_loss: 0.5659 - loss: 3.9323\n","Epoch 3: val_head_1_accuracy improved from 0.80993 to 0.86129, saving model to /kaggle/working/results/models/ViT_UNet_MultiHead_best.keras\n","\n","Epoch 3: finished saving model to /kaggle/working/results/models/ViT_UNet_MultiHead_best.keras\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m29s\u001b[0m 91ms/step - head_1_accuracy: 0.8108 - head_1_loss: 0.5387 - head_2_accuracy: 0.8103 - head_2_loss: 0.5369 - head_3_accuracy: 0.8095 - head_3_loss: 0.5360 - head_4_accuracy: 0.8081 - head_4_loss: 0.5384 - head_5_accuracy: 0.8092 - head_5_loss: 0.5377 - head_6_accuracy: 0.8096 - head_6_loss: 0.5370 - head_7_accuracy: 0.8091 - head_7_loss: 0.5410 - loss: 3.7662 - val_head_1_accuracy: 0.8613 - val_head_1_loss: 0.4044 - val_head_2_accuracy: 0.8605 - val_head_2_loss: 0.4058 - val_head_3_accuracy: 0.8594 - val_head_3_loss: 0.4057 - val_head_4_accuracy: 0.8608 - val_head_4_loss: 0.4022 - val_head_5_accuracy: 0.8608 - val_head_5_loss: 0.4072 - val_head_6_accuracy: 0.8594 - val_head_6_loss: 0.4057 - val_head_7_accuracy: 0.8623 - val_head_7_loss: 0.4067 - val_loss: 2.8484 - learning_rate: 5.9940e-04\n","\n","Epoch 4: LearningRateScheduler setting learning rate to 0.0005986418031757967.\n","Epoch 4/100\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 79ms/step - head_1_accuracy: 0.8260 - head_1_loss: 0.4959 - head_2_accuracy: 0.8270 - head_2_loss: 0.4951 - head_3_accuracy: 0.8261 - head_3_loss: 0.4933 - head_4_accuracy: 0.8273 - head_4_loss: 0.4967 - head_5_accuracy: 0.8275 - head_5_loss: 0.4942 - head_6_accuracy: 0.8280 - head_6_loss: 0.4928 - head_7_accuracy: 0.8253 - head_7_loss: 0.4975 - loss: 3.4655\n","Epoch 4: val_head_1_accuracy improved from 0.86129 to 0.86455, saving model to /kaggle/working/results/models/ViT_UNet_MultiHead_best.keras\n","\n","Epoch 4: finished saving model to /kaggle/working/results/models/ViT_UNet_MultiHead_best.keras\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m29s\u001b[0m 91ms/step - head_1_accuracy: 0.8301 - head_1_loss: 0.4861 - head_2_accuracy: 0.8297 - head_2_loss: 0.4853 - head_3_accuracy: 0.8289 - head_3_loss: 0.4835 - head_4_accuracy: 0.8306 - head_4_loss: 0.4858 - head_5_accuracy: 0.8306 - head_5_loss: 0.4854 - head_6_accuracy: 0.8311 - head_6_loss: 0.4832 - head_7_accuracy: 0.8295 - head_7_loss: 0.4867 - loss: 3.3960 - val_head_1_accuracy: 0.8646 - val_head_1_loss: 0.4009 - val_head_2_accuracy: 0.8633 - val_head_2_loss: 0.4020 - val_head_3_accuracy: 0.8626 - val_head_3_loss: 0.4025 - val_head_4_accuracy: 0.8673 - val_head_4_loss: 0.3982 - val_head_5_accuracy: 0.8637 - val_head_5_loss: 0.4021 - val_head_6_accuracy: 0.8609 - val_head_6_loss: 0.4042 - val_head_7_accuracy: 0.8633 - val_head_7_loss: 0.4027 - val_loss: 2.8217 - learning_rate: 5.9864e-04\n","\n","Epoch 5: LearningRateScheduler setting learning rate to 0.000597586846108597.\n","Epoch 5/100\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 79ms/step - head_1_accuracy: 0.8368 - head_1_loss: 0.4663 - head_2_accuracy: 0.8363 - head_2_loss: 0.4665 - head_3_accuracy: 0.8364 - head_3_loss: 0.4642 - head_4_accuracy: 0.8374 - head_4_loss: 0.4678 - head_5_accuracy: 0.8375 - head_5_loss: 0.4672 - head_6_accuracy: 0.8372 - head_6_loss: 0.4662 - head_7_accuracy: 0.8356 - head_7_loss: 0.4684 - loss: 3.2665\n","Epoch 5: val_head_1_accuracy did not improve from 0.86455\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m27s\u001b[0m 86ms/step - head_1_accuracy: 0.8366 - head_1_loss: 0.4645 - head_2_accuracy: 0.8361 - head_2_loss: 0.4636 - head_3_accuracy: 0.8364 - head_3_loss: 0.4620 - head_4_accuracy: 0.8367 - head_4_loss: 0.4657 - head_5_accuracy: 0.8366 - head_5_loss: 0.4656 - head_6_accuracy: 0.8368 - head_6_loss: 0.4650 - head_7_accuracy: 0.8351 - head_7_loss: 0.4663 - loss: 3.2530 - val_head_1_accuracy: 0.8542 - val_head_1_loss: 0.3950 - val_head_2_accuracy: 0.8531 - val_head_2_loss: 0.3963 - val_head_3_accuracy: 0.8533 - val_head_3_loss: 0.3965 - val_head_4_accuracy: 0.8542 - val_head_4_loss: 0.3978 - val_head_5_accuracy: 0.8525 - val_head_5_loss: 0.3995 - val_head_6_accuracy: 0.8542 - val_head_6_loss: 0.3993 - val_head_7_accuracy: 0.8536 - val_head_7_loss: 0.3958 - val_loss: 2.7900 - learning_rate: 5.9759e-04\n","\n","Epoch 6: LearningRateScheduler setting learning rate to 0.0005962322946584844.\n","Epoch 6/100\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 79ms/step - head_1_accuracy: 0.8371 - head_1_loss: 0.4636 - head_2_accuracy: 0.8379 - head_2_loss: 0.4623 - head_3_accuracy: 0.8379 - head_3_loss: 0.4608 - head_4_accuracy: 0.8369 - head_4_loss: 0.4649 - head_5_accuracy: 0.8373 - head_5_loss: 0.4618 - head_6_accuracy: 0.8373 - head_6_loss: 0.4632 - head_7_accuracy: 0.8370 - head_7_loss: 0.4624 - loss: 3.2391\n","Epoch 6: val_head_1_accuracy did not improve from 0.86455\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m27s\u001b[0m 85ms/step - head_1_accuracy: 0.8413 - head_1_loss: 0.4501 - head_2_accuracy: 0.8415 - head_2_loss: 0.4495 - head_3_accuracy: 0.8413 - head_3_loss: 0.4483 - head_4_accuracy: 0.8410 - head_4_loss: 0.4518 - head_5_accuracy: 0.8407 - head_5_loss: 0.4491 - head_6_accuracy: 0.8411 - head_6_loss: 0.4495 - head_7_accuracy: 0.8417 - head_7_loss: 0.4497 - loss: 3.1486 - val_head_1_accuracy: 0.8217 - val_head_1_loss: 0.4732 - val_head_2_accuracy: 0.8208 - val_head_2_loss: 0.4654 - val_head_3_accuracy: 0.8199 - val_head_3_loss: 0.4703 - val_head_4_accuracy: 0.8189 - val_head_4_loss: 0.4653 - val_head_5_accuracy: 0.8198 - val_head_5_loss: 0.4680 - val_head_6_accuracy: 0.8209 - val_head_6_loss: 0.4682 - val_head_7_accuracy: 0.8225 - val_head_7_loss: 0.4619 - val_loss: 3.2820 - learning_rate: 5.9623e-04\n","\n","Epoch 7: LearningRateScheduler setting learning rate to 0.0005945795127439488.\n","Epoch 7/100\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 79ms/step - head_1_accuracy: 0.7874 - head_1_loss: 0.6075 - head_2_accuracy: 0.7900 - head_2_loss: 0.6063 - head_3_accuracy: 0.7885 - head_3_loss: 0.6044 - head_4_accuracy: 0.7879 - head_4_loss: 0.6102 - head_5_accuracy: 0.7875 - head_5_loss: 0.6052 - head_6_accuracy: 0.7885 - head_6_loss: 0.6068 - head_7_accuracy: 0.7869 - head_7_loss: 0.6080 - loss: 4.2484\n","Epoch 7: val_head_1_accuracy improved from 0.86455 to 0.86663, saving model to /kaggle/working/results/models/ViT_UNet_MultiHead_best.keras\n","\n","Epoch 7: finished saving model to /kaggle/working/results/models/ViT_UNet_MultiHead_best.keras\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m29s\u001b[0m 91ms/step - head_1_accuracy: 0.8152 - head_1_loss: 0.5286 - head_2_accuracy: 0.8166 - head_2_loss: 0.5284 - head_3_accuracy: 0.8162 - head_3_loss: 0.5276 - head_4_accuracy: 0.8165 - head_4_loss: 0.5308 - head_5_accuracy: 0.8154 - head_5_loss: 0.5276 - head_6_accuracy: 0.8160 - head_6_loss: 0.5284 - head_7_accuracy: 0.8153 - head_7_loss: 0.5291 - loss: 3.7012 - val_head_1_accuracy: 0.8666 - val_head_1_loss: 0.3857 - val_head_2_accuracy: 0.8712 - val_head_2_loss: 0.3835 - val_head_3_accuracy: 0.8689 - val_head_3_loss: 0.3819 - val_head_4_accuracy: 0.8689 - val_head_4_loss: 0.3845 - val_head_5_accuracy: 0.8691 - val_head_5_loss: 0.3852 - val_head_6_accuracy: 0.8697 - val_head_6_loss: 0.3826 - val_head_7_accuracy: 0.8695 - val_head_7_loss: 0.3850 - val_loss: 2.6967 - learning_rate: 5.9458e-04\n","\n","Epoch 8: LearningRateScheduler setting learning rate to 0.0005926301645762779.\n","Epoch 8/100\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 79ms/step - head_1_accuracy: 0.8662 - head_1_loss: 0.3857 - head_2_accuracy: 0.8681 - head_2_loss: 0.3846 - head_3_accuracy: 0.8686 - head_3_loss: 0.3838 - head_4_accuracy: 0.8684 - head_4_loss: 0.3859 - head_5_accuracy: 0.8673 - head_5_loss: 0.3854 - head_6_accuracy: 0.8677 - head_6_loss: 0.3862 - head_7_accuracy: 0.8674 - head_7_loss: 0.3851 - loss: 2.6967\n","Epoch 8: val_head_1_accuracy improved from 0.86663 to 0.89335, saving model to /kaggle/working/results/models/ViT_UNet_MultiHead_best.keras\n","\n","Epoch 8: finished saving model to /kaggle/working/results/models/ViT_UNet_MultiHead_best.keras\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m29s\u001b[0m 91ms/step - head_1_accuracy: 0.8711 - head_1_loss: 0.3711 - head_2_accuracy: 0.8720 - head_2_loss: 0.3699 - head_3_accuracy: 0.8722 - head_3_loss: 0.3696 - head_4_accuracy: 0.8723 - head_4_loss: 0.3715 - head_5_accuracy: 0.8719 - head_5_loss: 0.3703 - head_6_accuracy: 0.8714 - head_6_loss: 0.3715 - head_7_accuracy: 0.8716 - head_7_loss: 0.3699 - loss: 2.5942 - val_head_1_accuracy: 0.8934 - val_head_1_loss: 0.2873 - val_head_2_accuracy: 0.8919 - val_head_2_loss: 0.2874 - val_head_3_accuracy: 0.8890 - val_head_3_loss: 0.2890 - val_head_4_accuracy: 0.8914 - val_head_4_loss: 0.2887 - val_head_5_accuracy: 0.8910 - val_head_5_loss: 0.2873 - val_head_6_accuracy: 0.8923 - val_head_6_loss: 0.2870 - val_head_7_accuracy: 0.8904 - val_head_7_loss: 0.2886 - val_loss: 2.0214 - learning_rate: 5.9263e-04\n","\n","Epoch 9: LearningRateScheduler setting learning rate to 0.0005903862129838371.\n","Epoch 9/100\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 79ms/step - head_1_accuracy: 0.8658 - head_1_loss: 0.3923 - head_2_accuracy: 0.8640 - head_2_loss: 0.3907 - head_3_accuracy: 0.8653 - head_3_loss: 0.3887 - head_4_accuracy: 0.8646 - head_4_loss: 0.3909 - head_5_accuracy: 0.8655 - head_5_loss: 0.3905 - head_6_accuracy: 0.8658 - head_6_loss: 0.3896 - head_7_accuracy: 0.8657 - head_7_loss: 0.3906 - loss: 2.7333\n","Epoch 9: val_head_1_accuracy improved from 0.89335 to 0.90671, saving model to /kaggle/working/results/models/ViT_UNet_MultiHead_best.keras\n","\n","Epoch 9: finished saving model to /kaggle/working/results/models/ViT_UNet_MultiHead_best.keras\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m29s\u001b[0m 91ms/step - head_1_accuracy: 0.8635 - head_1_loss: 0.3969 - head_2_accuracy: 0.8619 - head_2_loss: 0.3953 - head_3_accuracy: 0.8630 - head_3_loss: 0.3939 - head_4_accuracy: 0.8638 - head_4_loss: 0.3953 - head_5_accuracy: 0.8633 - head_5_loss: 0.3955 - head_6_accuracy: 0.8636 - head_6_loss: 0.3944 - head_7_accuracy: 0.8624 - head_7_loss: 0.3951 - loss: 2.7664 - val_head_1_accuracy: 0.9067 - val_head_1_loss: 0.2559 - val_head_2_accuracy: 0.9069 - val_head_2_loss: 0.2555 - val_head_3_accuracy: 0.9066 - val_head_3_loss: 0.2577 - val_head_4_accuracy: 0.9068 - val_head_4_loss: 0.2566 - val_head_5_accuracy: 0.9062 - val_head_5_loss: 0.2584 - val_head_6_accuracy: 0.9080 - val_head_6_loss: 0.2559 - val_head_7_accuracy: 0.9077 - val_head_7_loss: 0.2548 - val_loss: 1.7996 - learning_rate: 5.9039e-04\n","\n","Epoch 10: LearningRateScheduler setting learning rate to 0.0005878499174356684.\n","Epoch 10/100\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 78ms/step - head_1_accuracy: 0.8806 - head_1_loss: 0.3393 - head_2_accuracy: 0.8804 - head_2_loss: 0.3376 - head_3_accuracy: 0.8799 - head_3_loss: 0.3383 - head_4_accuracy: 0.8803 - head_4_loss: 0.3389 - head_5_accuracy: 0.8805 - head_5_loss: 0.3387 - head_6_accuracy: 0.8794 - head_6_loss: 0.3402 - head_7_accuracy: 0.8813 - head_7_loss: 0.3379 - loss: 2.3709\n","Epoch 10: val_head_1_accuracy did not improve from 0.90671\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m27s\u001b[0m 85ms/step - head_1_accuracy: 0.8579 - head_1_loss: 0.4018 - head_2_accuracy: 0.8577 - head_2_loss: 0.4004 - head_3_accuracy: 0.8571 - head_3_loss: 0.4007 - head_4_accuracy: 0.8577 - head_4_loss: 0.4023 - head_5_accuracy: 0.8580 - head_5_loss: 0.4008 - head_6_accuracy: 0.8567 - head_6_loss: 0.4017 - head_7_accuracy: 0.8578 - head_7_loss: 0.4008 - loss: 2.8085 - val_head_1_accuracy: 0.8906 - val_head_1_loss: 0.3023 - val_head_2_accuracy: 0.8917 - val_head_2_loss: 0.3017 - val_head_3_accuracy: 0.8906 - val_head_3_loss: 0.3026 - val_head_4_accuracy: 0.8917 - val_head_4_loss: 0.3030 - val_head_5_accuracy: 0.8911 - val_head_5_loss: 0.3041 - val_head_6_accuracy: 0.8926 - val_head_6_loss: 0.3027 - val_head_7_accuracy: 0.8913 - val_head_7_loss: 0.3035 - val_loss: 2.1272 - learning_rate: 5.8785e-04\n","\n","Epoch 11: LearningRateScheduler setting learning rate to 0.0005850238317663965.\n","Epoch 11/100\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 79ms/step - head_1_accuracy: 0.8694 - head_1_loss: 0.3751 - head_2_accuracy: 0.8693 - head_2_loss: 0.3742 - head_3_accuracy: 0.8699 - head_3_loss: 0.3745 - head_4_accuracy: 0.8698 - head_4_loss: 0.3747 - head_5_accuracy: 0.8701 - head_5_loss: 0.3734 - head_6_accuracy: 0.8711 - head_6_loss: 0.3741 - head_7_accuracy: 0.8692 - head_7_loss: 0.3749 - loss: 2.6210\n","Epoch 11: val_head_1_accuracy did not improve from 0.90671\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m27s\u001b[0m 85ms/step - head_1_accuracy: 0.8725 - head_1_loss: 0.3642 - head_2_accuracy: 0.8718 - head_2_loss: 0.3637 - head_3_accuracy: 0.8723 - head_3_loss: 0.3636 - head_4_accuracy: 0.8724 - head_4_loss: 0.3645 - head_5_accuracy: 0.8717 - head_5_loss: 0.3633 - head_6_accuracy: 0.8729 - head_6_loss: 0.3642 - head_7_accuracy: 0.8722 - head_7_loss: 0.3642 - loss: 2.5477 - val_head_1_accuracy: 0.8940 - val_head_1_loss: 0.2708 - val_head_2_accuracy: 0.8939 - val_head_2_loss: 0.2714 - val_head_3_accuracy: 0.8937 - val_head_3_loss: 0.2704 - val_head_4_accuracy: 0.8937 - val_head_4_loss: 0.2715 - val_head_5_accuracy: 0.8942 - val_head_5_loss: 0.2711 - val_head_6_accuracy: 0.8942 - val_head_6_loss: 0.2693 - val_head_7_accuracy: 0.8943 - val_head_7_loss: 0.2714 - val_loss: 1.8994 - learning_rate: 5.8502e-04\n","\n","Epoch 12: LearningRateScheduler setting learning rate to 0.0005819108016047332.\n","Epoch 12/100\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 79ms/step - head_1_accuracy: 0.8942 - head_1_loss: 0.2974 - head_2_accuracy: 0.8931 - head_2_loss: 0.2985 - head_3_accuracy: 0.8926 - head_3_loss: 0.2975 - head_4_accuracy: 0.8935 - head_4_loss: 0.2989 - head_5_accuracy: 0.8936 - head_5_loss: 0.2970 - head_6_accuracy: 0.8928 - head_6_loss: 0.2991 - head_7_accuracy: 0.8930 - head_7_loss: 0.2976 - loss: 2.0860\n","Epoch 12: val_head_1_accuracy improved from 0.90671 to 0.92385, saving model to /kaggle/working/results/models/ViT_UNet_MultiHead_best.keras\n","\n","Epoch 12: finished saving model to /kaggle/working/results/models/ViT_UNet_MultiHead_best.keras\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m29s\u001b[0m 91ms/step - head_1_accuracy: 0.8967 - head_1_loss: 0.2881 - head_2_accuracy: 0.8959 - head_2_loss: 0.2882 - head_3_accuracy: 0.8956 - head_3_loss: 0.2878 - head_4_accuracy: 0.8963 - head_4_loss: 0.2886 - head_5_accuracy: 0.8965 - head_5_loss: 0.2878 - head_6_accuracy: 0.8964 - head_6_loss: 0.2888 - head_7_accuracy: 0.8958 - head_7_loss: 0.2882 - loss: 2.0181 - val_head_1_accuracy: 0.9239 - val_head_1_loss: 0.1956 - val_head_2_accuracy: 0.9232 - val_head_2_loss: 0.1968 - val_head_3_accuracy: 0.9231 - val_head_3_loss: 0.1970 - val_head_4_accuracy: 0.9239 - val_head_4_loss: 0.1960 - val_head_5_accuracy: 0.9241 - val_head_5_loss: 0.1961 - val_head_6_accuracy: 0.9245 - val_head_6_loss: 0.1957 - val_head_7_accuracy: 0.9231 - val_head_7_loss: 0.1965 - val_loss: 1.3766 - learning_rate: 5.8191e-04\n","\n","Epoch 13: LearningRateScheduler setting learning rate to 0.0005785139615081709.\n","Epoch 13/100\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 79ms/step - head_1_accuracy: 0.9069 - head_1_loss: 0.2615 - head_2_accuracy: 0.9062 - head_2_loss: 0.2608 - head_3_accuracy: 0.9066 - head_3_loss: 0.2607 - head_4_accuracy: 0.9067 - head_4_loss: 0.2605 - head_5_accuracy: 0.9066 - head_5_loss: 0.2613 - head_6_accuracy: 0.9063 - head_6_loss: 0.2611 - head_7_accuracy: 0.9069 - head_7_loss: 0.2608 - loss: 1.8269\n","Epoch 13: val_head_1_accuracy improved from 0.92385 to 0.94122, saving model to /kaggle/working/results/models/ViT_UNet_MultiHead_best.keras\n","\n","Epoch 13: finished saving model to /kaggle/working/results/models/ViT_UNet_MultiHead_best.keras\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m29s\u001b[0m 91ms/step - head_1_accuracy: 0.9065 - head_1_loss: 0.2641 - head_2_accuracy: 0.9063 - head_2_loss: 0.2638 - head_3_accuracy: 0.9062 - head_3_loss: 0.2638 - head_4_accuracy: 0.9065 - head_4_loss: 0.2637 - head_5_accuracy: 0.9065 - head_5_loss: 0.2643 - head_6_accuracy: 0.9061 - head_6_loss: 0.2643 - head_7_accuracy: 0.9063 - head_7_loss: 0.2639 - loss: 1.8484 - val_head_1_accuracy: 0.9412 - val_head_1_loss: 0.1554 - val_head_2_accuracy: 0.9413 - val_head_2_loss: 0.1552 - val_head_3_accuracy: 0.9414 - val_head_3_loss: 0.1557 - val_head_4_accuracy: 0.9423 - val_head_4_loss: 0.1548 - val_head_5_accuracy: 0.9406 - val_head_5_loss: 0.1548 - val_head_6_accuracy: 0.9415 - val_head_6_loss: 0.1548 - val_head_7_accuracy: 0.9426 - val_head_7_loss: 0.1549 - val_loss: 1.0883 - learning_rate: 5.7851e-04\n","\n","Epoch 14: LearningRateScheduler setting learning rate to 0.0005748367318067492.\n","Epoch 14/100\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 80ms/step - head_1_accuracy: 0.8360 - head_1_loss: 0.4745 - head_2_accuracy: 0.8356 - head_2_loss: 0.4727 - head_3_accuracy: 0.8359 - head_3_loss: 0.4714 - head_4_accuracy: 0.8351 - head_4_loss: 0.4749 - head_5_accuracy: 0.8341 - head_5_loss: 0.4735 - head_6_accuracy: 0.8337 - head_6_loss: 0.4765 - head_7_accuracy: 0.8357 - head_7_loss: 0.4725 - loss: 3.3161\n","Epoch 14: val_head_1_accuracy did not improve from 0.94122\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m27s\u001b[0m 86ms/step - head_1_accuracy: 0.7690 - head_1_loss: 0.6704 - head_2_accuracy: 0.7681 - head_2_loss: 0.6700 - head_3_accuracy: 0.7685 - head_3_loss: 0.6679 - head_4_accuracy: 0.7675 - head_4_loss: 0.6719 - head_5_accuracy: 0.7657 - head_5_loss: 0.6704 - head_6_accuracy: 0.7653 - head_6_loss: 0.6743 - head_7_accuracy: 0.7686 - head_7_loss: 0.6698 - loss: 4.6959 - val_head_1_accuracy: 0.8193 - val_head_1_loss: 0.5040 - val_head_2_accuracy: 0.8191 - val_head_2_loss: 0.5065 - val_head_3_accuracy: 0.8199 - val_head_3_loss: 0.5039 - val_head_4_accuracy: 0.8185 - val_head_4_loss: 0.5069 - val_head_5_accuracy: 0.8221 - val_head_5_loss: 0.5030 - val_head_6_accuracy: 0.8214 - val_head_6_loss: 0.5044 - val_head_7_accuracy: 0.8153 - val_head_7_loss: 0.5090 - val_loss: 3.5495 - learning_rate: 5.7484e-04\n","\n","Epoch 15: LearningRateScheduler setting learning rate to 0.0005708828151590721.\n","Epoch 15/100\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 79ms/step - head_1_accuracy: 0.8201 - head_1_loss: 0.5378 - head_2_accuracy: 0.8200 - head_2_loss: 0.5397 - head_3_accuracy: 0.8197 - head_3_loss: 0.5391 - head_4_accuracy: 0.8199 - head_4_loss: 0.5406 - head_5_accuracy: 0.8201 - head_5_loss: 0.5405 - head_6_accuracy: 0.8208 - head_6_loss: 0.5389 - head_7_accuracy: 0.8195 - head_7_loss: 0.5402 - loss: 3.7769\n","Epoch 15: val_head_1_accuracy did not improve from 0.94122\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m27s\u001b[0m 86ms/step - head_1_accuracy: 0.8340 - head_1_loss: 0.4941 - head_2_accuracy: 0.8342 - head_2_loss: 0.4952 - head_3_accuracy: 0.8342 - head_3_loss: 0.4952 - head_4_accuracy: 0.8338 - head_4_loss: 0.4962 - head_5_accuracy: 0.8342 - head_5_loss: 0.4954 - head_6_accuracy: 0.8346 - head_6_loss: 0.4953 - head_7_accuracy: 0.8335 - head_7_loss: 0.4952 - loss: 3.4674 - val_head_1_accuracy: 0.8929 - val_head_1_loss: 0.3016 - val_head_2_accuracy: 0.8928 - val_head_2_loss: 0.3003 - val_head_3_accuracy: 0.8922 - val_head_3_loss: 0.3011 - val_head_4_accuracy: 0.8928 - val_head_4_loss: 0.3017 - val_head_5_accuracy: 0.8925 - val_head_5_loss: 0.3019 - val_head_6_accuracy: 0.8925 - val_head_6_loss: 0.3028 - val_head_7_accuracy: 0.8930 - val_head_7_loss: 0.3021 - val_loss: 2.1191 - learning_rate: 5.7088e-04\n","\n","Epoch 16: LearningRateScheduler setting learning rate to 0.0005666561928240443.\n","Epoch 16/100\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 79ms/step - head_1_accuracy: 0.8754 - head_1_loss: 0.3611 - head_2_accuracy: 0.8753 - head_2_loss: 0.3612 - head_3_accuracy: 0.8753 - head_3_loss: 0.3618 - head_4_accuracy: 0.8743 - head_4_loss: 0.3613 - head_5_accuracy: 0.8750 - head_5_loss: 0.3609 - head_6_accuracy: 0.8753 - head_6_loss: 0.3612 - head_7_accuracy: 0.8745 - head_7_loss: 0.3615 - loss: 2.5290\n","Epoch 16: val_head_1_accuracy did not improve from 0.94122\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m27s\u001b[0m 85ms/step - head_1_accuracy: 0.8772 - head_1_loss: 0.3494 - head_2_accuracy: 0.8770 - head_2_loss: 0.3492 - head_3_accuracy: 0.8774 - head_3_loss: 0.3495 - head_4_accuracy: 0.8764 - head_4_loss: 0.3496 - head_5_accuracy: 0.8770 - head_5_loss: 0.3495 - head_6_accuracy: 0.8769 - head_6_loss: 0.3498 - head_7_accuracy: 0.8772 - head_7_loss: 0.3494 - loss: 2.4468 - val_head_1_accuracy: 0.9177 - val_head_1_loss: 0.2170 - val_head_2_accuracy: 0.9183 - val_head_2_loss: 0.2178 - val_head_3_accuracy: 0.9178 - val_head_3_loss: 0.2168 - val_head_4_accuracy: 0.9174 - val_head_4_loss: 0.2171 - val_head_5_accuracy: 0.9167 - val_head_5_loss: 0.2179 - val_head_6_accuracy: 0.9171 - val_head_6_loss: 0.2169 - val_head_7_accuracy: 0.9167 - val_head_7_loss: 0.2176 - val_loss: 1.5254 - learning_rate: 5.6666e-04\n","\n","Epoch 17: LearningRateScheduler setting learning rate to 0.000562161120652082.\n","Epoch 17/100\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 79ms/step - head_1_accuracy: 0.8800 - head_1_loss: 0.3428 - head_2_accuracy: 0.8800 - head_2_loss: 0.3429 - head_3_accuracy: 0.8807 - head_3_loss: 0.3431 - head_4_accuracy: 0.8802 - head_4_loss: 0.3431 - head_5_accuracy: 0.8789 - head_5_loss: 0.3435 - head_6_accuracy: 0.8806 - head_6_loss: 0.3431 - head_7_accuracy: 0.8798 - head_7_loss: 0.3423 - loss: 2.4008\n","Epoch 17: val_head_1_accuracy did not improve from 0.94122\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m27s\u001b[0m 86ms/step - head_1_accuracy: 0.8706 - head_1_loss: 0.3671 - head_2_accuracy: 0.8708 - head_2_loss: 0.3674 - head_3_accuracy: 0.8709 - head_3_loss: 0.3671 - head_4_accuracy: 0.8706 - head_4_loss: 0.3669 - head_5_accuracy: 0.8701 - head_5_loss: 0.3677 - head_6_accuracy: 0.8711 - head_6_loss: 0.3676 - head_7_accuracy: 0.8705 - head_7_loss: 0.3668 - loss: 2.5711 - val_head_1_accuracy: 0.9261 - val_head_1_loss: 0.2011 - val_head_2_accuracy: 0.9282 - val_head_2_loss: 0.2015 - val_head_3_accuracy: 0.9263 - val_head_3_loss: 0.2013 - val_head_4_accuracy: 0.9277 - val_head_4_loss: 0.2008 - val_head_5_accuracy: 0.9266 - val_head_5_loss: 0.2012 - val_head_6_accuracy: 0.9262 - val_head_6_loss: 0.2015 - val_head_7_accuracy: 0.9276 - val_head_7_loss: 0.2008 - val_loss: 1.4119 - learning_rate: 5.6216e-04\n","\n","Epoch 18: LearningRateScheduler setting learning rate to 0.0005574021247998313.\n","Epoch 18/100\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 79ms/step - head_1_accuracy: 0.8960 - head_1_loss: 0.2970 - head_2_accuracy: 0.8974 - head_2_loss: 0.2971 - head_3_accuracy: 0.8962 - head_3_loss: 0.2969 - head_4_accuracy: 0.8966 - head_4_loss: 0.2974 - head_5_accuracy: 0.8956 - head_5_loss: 0.2969 - head_6_accuracy: 0.8962 - head_6_loss: 0.2976 - head_7_accuracy: 0.8967 - head_7_loss: 0.2960 - loss: 2.0789\n","Epoch 18: val_head_1_accuracy did not improve from 0.94122\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m27s\u001b[0m 86ms/step - head_1_accuracy: 0.8957 - head_1_loss: 0.2946 - head_2_accuracy: 0.8968 - head_2_loss: 0.2938 - head_3_accuracy: 0.8953 - head_3_loss: 0.2940 - head_4_accuracy: 0.8951 - head_4_loss: 0.2945 - head_5_accuracy: 0.8950 - head_5_loss: 0.2942 - head_6_accuracy: 0.8959 - head_6_loss: 0.2942 - head_7_accuracy: 0.8958 - head_7_loss: 0.2938 - loss: 2.0597 - val_head_1_accuracy: 0.9273 - val_head_1_loss: 0.1910 - val_head_2_accuracy: 0.9277 - val_head_2_loss: 0.1913 - val_head_3_accuracy: 0.9266 - val_head_3_loss: 0.1921 - val_head_4_accuracy: 0.9262 - val_head_4_loss: 0.1925 - val_head_5_accuracy: 0.9262 - val_head_5_loss: 0.1927 - val_head_6_accuracy: 0.9272 - val_head_6_loss: 0.1919 - val_head_7_accuracy: 0.9264 - val_head_7_loss: 0.1928 - val_loss: 1.3461 - learning_rate: 5.5740e-04\n","\n","Epoch 19: LearningRateScheduler setting learning rate to 0.0005523839971727127.\n","Epoch 19/100\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 79ms/step - head_1_accuracy: 0.9128 - head_1_loss: 0.2477 - head_2_accuracy: 0.9131 - head_2_loss: 0.2471 - head_3_accuracy: 0.9122 - head_3_loss: 0.2468 - head_4_accuracy: 0.9117 - head_4_loss: 0.2478 - head_5_accuracy: 0.9123 - head_5_loss: 0.2476 - head_6_accuracy: 0.9118 - head_6_loss: 0.2476 - head_7_accuracy: 0.9121 - head_7_loss: 0.2476 - loss: 1.7321\n","Epoch 19: val_head_1_accuracy did not improve from 0.94122\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m27s\u001b[0m 86ms/step - head_1_accuracy: 0.9070 - head_1_loss: 0.2609 - head_2_accuracy: 0.9074 - head_2_loss: 0.2602 - head_3_accuracy: 0.9069 - head_3_loss: 0.2602 - head_4_accuracy: 0.9066 - head_4_loss: 0.2607 - head_5_accuracy: 0.9070 - head_5_loss: 0.2605 - head_6_accuracy: 0.9069 - head_6_loss: 0.2606 - head_7_accuracy: 0.9065 - head_7_loss: 0.2603 - loss: 1.8238 - val_head_1_accuracy: 0.9262 - val_head_1_loss: 0.2052 - val_head_2_accuracy: 0.9258 - val_head_2_loss: 0.2060 - val_head_3_accuracy: 0.9262 - val_head_3_loss: 0.2058 - val_head_4_accuracy: 0.9253 - val_head_4_loss: 0.2054 - val_head_5_accuracy: 0.9258 - val_head_5_loss: 0.2055 - val_head_6_accuracy: 0.9261 - val_head_6_loss: 0.2059 - val_head_7_accuracy: 0.9257 - val_head_7_loss: 0.2053 - val_loss: 1.4435 - learning_rate: 5.5238e-04\n","\n","Epoch 20: LearningRateScheduler setting learning rate to 0.0005471117905998783.\n","Epoch 20/100\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 79ms/step - head_1_accuracy: 0.9166 - head_1_loss: 0.2269 - head_2_accuracy: 0.9171 - head_2_loss: 0.2260 - head_3_accuracy: 0.9175 - head_3_loss: 0.2261 - head_4_accuracy: 0.9166 - head_4_loss: 0.2269 - head_5_accuracy: 0.9171 - head_5_loss: 0.2256 - head_6_accuracy: 0.9165 - head_6_loss: 0.2261 - head_7_accuracy: 0.9170 - head_7_loss: 0.2260 - loss: 1.5835\n","Epoch 20: val_head_1_accuracy did not improve from 0.94122\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m27s\u001b[0m 85ms/step - head_1_accuracy: 0.9155 - head_1_loss: 0.2311 - head_2_accuracy: 0.9159 - head_2_loss: 0.2302 - head_3_accuracy: 0.9158 - head_3_loss: 0.2302 - head_4_accuracy: 0.9154 - head_4_loss: 0.2309 - head_5_accuracy: 0.9157 - head_5_loss: 0.2302 - head_6_accuracy: 0.9156 - head_6_loss: 0.2303 - head_7_accuracy: 0.9158 - head_7_loss: 0.2302 - loss: 1.6135 - val_head_1_accuracy: 0.9322 - val_head_1_loss: 0.1728 - val_head_2_accuracy: 0.9326 - val_head_2_loss: 0.1725 - val_head_3_accuracy: 0.9322 - val_head_3_loss: 0.1714 - val_head_4_accuracy: 0.9321 - val_head_4_loss: 0.1725 - val_head_5_accuracy: 0.9325 - val_head_5_loss: 0.1725 - val_head_6_accuracy: 0.9322 - val_head_6_loss: 0.1725 - val_head_7_accuracy: 0.9327 - val_head_7_loss: 0.1721 - val_loss: 1.2100 - learning_rate: 5.4711e-04\n","\n","Epoch 21: LearningRateScheduler setting learning rate to 0.000541590813746441.\n","Epoch 21/100\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 78ms/step - head_1_accuracy: 0.9212 - head_1_loss: 0.2120 - head_2_accuracy: 0.9211 - head_2_loss: 0.2125 - head_3_accuracy: 0.9214 - head_3_loss: 0.2123 - head_4_accuracy: 0.9213 - head_4_loss: 0.2122 - head_5_accuracy: 0.9210 - head_5_loss: 0.2122 - head_6_accuracy: 0.9219 - head_6_loss: 0.2117 - head_7_accuracy: 0.9205 - head_7_loss: 0.2123 - loss: 1.4852\n","Epoch 21: val_head_1_accuracy did not improve from 0.94122\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m27s\u001b[0m 85ms/step - head_1_accuracy: 0.9249 - head_1_loss: 0.2060 - head_2_accuracy: 0.9253 - head_2_loss: 0.2062 - head_3_accuracy: 0.9245 - head_3_loss: 0.2062 - head_4_accuracy: 0.9245 - head_4_loss: 0.2065 - head_5_accuracy: 0.9247 - head_5_loss: 0.2061 - head_6_accuracy: 0.9252 - head_6_loss: 0.2060 - head_7_accuracy: 0.9245 - head_7_loss: 0.2062 - loss: 1.4435 - val_head_1_accuracy: 0.9408 - val_head_1_loss: 0.1473 - val_head_2_accuracy: 0.9406 - val_head_2_loss: 0.1476 - val_head_3_accuracy: 0.9412 - val_head_3_loss: 0.1467 - val_head_4_accuracy: 0.9415 - val_head_4_loss: 0.1467 - val_head_5_accuracy: 0.9410 - val_head_5_loss: 0.1470 - val_head_6_accuracy: 0.9409 - val_head_6_loss: 0.1474 - val_head_7_accuracy: 0.9409 - val_head_7_loss: 0.1469 - val_loss: 1.0334 - learning_rate: 5.4159e-04\n","\n","Epoch 22: LearningRateScheduler setting learning rate to 0.0005358266257680991.\n","Epoch 22/100\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 79ms/step - head_1_accuracy: 0.8996 - head_1_loss: 0.2862 - head_2_accuracy: 0.8999 - head_2_loss: 0.2865 - head_3_accuracy: 0.8997 - head_3_loss: 0.2860 - head_4_accuracy: 0.8999 - head_4_loss: 0.2866 - head_5_accuracy: 0.8995 - head_5_loss: 0.2867 - head_6_accuracy: 0.8992 - head_6_loss: 0.2859 - head_7_accuracy: 0.9002 - head_7_loss: 0.2858 - loss: 2.0036\n","Epoch 22: val_head_1_accuracy improved from 0.94122 to 0.94515, saving model to /kaggle/working/results/models/ViT_UNet_MultiHead_best.keras\n","\n","Epoch 22: finished saving model to /kaggle/working/results/models/ViT_UNet_MultiHead_best.keras\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m29s\u001b[0m 90ms/step - head_1_accuracy: 0.8995 - head_1_loss: 0.2857 - head_2_accuracy: 0.8997 - head_2_loss: 0.2855 - head_3_accuracy: 0.8999 - head_3_loss: 0.2850 - head_4_accuracy: 0.9000 - head_4_loss: 0.2858 - head_5_accuracy: 0.8997 - head_5_loss: 0.2859 - head_6_accuracy: 0.8995 - head_6_loss: 0.2851 - head_7_accuracy: 0.8997 - head_7_loss: 0.2853 - loss: 1.9989 - val_head_1_accuracy: 0.9452 - val_head_1_loss: 0.1399 - val_head_2_accuracy: 0.9454 - val_head_2_loss: 0.1398 - val_head_3_accuracy: 0.9465 - val_head_3_loss: 0.1390 - val_head_4_accuracy: 0.9463 - val_head_4_loss: 0.1391 - val_head_5_accuracy: 0.9457 - val_head_5_loss: 0.1392 - val_head_6_accuracy: 0.9461 - val_head_6_loss: 0.1397 - val_head_7_accuracy: 0.9457 - val_head_7_loss: 0.1396 - val_loss: 0.9789 - learning_rate: 5.3583e-04\n","\n","Epoch 23: LearningRateScheduler setting learning rate to 0.0005298250307135374.\n","Epoch 23/100\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 79ms/step - head_1_accuracy: 0.9179 - head_1_loss: 0.2347 - head_2_accuracy: 0.9177 - head_2_loss: 0.2340 - head_3_accuracy: 0.9173 - head_3_loss: 0.2344 - head_4_accuracy: 0.9179 - head_4_loss: 0.2343 - head_5_accuracy: 0.9176 - head_5_loss: 0.2345 - head_6_accuracy: 0.9172 - head_6_loss: 0.2345 - head_7_accuracy: 0.9175 - head_7_loss: 0.2340 - loss: 1.6404\n","Epoch 23: val_head_1_accuracy did not improve from 0.94515\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m27s\u001b[0m 85ms/step - head_1_accuracy: 0.9185 - head_1_loss: 0.2285 - head_2_accuracy: 0.9185 - head_2_loss: 0.2281 - head_3_accuracy: 0.9182 - head_3_loss: 0.2286 - head_4_accuracy: 0.9184 - head_4_loss: 0.2286 - head_5_accuracy: 0.9184 - head_5_loss: 0.2286 - head_6_accuracy: 0.9185 - head_6_loss: 0.2285 - head_7_accuracy: 0.9183 - head_7_loss: 0.2283 - loss: 1.5989 - val_head_1_accuracy: 0.9331 - val_head_1_loss: 0.1744 - val_head_2_accuracy: 0.9325 - val_head_2_loss: 0.1737 - val_head_3_accuracy: 0.9330 - val_head_3_loss: 0.1742 - val_head_4_accuracy: 0.9317 - val_head_4_loss: 0.1750 - val_head_5_accuracy: 0.9320 - val_head_5_loss: 0.1742 - val_head_6_accuracy: 0.9323 - val_head_6_loss: 0.1740 - val_head_7_accuracy: 0.9324 - val_head_7_loss: 0.1738 - val_loss: 1.2223 - learning_rate: 5.2983e-04\n","\n","Epoch 24: LearningRateScheduler setting learning rate to 0.0005235920716802426.\n","Epoch 24/100\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 79ms/step - head_1_accuracy: 0.9260 - head_1_loss: 0.2034 - head_2_accuracy: 0.9258 - head_2_loss: 0.2034 - head_3_accuracy: 0.9253 - head_3_loss: 0.2031 - head_4_accuracy: 0.9250 - head_4_loss: 0.2039 - head_5_accuracy: 0.9257 - head_5_loss: 0.2031 - head_6_accuracy: 0.9253 - head_6_loss: 0.2035 - head_7_accuracy: 0.9257 - head_7_loss: 0.2038 - loss: 1.4243\n","Epoch 24: val_head_1_accuracy did not improve from 0.94515\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m27s\u001b[0m 86ms/step - head_1_accuracy: 0.9217 - head_1_loss: 0.2144 - head_2_accuracy: 0.9214 - head_2_loss: 0.2142 - head_3_accuracy: 0.9217 - head_3_loss: 0.2141 - head_4_accuracy: 0.9212 - head_4_loss: 0.2147 - head_5_accuracy: 0.9217 - head_5_loss: 0.2142 - head_6_accuracy: 0.9212 - head_6_loss: 0.2146 - head_7_accuracy: 0.9216 - head_7_loss: 0.2145 - loss: 1.5006 - val_head_1_accuracy: 0.9406 - val_head_1_loss: 0.1569 - val_head_2_accuracy: 0.9407 - val_head_2_loss: 0.1569 - val_head_3_accuracy: 0.9403 - val_head_3_loss: 0.1564 - val_head_4_accuracy: 0.9400 - val_head_4_loss: 0.1567 - val_head_5_accuracy: 0.9397 - val_head_5_loss: 0.1576 - val_head_6_accuracy: 0.9407 - val_head_6_loss: 0.1568 - val_head_7_accuracy: 0.9413 - val_head_7_loss: 0.1565 - val_loss: 1.0994 - learning_rate: 5.2359e-04\n","\n","Epoch 25: LearningRateScheduler setting learning rate to 0.0005171340247296157.\n","Epoch 25/100\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 79ms/step - head_1_accuracy: 0.9235 - head_1_loss: 0.2077 - head_2_accuracy: 0.9228 - head_2_loss: 0.2076 - head_3_accuracy: 0.9226 - head_3_loss: 0.2077 - head_4_accuracy: 0.9231 - head_4_loss: 0.2084 - head_5_accuracy: 0.9229 - head_5_loss: 0.2077 - head_6_accuracy: 0.9227 - head_6_loss: 0.2082 - head_7_accuracy: 0.9235 - head_7_loss: 0.2073 - loss: 1.4545\n","Epoch 25: val_head_1_accuracy did not improve from 0.94515\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m27s\u001b[0m 85ms/step - head_1_accuracy: 0.9165 - head_1_loss: 0.2275 - head_2_accuracy: 0.9165 - head_2_loss: 0.2274 - head_3_accuracy: 0.9161 - head_3_loss: 0.2273 - head_4_accuracy: 0.9163 - head_4_loss: 0.2278 - head_5_accuracy: 0.9161 - head_5_loss: 0.2272 - head_6_accuracy: 0.9159 - head_6_loss: 0.2279 - head_7_accuracy: 0.9166 - head_7_loss: 0.2273 - loss: 1.5916 - val_head_1_accuracy: 0.8884 - val_head_1_loss: 0.2899 - val_head_2_accuracy: 0.8886 - val_head_2_loss: 0.2898 - val_head_3_accuracy: 0.8881 - val_head_3_loss: 0.2897 - val_head_4_accuracy: 0.8883 - val_head_4_loss: 0.2896 - val_head_5_accuracy: 0.8882 - val_head_5_loss: 0.2889 - val_head_6_accuracy: 0.8886 - val_head_6_loss: 0.2921 - val_head_7_accuracy: 0.8886 - val_head_7_loss: 0.2899 - val_loss: 2.0337 - learning_rate: 5.1713e-04\n","\n","Epoch 26: LearningRateScheduler setting learning rate to 0.000510457392567511.\n","Epoch 26/100\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 79ms/step - head_1_accuracy: 0.8309 - head_1_loss: 0.4774 - head_2_accuracy: 0.8296 - head_2_loss: 0.4770 - head_3_accuracy: 0.8309 - head_3_loss: 0.4777 - head_4_accuracy: 0.8301 - head_4_loss: 0.4772 - head_5_accuracy: 0.8299 - head_5_loss: 0.4775 - head_6_accuracy: 0.8294 - head_6_loss: 0.4781 - head_7_accuracy: 0.8303 - head_7_loss: 0.4770 - loss: 3.3417\n","Epoch 26: val_head_1_accuracy did not improve from 0.94515\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m27s\u001b[0m 85ms/step - head_1_accuracy: 0.8702 - head_1_loss: 0.3641 - head_2_accuracy: 0.8695 - head_2_loss: 0.3640 - head_3_accuracy: 0.8702 - head_3_loss: 0.3642 - head_4_accuracy: 0.8702 - head_4_loss: 0.3642 - head_5_accuracy: 0.8696 - head_5_loss: 0.3641 - head_6_accuracy: 0.8694 - head_6_loss: 0.3644 - head_7_accuracy: 0.8697 - head_7_loss: 0.3641 - loss: 2.5498 - val_head_1_accuracy: 0.9176 - val_head_1_loss: 0.2000 - val_head_2_accuracy: 0.9165 - val_head_2_loss: 0.2014 - val_head_3_accuracy: 0.9160 - val_head_3_loss: 0.2018 - val_head_4_accuracy: 0.9165 - val_head_4_loss: 0.2019 - val_head_5_accuracy: 0.9171 - val_head_5_loss: 0.2013 - val_head_6_accuracy: 0.9175 - val_head_6_loss: 0.2016 - val_head_7_accuracy: 0.9168 - val_head_7_loss: 0.2003 - val_loss: 1.4120 - learning_rate: 5.1046e-04\n","\n","Epoch 27: LearningRateScheduler setting learning rate to 0.0005035688979965618.\n","Epoch 27/100\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 79ms/step - head_1_accuracy: 0.9200 - head_1_loss: 0.2170 - head_2_accuracy: 0.9203 - head_2_loss: 0.2168 - head_3_accuracy: 0.9204 - head_3_loss: 0.2171 - head_4_accuracy: 0.9205 - head_4_loss: 0.2172 - head_5_accuracy: 0.9211 - head_5_loss: 0.2171 - head_6_accuracy: 0.9203 - head_6_loss: 0.2174 - head_7_accuracy: 0.9207 - head_7_loss: 0.2171 - loss: 1.5197\n","Epoch 27: val_head_1_accuracy did not improve from 0.94515\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m27s\u001b[0m 86ms/step - head_1_accuracy: 0.9220 - head_1_loss: 0.2118 - head_2_accuracy: 0.9223 - head_2_loss: 0.2115 - head_3_accuracy: 0.9218 - head_3_loss: 0.2116 - head_4_accuracy: 0.9222 - head_4_loss: 0.2117 - head_5_accuracy: 0.9221 - head_5_loss: 0.2117 - head_6_accuracy: 0.9221 - head_6_loss: 0.2118 - head_7_accuracy: 0.9223 - head_7_loss: 0.2116 - loss: 1.4818 - val_head_1_accuracy: 0.9313 - val_head_1_loss: 0.1730 - val_head_2_accuracy: 0.9319 - val_head_2_loss: 0.1730 - val_head_3_accuracy: 0.9312 - val_head_3_loss: 0.1726 - val_head_4_accuracy: 0.9314 - val_head_4_loss: 0.1729 - val_head_5_accuracy: 0.9317 - val_head_5_loss: 0.1730 - val_head_6_accuracy: 0.9322 - val_head_6_loss: 0.1728 - val_head_7_accuracy: 0.9319 - val_head_7_loss: 0.1725 - val_loss: 1.2131 - learning_rate: 5.0357e-04\n","\n","Epoch 28: LearningRateScheduler setting learning rate to 0.0004964754771468882.\n","Epoch 28/100\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 79ms/step - head_1_accuracy: 0.9209 - head_1_loss: 0.2187 - head_2_accuracy: 0.9218 - head_2_loss: 0.2187 - head_3_accuracy: 0.9213 - head_3_loss: 0.2181 - head_4_accuracy: 0.9214 - head_4_loss: 0.2185 - head_5_accuracy: 0.9214 - head_5_loss: 0.2188 - head_6_accuracy: 0.9214 - head_6_loss: 0.2185 - head_7_accuracy: 0.9218 - head_7_loss: 0.2184 - loss: 1.5298\n","Epoch 28: val_head_1_accuracy did not improve from 0.94515\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m27s\u001b[0m 86ms/step - head_1_accuracy: 0.9107 - head_1_loss: 0.2511 - head_2_accuracy: 0.9114 - head_2_loss: 0.2512 - head_3_accuracy: 0.9107 - head_3_loss: 0.2509 - head_4_accuracy: 0.9109 - head_4_loss: 0.2514 - head_5_accuracy: 0.9111 - head_5_loss: 0.2512 - head_6_accuracy: 0.9109 - head_6_loss: 0.2513 - head_7_accuracy: 0.9114 - head_7_loss: 0.2513 - loss: 1.7585 - val_head_1_accuracy: 0.9383 - val_head_1_loss: 0.1605 - val_head_2_accuracy: 0.9383 - val_head_2_loss: 0.1607 - val_head_3_accuracy: 0.9377 - val_head_3_loss: 0.1604 - val_head_4_accuracy: 0.9380 - val_head_4_loss: 0.1604 - val_head_5_accuracy: 0.9380 - val_head_5_loss: 0.1607 - val_head_6_accuracy: 0.9385 - val_head_6_loss: 0.1607 - val_head_7_accuracy: 0.9385 - val_head_7_loss: 0.1607 - val_loss: 1.1256 - learning_rate: 4.9648e-04\n","\n","Epoch 29: LearningRateScheduler setting learning rate to 0.0004891842724920025.\n","Epoch 29/100\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 79ms/step - head_1_accuracy: 0.9185 - head_1_loss: 0.2270 - head_2_accuracy: 0.9188 - head_2_loss: 0.2270 - head_3_accuracy: 0.9184 - head_3_loss: 0.2270 - head_4_accuracy: 0.9189 - head_4_loss: 0.2268 - head_5_accuracy: 0.9187 - head_5_loss: 0.2267 - head_6_accuracy: 0.9187 - head_6_loss: 0.2276 - head_7_accuracy: 0.9186 - head_7_loss: 0.2268 - loss: 1.5888\n","Epoch 29: val_head_1_accuracy did not improve from 0.94515\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m27s\u001b[0m 86ms/step - head_1_accuracy: 0.9123 - head_1_loss: 0.2417 - head_2_accuracy: 0.9124 - head_2_loss: 0.2417 - head_3_accuracy: 0.9125 - head_3_loss: 0.2417 - head_4_accuracy: 0.9126 - head_4_loss: 0.2417 - head_5_accuracy: 0.9120 - head_5_loss: 0.2414 - head_6_accuracy: 0.9127 - head_6_loss: 0.2419 - head_7_accuracy: 0.9122 - head_7_loss: 0.2416 - loss: 1.6922 - val_head_1_accuracy: 0.9414 - val_head_1_loss: 0.1505 - val_head_2_accuracy: 0.9423 - val_head_2_loss: 0.1504 - val_head_3_accuracy: 0.9420 - val_head_3_loss: 0.1501 - val_head_4_accuracy: 0.9418 - val_head_4_loss: 0.1500 - val_head_5_accuracy: 0.9419 - val_head_5_loss: 0.1505 - val_head_6_accuracy: 0.9427 - val_head_6_loss: 0.1502 - val_head_7_accuracy: 0.9429 - val_head_7_loss: 0.1502 - val_loss: 1.0545 - learning_rate: 4.8918e-04\n","\n","Epoch 30: LearningRateScheduler setting learning rate to 0.0004817026256569431.\n","Epoch 30/100\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 79ms/step - head_1_accuracy: 0.9236 - head_1_loss: 0.2147 - head_2_accuracy: 0.9234 - head_2_loss: 0.2144 - head_3_accuracy: 0.9231 - head_3_loss: 0.2142 - head_4_accuracy: 0.9230 - head_4_loss: 0.2145 - head_5_accuracy: 0.9236 - head_5_loss: 0.2142 - head_6_accuracy: 0.9230 - head_6_loss: 0.2139 - head_7_accuracy: 0.9231 - head_7_loss: 0.2138 - loss: 1.4997\n","Epoch 30: val_head_1_accuracy did not improve from 0.94515\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m27s\u001b[0m 85ms/step - head_1_accuracy: 0.9206 - head_1_loss: 0.2223 - head_2_accuracy: 0.9207 - head_2_loss: 0.2221 - head_3_accuracy: 0.9205 - head_3_loss: 0.2222 - head_4_accuracy: 0.9199 - head_4_loss: 0.2225 - head_5_accuracy: 0.9205 - head_5_loss: 0.2223 - head_6_accuracy: 0.9203 - head_6_loss: 0.2220 - head_7_accuracy: 0.9202 - head_7_loss: 0.2220 - loss: 1.5552 - val_head_1_accuracy: 0.9149 - val_head_1_loss: 0.2162 - val_head_2_accuracy: 0.9150 - val_head_2_loss: 0.2175 - val_head_3_accuracy: 0.9147 - val_head_3_loss: 0.2175 - val_head_4_accuracy: 0.9145 - val_head_4_loss: 0.2178 - val_head_5_accuracy: 0.9147 - val_head_5_loss: 0.2177 - val_head_6_accuracy: 0.9153 - val_head_6_loss: 0.2177 - val_head_7_accuracy: 0.9147 - val_head_7_loss: 0.2173 - val_loss: 1.5205 - learning_rate: 4.8170e-04\n","\n","Epoch 31: LearningRateScheduler setting learning rate to 0.0004740380700258809.\n","Epoch 31/100\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 79ms/step - head_1_accuracy: 0.8914 - head_1_loss: 0.2996 - head_2_accuracy: 0.8909 - head_2_loss: 0.2985 - head_3_accuracy: 0.8910 - head_3_loss: 0.2995 - head_4_accuracy: 0.8916 - head_4_loss: 0.2988 - head_5_accuracy: 0.8913 - head_5_loss: 0.2990 - head_6_accuracy: 0.8908 - head_6_loss: 0.2989 - head_7_accuracy: 0.8912 - head_7_loss: 0.2988 - loss: 2.0931\n","Epoch 31: val_head_1_accuracy did not improve from 0.94515\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m27s\u001b[0m 86ms/step - head_1_accuracy: 0.9053 - head_1_loss: 0.2626 - head_2_accuracy: 0.9048 - head_2_loss: 0.2617 - head_3_accuracy: 0.9048 - head_3_loss: 0.2622 - head_4_accuracy: 0.9050 - head_4_loss: 0.2617 - head_5_accuracy: 0.9052 - head_5_loss: 0.2622 - head_6_accuracy: 0.9047 - head_6_loss: 0.2620 - head_7_accuracy: 0.9049 - head_7_loss: 0.2619 - loss: 1.8350 - val_head_1_accuracy: 0.9377 - val_head_1_loss: 0.1607 - val_head_2_accuracy: 0.9385 - val_head_2_loss: 0.1607 - val_head_3_accuracy: 0.9377 - val_head_3_loss: 0.1607 - val_head_4_accuracy: 0.9374 - val_head_4_loss: 0.1608 - val_head_5_accuracy: 0.9380 - val_head_5_loss: 0.1612 - val_head_6_accuracy: 0.9383 - val_head_6_loss: 0.1605 - val_head_7_accuracy: 0.9381 - val_head_7_loss: 0.1608 - val_loss: 1.1271 - learning_rate: 4.7404e-04\n","\n","Epoch 32: LearningRateScheduler setting learning rate to 0.0004661983231566398.\n","Epoch 32/100\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 79ms/step - head_1_accuracy: 0.9330 - head_1_loss: 0.1811 - head_2_accuracy: 0.9342 - head_2_loss: 0.1814 - head_3_accuracy: 0.9329 - head_3_loss: 0.1812 - head_4_accuracy: 0.9331 - head_4_loss: 0.1814 - head_5_accuracy: 0.9335 - head_5_loss: 0.1813 - head_6_accuracy: 0.9333 - head_6_loss: 0.1813 - head_7_accuracy: 0.9335 - head_7_loss: 0.1814 - loss: 1.2692\n","Epoch 32: val_head_1_accuracy improved from 0.94515 to 0.94938, saving model to /kaggle/working/results/models/ViT_UNet_MultiHead_best.keras\n","\n","Epoch 32: finished saving model to /kaggle/working/results/models/ViT_UNet_MultiHead_best.keras\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m29s\u001b[0m 91ms/step - head_1_accuracy: 0.9278 - head_1_loss: 0.1996 - head_2_accuracy: 0.9283 - head_2_loss: 0.1997 - head_3_accuracy: 0.9278 - head_3_loss: 0.1997 - head_4_accuracy: 0.9279 - head_4_loss: 0.1998 - head_5_accuracy: 0.9278 - head_5_loss: 0.1997 - head_6_accuracy: 0.9277 - head_6_loss: 0.1997 - head_7_accuracy: 0.9279 - head_7_loss: 0.1995 - loss: 1.3978 - val_head_1_accuracy: 0.9494 - val_head_1_loss: 0.1330 - val_head_2_accuracy: 0.9500 - val_head_2_loss: 0.1331 - val_head_3_accuracy: 0.9493 - val_head_3_loss: 0.1331 - val_head_4_accuracy: 0.9497 - val_head_4_loss: 0.1332 - val_head_5_accuracy: 0.9497 - val_head_5_loss: 0.1330 - val_head_6_accuracy: 0.9498 - val_head_6_loss: 0.1329 - val_head_7_accuracy: 0.9496 - val_head_7_loss: 0.1331 - val_loss: 0.9335 - learning_rate: 4.6620e-04\n","\n","Epoch 33: LearningRateScheduler setting learning rate to 0.0004581912790097702.\n","Epoch 33/100\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 79ms/step - head_1_accuracy: 0.9350 - head_1_loss: 0.1774 - head_2_accuracy: 0.9341 - head_2_loss: 0.1772 - head_3_accuracy: 0.9352 - head_3_loss: 0.1770 - head_4_accuracy: 0.9349 - head_4_loss: 0.1765 - head_5_accuracy: 0.9348 - head_5_loss: 0.1769 - head_6_accuracy: 0.9347 - head_6_loss: 0.1768 - head_7_accuracy: 0.9352 - head_7_loss: 0.1769 - loss: 1.2387\n","Epoch 33: val_head_1_accuracy improved from 0.94938 to 0.95413, saving model to /kaggle/working/results/models/ViT_UNet_MultiHead_best.keras\n","\n","Epoch 33: finished saving model to /kaggle/working/results/models/ViT_UNet_MultiHead_best.keras\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m29s\u001b[0m 91ms/step - head_1_accuracy: 0.9339 - head_1_loss: 0.1804 - head_2_accuracy: 0.9338 - head_2_loss: 0.1802 - head_3_accuracy: 0.9342 - head_3_loss: 0.1802 - head_4_accuracy: 0.9341 - head_4_loss: 0.1799 - head_5_accuracy: 0.9339 - head_5_loss: 0.1801 - head_6_accuracy: 0.9337 - head_6_loss: 0.1801 - head_7_accuracy: 0.9341 - head_7_loss: 0.1802 - loss: 1.2614 - val_head_1_accuracy: 0.9541 - val_head_1_loss: 0.1151 - val_head_2_accuracy: 0.9541 - val_head_2_loss: 0.1150 - val_head_3_accuracy: 0.9540 - val_head_3_loss: 0.1148 - val_head_4_accuracy: 0.9540 - val_head_4_loss: 0.1147 - val_head_5_accuracy: 0.9544 - val_head_5_loss: 0.1148 - val_head_6_accuracy: 0.9541 - val_head_6_loss: 0.1149 - val_head_7_accuracy: 0.9540 - val_head_7_loss: 0.1149 - val_loss: 0.8065 - learning_rate: 4.5819e-04\n","\n","Epoch 34: LearningRateScheduler setting learning rate to 0.000450025.\n","Epoch 34/100\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 79ms/step - head_1_accuracy: 0.9427 - head_1_loss: 0.1551 - head_2_accuracy: 0.9421 - head_2_loss: 0.1552 - head_3_accuracy: 0.9421 - head_3_loss: 0.1551 - head_4_accuracy: 0.9423 - head_4_loss: 0.1550 - head_5_accuracy: 0.9422 - head_5_loss: 0.1549 - head_6_accuracy: 0.9424 - head_6_loss: 0.1551 - head_7_accuracy: 0.9422 - head_7_loss: 0.1549 - loss: 1.0853\n","Epoch 34: val_head_1_accuracy did not improve from 0.95413\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m27s\u001b[0m 85ms/step - head_1_accuracy: 0.9439 - head_1_loss: 0.1513 - head_2_accuracy: 0.9434 - head_2_loss: 0.1513 - head_3_accuracy: 0.9438 - head_3_loss: 0.1511 - head_4_accuracy: 0.9437 - head_4_loss: 0.1511 - head_5_accuracy: 0.9438 - head_5_loss: 0.1510 - head_6_accuracy: 0.9440 - head_6_loss: 0.1513 - head_7_accuracy: 0.9438 - head_7_loss: 0.1510 - loss: 1.0582 - val_head_1_accuracy: 0.9521 - val_head_1_loss: 0.1154 - val_head_2_accuracy: 0.9524 - val_head_2_loss: 0.1157 - val_head_3_accuracy: 0.9522 - val_head_3_loss: 0.1154 - val_head_4_accuracy: 0.9522 - val_head_4_loss: 0.1156 - val_head_5_accuracy: 0.9522 - val_head_5_loss: 0.1155 - val_head_6_accuracy: 0.9526 - val_head_6_loss: 0.1152 - val_head_7_accuracy: 0.9523 - val_head_7_loss: 0.1155 - val_loss: 0.8070 - learning_rate: 4.5002e-04\n","\n","Epoch 35: LearningRateScheduler setting learning rate to 0.0004417077088780662.\n","Epoch 35/100\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 78ms/step - head_1_accuracy: 0.9431 - head_1_loss: 0.1545 - head_2_accuracy: 0.9433 - head_2_loss: 0.1541 - head_3_accuracy: 0.9432 - head_3_loss: 0.1541 - head_4_accuracy: 0.9432 - head_4_loss: 0.1551 - head_5_accuracy: 0.9433 - head_5_loss: 0.1544 - head_6_accuracy: 0.9433 - head_6_loss: 0.1543 - head_7_accuracy: 0.9432 - head_7_loss: 0.1545 - loss: 1.0809\n","Epoch 35: val_head_1_accuracy improved from 0.95413 to 0.95955, saving model to /kaggle/working/results/models/ViT_UNet_MultiHead_best.keras\n","\n","Epoch 35: finished saving model to /kaggle/working/results/models/ViT_UNet_MultiHead_best.keras\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m28s\u001b[0m 90ms/step - head_1_accuracy: 0.9395 - head_1_loss: 0.1699 - head_2_accuracy: 0.9397 - head_2_loss: 0.1698 - head_3_accuracy: 0.9395 - head_3_loss: 0.1698 - head_4_accuracy: 0.9397 - head_4_loss: 0.1701 - head_5_accuracy: 0.9396 - head_5_loss: 0.1698 - head_6_accuracy: 0.9398 - head_6_loss: 0.1698 - head_7_accuracy: 0.9395 - head_7_loss: 0.1699 - loss: 1.1890 - val_head_1_accuracy: 0.9596 - val_head_1_loss: 0.1080 - val_head_2_accuracy: 0.9597 - val_head_2_loss: 0.1085 - val_head_3_accuracy: 0.9596 - val_head_3_loss: 0.1084 - val_head_4_accuracy: 0.9598 - val_head_4_loss: 0.1081 - val_head_5_accuracy: 0.9598 - val_head_5_loss: 0.1081 - val_head_6_accuracy: 0.9599 - val_head_6_loss: 0.1083 - val_head_7_accuracy: 0.9597 - val_head_7_loss: 0.1082 - val_loss: 0.7590 - learning_rate: 4.4171e-04\n","\n","Epoch 36: LearningRateScheduler setting learning rate to 0.00043324778045110196.\n","Epoch 36/100\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 79ms/step - head_1_accuracy: 0.8920 - head_1_loss: 0.3441 - head_2_accuracy: 0.8922 - head_2_loss: 0.3434 - head_3_accuracy: 0.8922 - head_3_loss: 0.3438 - head_4_accuracy: 0.8916 - head_4_loss: 0.3447 - head_5_accuracy: 0.8919 - head_5_loss: 0.3434 - head_6_accuracy: 0.8924 - head_6_loss: 0.3436 - head_7_accuracy: 0.8920 - head_7_loss: 0.3438 - loss: 2.4067\n","Epoch 36: val_head_1_accuracy did not improve from 0.95955\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m27s\u001b[0m 86ms/step - head_1_accuracy: 0.9179 - head_1_loss: 0.2439 - head_2_accuracy: 0.9179 - head_2_loss: 0.2437 - head_3_accuracy: 0.9181 - head_3_loss: 0.2438 - head_4_accuracy: 0.9179 - head_4_loss: 0.2441 - head_5_accuracy: 0.9179 - head_5_loss: 0.2437 - head_6_accuracy: 0.9182 - head_6_loss: 0.2436 - head_7_accuracy: 0.9178 - head_7_loss: 0.2437 - loss: 1.7071 - val_head_1_accuracy: 0.9452 - val_head_1_loss: 0.1412 - val_head_2_accuracy: 0.9454 - val_head_2_loss: 0.1414 - val_head_3_accuracy: 0.9452 - val_head_3_loss: 0.1414 - val_head_4_accuracy: 0.9455 - val_head_4_loss: 0.1415 - val_head_5_accuracy: 0.9454 - val_head_5_loss: 0.1409 - val_head_6_accuracy: 0.9457 - val_head_6_loss: 0.1412 - val_head_7_accuracy: 0.9456 - val_head_7_loss: 0.1410 - val_loss: 0.9897 - learning_rate: 4.3325e-04\n","\n","Epoch 37: LearningRateScheduler setting learning rate to 0.0004246537331499159.\n","Epoch 37/100\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 79ms/step - head_1_accuracy: 0.9470 - head_1_loss: 0.1466 - head_2_accuracy: 0.9470 - head_2_loss: 0.1467 - head_3_accuracy: 0.9468 - head_3_loss: 0.1465 - head_4_accuracy: 0.9470 - head_4_loss: 0.1467 - head_5_accuracy: 0.9469 - head_5_loss: 0.1464 - head_6_accuracy: 0.9466 - head_6_loss: 0.1465 - head_7_accuracy: 0.9474 - head_7_loss: 0.1463 - loss: 1.0257\n","Epoch 37: val_head_1_accuracy did not improve from 0.95955\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m27s\u001b[0m 86ms/step - head_1_accuracy: 0.9476 - head_1_loss: 0.1440 - head_2_accuracy: 0.9477 - head_2_loss: 0.1438 - head_3_accuracy: 0.9475 - head_3_loss: 0.1438 - head_4_accuracy: 0.9476 - head_4_loss: 0.1439 - head_5_accuracy: 0.9476 - head_5_loss: 0.1437 - head_6_accuracy: 0.9475 - head_6_loss: 0.1436 - head_7_accuracy: 0.9477 - head_7_loss: 0.1438 - loss: 1.0067 - val_head_1_accuracy: 0.9595 - val_head_1_loss: 0.1033 - val_head_2_accuracy: 0.9593 - val_head_2_loss: 0.1038 - val_head_3_accuracy: 0.9595 - val_head_3_loss: 0.1036 - val_head_4_accuracy: 0.9593 - val_head_4_loss: 0.1034 - val_head_5_accuracy: 0.9592 - val_head_5_loss: 0.1036 - val_head_6_accuracy: 0.9594 - val_head_6_loss: 0.1036 - val_head_7_accuracy: 0.9594 - val_head_7_loss: 0.1035 - val_loss: 0.7257 - learning_rate: 4.2465e-04\n","\n","Epoch 38: LearningRateScheduler setting learning rate to 0.0004159342204516539.\n","Epoch 38/100\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 79ms/step - head_1_accuracy: 0.9508 - head_1_loss: 0.1280 - head_2_accuracy: 0.9506 - head_2_loss: 0.1283 - head_3_accuracy: 0.9509 - head_3_loss: 0.1282 - head_4_accuracy: 0.9510 - head_4_loss: 0.1278 - head_5_accuracy: 0.9508 - head_5_loss: 0.1279 - head_6_accuracy: 0.9512 - head_6_loss: 0.1281 - head_7_accuracy: 0.9508 - head_7_loss: 0.1280 - loss: 0.8964\n","Epoch 38: val_head_1_accuracy improved from 0.95955 to 0.95963, saving model to /kaggle/working/results/models/ViT_UNet_MultiHead_best.keras\n","\n","Epoch 38: finished saving model to /kaggle/working/results/models/ViT_UNet_MultiHead_best.keras\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m29s\u001b[0m 91ms/step - head_1_accuracy: 0.9512 - head_1_loss: 0.1291 - head_2_accuracy: 0.9513 - head_2_loss: 0.1292 - head_3_accuracy: 0.9513 - head_3_loss: 0.1291 - head_4_accuracy: 0.9514 - head_4_loss: 0.1290 - head_5_accuracy: 0.9514 - head_5_loss: 0.1289 - head_6_accuracy: 0.9515 - head_6_loss: 0.1292 - head_7_accuracy: 0.9512 - head_7_loss: 0.1291 - loss: 0.9038 - val_head_1_accuracy: 0.9596 - val_head_1_loss: 0.1013 - val_head_2_accuracy: 0.9601 - val_head_2_loss: 0.1011 - val_head_3_accuracy: 0.9599 - val_head_3_loss: 0.1012 - val_head_4_accuracy: 0.9605 - val_head_4_loss: 0.1011 - val_head_5_accuracy: 0.9599 - val_head_5_loss: 0.1012 - val_head_6_accuracy: 0.9598 - val_head_6_loss: 0.1012 - val_head_7_accuracy: 0.9598 - val_head_7_loss: 0.1013 - val_loss: 0.7049 - learning_rate: 4.1593e-04\n","\n","Epoch 39: LearningRateScheduler setting learning rate to 0.0004070980221664819.\n","Epoch 39/100\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 79ms/step - head_1_accuracy: 0.9498 - head_1_loss: 0.1353 - head_2_accuracy: 0.9498 - head_2_loss: 0.1353 - head_3_accuracy: 0.9499 - head_3_loss: 0.1351 - head_4_accuracy: 0.9499 - head_4_loss: 0.1352 - head_5_accuracy: 0.9501 - head_5_loss: 0.1348 - head_6_accuracy: 0.9499 - head_6_loss: 0.1349 - head_7_accuracy: 0.9498 - head_7_loss: 0.1350 - loss: 0.9455\n","Epoch 39: val_head_1_accuracy did not improve from 0.95963\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m27s\u001b[0m 86ms/step - head_1_accuracy: 0.9453 - head_1_loss: 0.1491 - head_2_accuracy: 0.9452 - head_2_loss: 0.1492 - head_3_accuracy: 0.9455 - head_3_loss: 0.1491 - head_4_accuracy: 0.9454 - head_4_loss: 0.1492 - head_5_accuracy: 0.9456 - head_5_loss: 0.1489 - head_6_accuracy: 0.9455 - head_6_loss: 0.1491 - head_7_accuracy: 0.9454 - head_7_loss: 0.1490 - loss: 1.0439 - val_head_1_accuracy: 0.9578 - val_head_1_loss: 0.1113 - val_head_2_accuracy: 0.9584 - val_head_2_loss: 0.1116 - val_head_3_accuracy: 0.9586 - val_head_3_loss: 0.1115 - val_head_4_accuracy: 0.9584 - val_head_4_loss: 0.1111 - val_head_5_accuracy: 0.9582 - val_head_5_loss: 0.1112 - val_head_6_accuracy: 0.9581 - val_head_6_loss: 0.1113 - val_head_7_accuracy: 0.9584 - val_head_7_loss: 0.1115 - val_loss: 0.7809 - learning_rate: 4.0710e-04\n","\n","Epoch 40: LearningRateScheduler setting learning rate to 0.0003981540355970607.\n","Epoch 40/100\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 78ms/step - head_1_accuracy: 0.9375 - head_1_loss: 0.1817 - head_2_accuracy: 0.9368 - head_2_loss: 0.1821 - head_3_accuracy: 0.9373 - head_3_loss: 0.1821 - head_4_accuracy: 0.9373 - head_4_loss: 0.1820 - head_5_accuracy: 0.9376 - head_5_loss: 0.1822 - head_6_accuracy: 0.9373 - head_6_loss: 0.1820 - head_7_accuracy: 0.9372 - head_7_loss: 0.1820 - loss: 1.2741\n","Epoch 40: val_head_1_accuracy did not improve from 0.95963\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m27s\u001b[0m 85ms/step - head_1_accuracy: 0.9276 - head_1_loss: 0.2142 - head_2_accuracy: 0.9273 - head_2_loss: 0.2144 - head_3_accuracy: 0.9275 - head_3_loss: 0.2144 - head_4_accuracy: 0.9273 - head_4_loss: 0.2143 - head_5_accuracy: 0.9276 - head_5_loss: 0.2145 - head_6_accuracy: 0.9273 - head_6_loss: 0.2143 - head_7_accuracy: 0.9272 - head_7_loss: 0.2143 - loss: 1.5006 - val_head_1_accuracy: 0.9468 - val_head_1_loss: 0.1357 - val_head_2_accuracy: 0.9464 - val_head_2_loss: 0.1358 - val_head_3_accuracy: 0.9464 - val_head_3_loss: 0.1355 - val_head_4_accuracy: 0.9469 - val_head_4_loss: 0.1353 - val_head_5_accuracy: 0.9466 - val_head_5_loss: 0.1356 - val_head_6_accuracy: 0.9469 - val_head_6_loss: 0.1358 - val_head_7_accuracy: 0.9466 - val_head_7_loss: 0.1357 - val_loss: 0.9499 - learning_rate: 3.9815e-04\n","\n","Epoch 41: LearningRateScheduler setting learning rate to 0.0003891112665797161.\n","Epoch 41/100\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 79ms/step - head_1_accuracy: 0.9405 - head_1_loss: 0.1613 - head_2_accuracy: 0.9400 - head_2_loss: 0.1616 - head_3_accuracy: 0.9399 - head_3_loss: 0.1616 - head_4_accuracy: 0.9406 - head_4_loss: 0.1616 - head_5_accuracy: 0.9402 - head_5_loss: 0.1613 - head_6_accuracy: 0.9401 - head_6_loss: 0.1617 - head_7_accuracy: 0.9404 - head_7_loss: 0.1615 - loss: 1.1307\n","Epoch 41: val_head_1_accuracy did not improve from 0.95963\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m27s\u001b[0m 86ms/step - head_1_accuracy: 0.9457 - head_1_loss: 0.1469 - head_2_accuracy: 0.9456 - head_2_loss: 0.1470 - head_3_accuracy: 0.9455 - head_3_loss: 0.1472 - head_4_accuracy: 0.9456 - head_4_loss: 0.1472 - head_5_accuracy: 0.9456 - head_5_loss: 0.1470 - head_6_accuracy: 0.9455 - head_6_loss: 0.1471 - head_7_accuracy: 0.9458 - head_7_loss: 0.1471 - loss: 1.0295 - val_head_1_accuracy: 0.9363 - val_head_1_loss: 0.1693 - val_head_2_accuracy: 0.9364 - val_head_2_loss: 0.1680 - val_head_3_accuracy: 0.9362 - val_head_3_loss: 0.1688 - val_head_4_accuracy: 0.9360 - val_head_4_loss: 0.1686 - val_head_5_accuracy: 0.9360 - val_head_5_loss: 0.1696 - val_head_6_accuracy: 0.9363 - val_head_6_loss: 0.1684 - val_head_7_accuracy: 0.9361 - val_head_7_loss: 0.1685 - val_loss: 1.1832 - learning_rate: 3.8911e-04\n","\n","Epoch 42: LearningRateScheduler setting learning rate to 0.000379978820416326.\n","Epoch 42/100\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 79ms/step - head_1_accuracy: 0.9512 - head_1_loss: 0.1352 - head_2_accuracy: 0.9509 - head_2_loss: 0.1352 - head_3_accuracy: 0.9514 - head_3_loss: 0.1349 - head_4_accuracy: 0.9509 - head_4_loss: 0.1351 - head_5_accuracy: 0.9510 - head_5_loss: 0.1350 - head_6_accuracy: 0.9508 - head_6_loss: 0.1354 - head_7_accuracy: 0.9516 - head_7_loss: 0.1350 - loss: 0.9458\n","Epoch 42: val_head_1_accuracy did not improve from 0.95963\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m27s\u001b[0m 85ms/step - head_1_accuracy: 0.9532 - head_1_loss: 0.1264 - head_2_accuracy: 0.9531 - head_2_loss: 0.1263 - head_3_accuracy: 0.9535 - head_3_loss: 0.1262 - head_4_accuracy: 0.9531 - head_4_loss: 0.1264 - head_5_accuracy: 0.9532 - head_5_loss: 0.1262 - head_6_accuracy: 0.9531 - head_6_loss: 0.1264 - head_7_accuracy: 0.9536 - head_7_loss: 0.1261 - loss: 0.8843 - val_head_1_accuracy: 0.9491 - val_head_1_loss: 0.1264 - val_head_2_accuracy: 0.9488 - val_head_2_loss: 0.1264 - val_head_3_accuracy: 0.9489 - val_head_3_loss: 0.1265 - val_head_4_accuracy: 0.9494 - val_head_4_loss: 0.1264 - val_head_5_accuracy: 0.9492 - val_head_5_loss: 0.1264 - val_head_6_accuracy: 0.9489 - val_head_6_loss: 0.1263 - val_head_7_accuracy: 0.9490 - val_head_7_loss: 0.1262 - val_loss: 0.8796 - learning_rate: 3.7998e-04\n","\n","Epoch 43: LearningRateScheduler setting learning rate to 0.00037076589270605275.\n","Epoch 43/100\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 79ms/step - head_1_accuracy: 0.9514 - head_1_loss: 0.1305 - head_2_accuracy: 0.9517 - head_2_loss: 0.1305 - head_3_accuracy: 0.9515 - head_3_loss: 0.1308 - head_4_accuracy: 0.9513 - head_4_loss: 0.1308 - head_5_accuracy: 0.9518 - head_5_loss: 0.1307 - head_6_accuracy: 0.9517 - head_6_loss: 0.1305 - head_7_accuracy: 0.9512 - head_7_loss: 0.1305 - loss: 0.9144\n","Epoch 43: val_head_1_accuracy improved from 0.95963 to 0.96163, saving model to /kaggle/working/results/models/ViT_UNet_MultiHead_best.keras\n","\n","Epoch 43: finished saving model to /kaggle/working/results/models/ViT_UNet_MultiHead_best.keras\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m29s\u001b[0m 90ms/step - head_1_accuracy: 0.9508 - head_1_loss: 0.1309 - head_2_accuracy: 0.9510 - head_2_loss: 0.1309 - head_3_accuracy: 0.9507 - head_3_loss: 0.1310 - head_4_accuracy: 0.9508 - head_4_loss: 0.1311 - head_5_accuracy: 0.9511 - head_5_loss: 0.1310 - head_6_accuracy: 0.9512 - head_6_loss: 0.1309 - head_7_accuracy: 0.9508 - head_7_loss: 0.1308 - loss: 0.9168 - val_head_1_accuracy: 0.9616 - val_head_1_loss: 0.1012 - val_head_2_accuracy: 0.9617 - val_head_2_loss: 0.1009 - val_head_3_accuracy: 0.9613 - val_head_3_loss: 0.1011 - val_head_4_accuracy: 0.9617 - val_head_4_loss: 0.1008 - val_head_5_accuracy: 0.9617 - val_head_5_loss: 0.1010 - val_head_6_accuracy: 0.9619 - val_head_6_loss: 0.1007 - val_head_7_accuracy: 0.9616 - val_head_7_loss: 0.1009 - val_loss: 0.7076 - learning_rate: 3.7077e-04\n","\n","Epoch 44: LearningRateScheduler setting learning rate to 0.000361481760086154.\n","Epoch 44/100\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 78ms/step - head_1_accuracy: 0.9523 - head_1_loss: 0.1350 - head_2_accuracy: 0.9519 - head_2_loss: 0.1347 - head_3_accuracy: 0.9520 - head_3_loss: 0.1348 - head_4_accuracy: 0.9519 - head_4_loss: 0.1351 - head_5_accuracy: 0.9523 - head_5_loss: 0.1347 - head_6_accuracy: 0.9521 - head_6_loss: 0.1347 - head_7_accuracy: 0.9519 - head_7_loss: 0.1347 - loss: 0.9437\n","Epoch 44: val_head_1_accuracy improved from 0.96163 to 0.96742, saving model to /kaggle/working/results/models/ViT_UNet_MultiHead_best.keras\n","\n","Epoch 44: finished saving model to /kaggle/working/results/models/ViT_UNet_MultiHead_best.keras\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m28s\u001b[0m 90ms/step - head_1_accuracy: 0.9474 - head_1_loss: 0.1476 - head_2_accuracy: 0.9472 - head_2_loss: 0.1475 - head_3_accuracy: 0.9471 - head_3_loss: 0.1474 - head_4_accuracy: 0.9469 - head_4_loss: 0.1476 - head_5_accuracy: 0.9473 - head_5_loss: 0.1473 - head_6_accuracy: 0.9473 - head_6_loss: 0.1473 - head_7_accuracy: 0.9473 - head_7_loss: 0.1474 - loss: 1.0324 - val_head_1_accuracy: 0.9674 - val_head_1_loss: 0.0834 - val_head_2_accuracy: 0.9674 - val_head_2_loss: 0.0833 - val_head_3_accuracy: 0.9676 - val_head_3_loss: 0.0831 - val_head_4_accuracy: 0.9672 - val_head_4_loss: 0.0830 - val_head_5_accuracy: 0.9673 - val_head_5_loss: 0.0834 - val_head_6_accuracy: 0.9676 - val_head_6_loss: 0.0834 - val_head_7_accuracy: 0.9673 - val_head_7_loss: 0.0832 - val_loss: 0.5821 - learning_rate: 3.6148e-04\n","\n","Epoch 45: LearningRateScheduler setting learning rate to 0.0003521357708911958.\n","Epoch 45/100\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 79ms/step - head_1_accuracy: 0.9570 - head_1_loss: 0.1146 - head_2_accuracy: 0.9577 - head_2_loss: 0.1146 - head_3_accuracy: 0.9574 - head_3_loss: 0.1146 - head_4_accuracy: 0.9572 - head_4_loss: 0.1144 - head_5_accuracy: 0.9574 - head_5_loss: 0.1144 - head_6_accuracy: 0.9572 - head_6_loss: 0.1145 - head_7_accuracy: 0.9571 - head_7_loss: 0.1142 - loss: 0.8014\n","Epoch 45: val_head_1_accuracy did not improve from 0.96742\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m27s\u001b[0m 85ms/step - head_1_accuracy: 0.9592 - head_1_loss: 0.1092 - head_2_accuracy: 0.9595 - head_2_loss: 0.1092 - head_3_accuracy: 0.9593 - head_3_loss: 0.1092 - head_4_accuracy: 0.9591 - head_4_loss: 0.1091 - head_5_accuracy: 0.9593 - head_5_loss: 0.1090 - head_6_accuracy: 0.9591 - head_6_loss: 0.1092 - head_7_accuracy: 0.9593 - head_7_loss: 0.1090 - loss: 0.7641 - val_head_1_accuracy: 0.9650 - val_head_1_loss: 0.0857 - val_head_2_accuracy: 0.9651 - val_head_2_loss: 0.0855 - val_head_3_accuracy: 0.9653 - val_head_3_loss: 0.0855 - val_head_4_accuracy: 0.9653 - val_head_4_loss: 0.0856 - val_head_5_accuracy: 0.9652 - val_head_5_loss: 0.0856 - val_head_6_accuracy: 0.9651 - val_head_6_loss: 0.0855 - val_head_7_accuracy: 0.9650 - val_head_7_loss: 0.0855 - val_loss: 0.5969 - learning_rate: 3.5214e-04\n","\n","Epoch 46: LearningRateScheduler setting learning rate to 0.00034273733574007183.\n","Epoch 46/100\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 79ms/step - head_1_accuracy: 0.9640 - head_1_loss: 0.0963 - head_2_accuracy: 0.9640 - head_2_loss: 0.0965 - head_3_accuracy: 0.9643 - head_3_loss: 0.0966 - head_4_accuracy: 0.9638 - head_4_loss: 0.0967 - head_5_accuracy: 0.9645 - head_5_loss: 0.0965 - head_6_accuracy: 0.9638 - head_6_loss: 0.0964 - head_7_accuracy: 0.9639 - head_7_loss: 0.0963 - loss: 0.6753\n","Epoch 46: val_head_1_accuracy did not improve from 0.96742\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m27s\u001b[0m 85ms/step - head_1_accuracy: 0.9638 - head_1_loss: 0.0957 - head_2_accuracy: 0.9636 - head_2_loss: 0.0959 - head_3_accuracy: 0.9639 - head_3_loss: 0.0960 - head_4_accuracy: 0.9637 - head_4_loss: 0.0960 - head_5_accuracy: 0.9642 - head_5_loss: 0.0958 - head_6_accuracy: 0.9637 - head_6_loss: 0.0957 - head_7_accuracy: 0.9637 - head_7_loss: 0.0958 - loss: 0.6710 - val_head_1_accuracy: 0.9391 - val_head_1_loss: 0.1840 - val_head_2_accuracy: 0.9394 - val_head_2_loss: 0.1823 - val_head_3_accuracy: 0.9386 - val_head_3_loss: 0.1841 - val_head_4_accuracy: 0.9388 - val_head_4_loss: 0.1840 - val_head_5_accuracy: 0.9388 - val_head_5_loss: 0.1846 - val_head_6_accuracy: 0.9391 - val_head_6_loss: 0.1848 - val_head_7_accuracy: 0.9385 - val_head_7_loss: 0.1831 - val_loss: 1.2904 - learning_rate: 3.4274e-04\n","\n","Epoch 47: LearningRateScheduler setting learning rate to 0.0003332959180603083.\n","Epoch 47/100\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 79ms/step - head_1_accuracy: 0.9524 - head_1_loss: 0.1360 - head_2_accuracy: 0.9520 - head_2_loss: 0.1360 - head_3_accuracy: 0.9521 - head_3_loss: 0.1363 - head_4_accuracy: 0.9525 - head_4_loss: 0.1364 - head_5_accuracy: 0.9526 - head_5_loss: 0.1361 - head_6_accuracy: 0.9523 - head_6_loss: 0.1362 - head_7_accuracy: 0.9524 - head_7_loss: 0.1359 - loss: 0.9530\n","Epoch 47: val_head_1_accuracy did not improve from 0.96742\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m27s\u001b[0m 86ms/step - head_1_accuracy: 0.9461 - head_1_loss: 0.1597 - head_2_accuracy: 0.9460 - head_2_loss: 0.1598 - head_3_accuracy: 0.9460 - head_3_loss: 0.1599 - head_4_accuracy: 0.9462 - head_4_loss: 0.1600 - head_5_accuracy: 0.9464 - head_5_loss: 0.1598 - head_6_accuracy: 0.9462 - head_6_loss: 0.1598 - head_7_accuracy: 0.9463 - head_7_loss: 0.1597 - loss: 1.1188 - val_head_1_accuracy: 0.9632 - val_head_1_loss: 0.0981 - val_head_2_accuracy: 0.9633 - val_head_2_loss: 0.0982 - val_head_3_accuracy: 0.9631 - val_head_3_loss: 0.0982 - val_head_4_accuracy: 0.9636 - val_head_4_loss: 0.0978 - val_head_5_accuracy: 0.9633 - val_head_5_loss: 0.0982 - val_head_6_accuracy: 0.9633 - val_head_6_loss: 0.0980 - val_head_7_accuracy: 0.9636 - val_head_7_loss: 0.0981 - val_loss: 0.6881 - learning_rate: 3.3330e-04\n","\n","Epoch 48: LearningRateScheduler setting learning rate to 0.00032382102455919367.\n","Epoch 48/100\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 79ms/step - head_1_accuracy: 0.9501 - head_1_loss: 0.1446 - head_2_accuracy: 0.9503 - head_2_loss: 0.1443 - head_3_accuracy: 0.9507 - head_3_loss: 0.1444 - head_4_accuracy: 0.9504 - head_4_loss: 0.1448 - head_5_accuracy: 0.9504 - head_5_loss: 0.1447 - head_6_accuracy: 0.9502 - head_6_loss: 0.1445 - head_7_accuracy: 0.9503 - head_7_loss: 0.1448 - loss: 1.0120\n","Epoch 48: val_head_1_accuracy did not improve from 0.96742\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m27s\u001b[0m 85ms/step - head_1_accuracy: 0.9557 - head_1_loss: 0.1270 - head_2_accuracy: 0.9557 - head_2_loss: 0.1268 - head_3_accuracy: 0.9559 - head_3_loss: 0.1268 - head_4_accuracy: 0.9558 - head_4_loss: 0.1270 - head_5_accuracy: 0.9559 - head_5_loss: 0.1269 - head_6_accuracy: 0.9556 - head_6_loss: 0.1268 - head_7_accuracy: 0.9559 - head_7_loss: 0.1270 - loss: 0.8886 - val_head_1_accuracy: 0.9578 - val_head_1_loss: 0.1071 - val_head_2_accuracy: 0.9578 - val_head_2_loss: 0.1068 - val_head_3_accuracy: 0.9576 - val_head_3_loss: 0.1069 - val_head_4_accuracy: 0.9576 - val_head_4_loss: 0.1072 - val_head_5_accuracy: 0.9570 - val_head_5_loss: 0.1076 - val_head_6_accuracy: 0.9577 - val_head_6_loss: 0.1067 - val_head_7_accuracy: 0.9575 - val_head_7_loss: 0.1069 - val_loss: 0.7502 - learning_rate: 3.2382e-04\n","\n","Epoch 49: LearningRateScheduler setting learning rate to 0.0003143221956513316.\n","Epoch 49/100\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 79ms/step - head_1_accuracy: 0.9652 - head_1_loss: 0.0960 - head_2_accuracy: 0.9654 - head_2_loss: 0.0957 - head_3_accuracy: 0.9654 - head_3_loss: 0.0956 - head_4_accuracy: 0.9651 - head_4_loss: 0.0956 - head_5_accuracy: 0.9650 - head_5_loss: 0.0956 - head_6_accuracy: 0.9650 - head_6_loss: 0.0956 - head_7_accuracy: 0.9651 - head_7_loss: 0.0954 - loss: 0.6695\n","Epoch 49: val_head_1_accuracy did not improve from 0.96742\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m27s\u001b[0m 85ms/step - head_1_accuracy: 0.9656 - head_1_loss: 0.0949 - head_2_accuracy: 0.9655 - head_2_loss: 0.0947 - head_3_accuracy: 0.9655 - head_3_loss: 0.0946 - head_4_accuracy: 0.9654 - head_4_loss: 0.0946 - head_5_accuracy: 0.9653 - head_5_loss: 0.0946 - head_6_accuracy: 0.9653 - head_6_loss: 0.0946 - head_7_accuracy: 0.9654 - head_7_loss: 0.0946 - loss: 0.6628 - val_head_1_accuracy: 0.9628 - val_head_1_loss: 0.0965 - val_head_2_accuracy: 0.9629 - val_head_2_loss: 0.0968 - val_head_3_accuracy: 0.9627 - val_head_3_loss: 0.0966 - val_head_4_accuracy: 0.9624 - val_head_4_loss: 0.0965 - val_head_5_accuracy: 0.9624 - val_head_5_loss: 0.0971 - val_head_6_accuracy: 0.9630 - val_head_6_loss: 0.0964 - val_head_7_accuracy: 0.9628 - val_head_7_loss: 0.0966 - val_loss: 0.6783 - learning_rate: 3.1432e-04\n","\n","Epoch 50: LearningRateScheduler setting learning rate to 0.0003048089958522507.\n","Epoch 50/100\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 79ms/step - head_1_accuracy: 0.9650 - head_1_loss: 0.0955 - head_2_accuracy: 0.9650 - head_2_loss: 0.0955 - head_3_accuracy: 0.9652 - head_3_loss: 0.0954 - head_4_accuracy: 0.9647 - head_4_loss: 0.0955 - head_5_accuracy: 0.9648 - head_5_loss: 0.0956 - head_6_accuracy: 0.9653 - head_6_loss: 0.0953 - head_7_accuracy: 0.9649 - head_7_loss: 0.0955 - loss: 0.6682\n","Epoch 50: val_head_1_accuracy improved from 0.96742 to 0.98041, saving model to /kaggle/working/results/models/ViT_UNet_MultiHead_best.keras\n","\n","Epoch 50: finished saving model to /kaggle/working/results/models/ViT_UNet_MultiHead_best.keras\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m28s\u001b[0m 90ms/step - head_1_accuracy: 0.9654 - head_1_loss: 0.0936 - head_2_accuracy: 0.9654 - head_2_loss: 0.0937 - head_3_accuracy: 0.9654 - head_3_loss: 0.0936 - head_4_accuracy: 0.9654 - head_4_loss: 0.0936 - head_5_accuracy: 0.9654 - head_5_loss: 0.0936 - head_6_accuracy: 0.9656 - head_6_loss: 0.0934 - head_7_accuracy: 0.9655 - head_7_loss: 0.0935 - loss: 0.6552 - val_head_1_accuracy: 0.9804 - val_head_1_loss: 0.0512 - val_head_2_accuracy: 0.9810 - val_head_2_loss: 0.0511 - val_head_3_accuracy: 0.9808 - val_head_3_loss: 0.0510 - val_head_4_accuracy: 0.9810 - val_head_4_loss: 0.0508 - val_head_5_accuracy: 0.9806 - val_head_5_loss: 0.0511 - val_head_6_accuracy: 0.9809 - val_head_6_loss: 0.0510 - val_head_7_accuracy: 0.9806 - val_head_7_loss: 0.0511 - val_loss: 0.3552 - learning_rate: 3.0481e-04\n","\n","Epoch 51: LearningRateScheduler setting learning rate to 0.00029529100414774935.\n","Epoch 51/100\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 79ms/step - head_1_accuracy: 0.9673 - head_1_loss: 0.0900 - head_2_accuracy: 0.9669 - head_2_loss: 0.0900 - head_3_accuracy: 0.9671 - head_3_loss: 0.0899 - head_4_accuracy: 0.9670 - head_4_loss: 0.0899 - head_5_accuracy: 0.9669 - head_5_loss: 0.0898 - head_6_accuracy: 0.9672 - head_6_loss: 0.0900 - head_7_accuracy: 0.9671 - head_7_loss: 0.0900 - loss: 0.6295\n","Epoch 51: val_head_1_accuracy did not improve from 0.98041\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m27s\u001b[0m 85ms/step - head_1_accuracy: 0.9637 - head_1_loss: 0.0991 - head_2_accuracy: 0.9633 - head_2_loss: 0.0992 - head_3_accuracy: 0.9636 - head_3_loss: 0.0991 - head_4_accuracy: 0.9634 - head_4_loss: 0.0993 - head_5_accuracy: 0.9633 - head_5_loss: 0.0992 - head_6_accuracy: 0.9633 - head_6_loss: 0.0992 - head_7_accuracy: 0.9635 - head_7_loss: 0.0992 - loss: 0.6943 - val_head_1_accuracy: 0.9651 - val_head_1_loss: 0.0923 - val_head_2_accuracy: 0.9650 - val_head_2_loss: 0.0923 - val_head_3_accuracy: 0.9651 - val_head_3_loss: 0.0920 - val_head_4_accuracy: 0.9651 - val_head_4_loss: 0.0919 - val_head_5_accuracy: 0.9649 - val_head_5_loss: 0.0921 - val_head_6_accuracy: 0.9653 - val_head_6_loss: 0.0921 - val_head_7_accuracy: 0.9647 - val_head_7_loss: 0.0921 - val_loss: 0.6443 - learning_rate: 2.9529e-04\n","\n","Epoch 52: LearningRateScheduler setting learning rate to 0.00028577780434866853.\n","Epoch 52/100\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 78ms/step - head_1_accuracy: 0.9670 - head_1_loss: 0.0891 - head_2_accuracy: 0.9672 - head_2_loss: 0.0893 - head_3_accuracy: 0.9674 - head_3_loss: 0.0892 - head_4_accuracy: 0.9672 - head_4_loss: 0.0891 - head_5_accuracy: 0.9671 - head_5_loss: 0.0890 - head_6_accuracy: 0.9670 - head_6_loss: 0.0893 - head_7_accuracy: 0.9671 - head_7_loss: 0.0892 - loss: 0.6241\n","Epoch 52: val_head_1_accuracy did not improve from 0.98041\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m27s\u001b[0m 85ms/step - head_1_accuracy: 0.9658 - head_1_loss: 0.0943 - head_2_accuracy: 0.9659 - head_2_loss: 0.0946 - head_3_accuracy: 0.9658 - head_3_loss: 0.0945 - head_4_accuracy: 0.9659 - head_4_loss: 0.0945 - head_5_accuracy: 0.9658 - head_5_loss: 0.0944 - head_6_accuracy: 0.9659 - head_6_loss: 0.0945 - head_7_accuracy: 0.9658 - head_7_loss: 0.0944 - loss: 0.6615 - val_head_1_accuracy: 0.9774 - val_head_1_loss: 0.0581 - val_head_2_accuracy: 0.9775 - val_head_2_loss: 0.0581 - val_head_3_accuracy: 0.9774 - val_head_3_loss: 0.0581 - val_head_4_accuracy: 0.9773 - val_head_4_loss: 0.0580 - val_head_5_accuracy: 0.9774 - val_head_5_loss: 0.0580 - val_head_6_accuracy: 0.9774 - val_head_6_loss: 0.0580 - val_head_7_accuracy: 0.9772 - val_head_7_loss: 0.0580 - val_loss: 0.4057 - learning_rate: 2.8578e-04\n","\n","Epoch 53: LearningRateScheduler setting learning rate to 0.0002762789754408063.\n","Epoch 53/100\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 79ms/step - head_1_accuracy: 0.9574 - head_1_loss: 0.1257 - head_2_accuracy: 0.9575 - head_2_loss: 0.1258 - head_3_accuracy: 0.9575 - head_3_loss: 0.1258 - head_4_accuracy: 0.9575 - head_4_loss: 0.1259 - head_5_accuracy: 0.9577 - head_5_loss: 0.1259 - head_6_accuracy: 0.9573 - head_6_loss: 0.1257 - head_7_accuracy: 0.9576 - head_7_loss: 0.1257 - loss: 0.8805\n","Epoch 53: val_head_1_accuracy did not improve from 0.98041\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m27s\u001b[0m 85ms/step - head_1_accuracy: 0.9033 - head_1_loss: 0.2879 - head_2_accuracy: 0.9035 - head_2_loss: 0.2880 - head_3_accuracy: 0.9035 - head_3_loss: 0.2880 - head_4_accuracy: 0.9034 - head_4_loss: 0.2882 - head_5_accuracy: 0.9036 - head_5_loss: 0.2882 - head_6_accuracy: 0.9033 - head_6_loss: 0.2879 - head_7_accuracy: 0.9035 - head_7_loss: 0.2880 - loss: 2.0154 - val_head_1_accuracy: 0.8634 - val_head_1_loss: 0.4052 - val_head_2_accuracy: 0.8634 - val_head_2_loss: 0.4048 - val_head_3_accuracy: 0.8637 - val_head_3_loss: 0.4050 - val_head_4_accuracy: 0.8636 - val_head_4_loss: 0.4049 - val_head_5_accuracy: 0.8636 - val_head_5_loss: 0.4043 - val_head_6_accuracy: 0.8634 - val_head_6_loss: 0.4055 - val_head_7_accuracy: 0.8630 - val_head_7_loss: 0.4049 - val_loss: 2.8461 - learning_rate: 2.7628e-04\n","\n","Epoch 54: LearningRateScheduler setting learning rate to 0.0002668040819396918.\n","Epoch 54/100\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 78ms/step - head_1_accuracy: 0.8444 - head_1_loss: 0.4649 - head_2_accuracy: 0.8441 - head_2_loss: 0.4650 - head_3_accuracy: 0.8445 - head_3_loss: 0.4655 - head_4_accuracy: 0.8443 - head_4_loss: 0.4656 - head_5_accuracy: 0.8449 - head_5_loss: 0.4650 - head_6_accuracy: 0.8442 - head_6_loss: 0.4655 - head_7_accuracy: 0.8453 - head_7_loss: 0.4651 - loss: 3.2566\n","Epoch 54: val_head_1_accuracy did not improve from 0.98041\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m27s\u001b[0m 85ms/step - head_1_accuracy: 0.8608 - head_1_loss: 0.4051 - head_2_accuracy: 0.8604 - head_2_loss: 0.4051 - head_3_accuracy: 0.8606 - head_3_loss: 0.4054 - head_4_accuracy: 0.8606 - head_4_loss: 0.4053 - head_5_accuracy: 0.8608 - head_5_loss: 0.4051 - head_6_accuracy: 0.8604 - head_6_loss: 0.4054 - head_7_accuracy: 0.8613 - head_7_loss: 0.4051 - loss: 2.8375 - val_head_1_accuracy: 0.9267 - val_head_1_loss: 0.2034 - val_head_2_accuracy: 0.9265 - val_head_2_loss: 0.2030 - val_head_3_accuracy: 0.9270 - val_head_3_loss: 0.2027 - val_head_4_accuracy: 0.9267 - val_head_4_loss: 0.2030 - val_head_5_accuracy: 0.9264 - val_head_5_loss: 0.2029 - val_head_6_accuracy: 0.9271 - val_head_6_loss: 0.2032 - val_head_7_accuracy: 0.9271 - val_head_7_loss: 0.2035 - val_loss: 1.4268 - learning_rate: 2.6680e-04\n","\n","Epoch 55: LearningRateScheduler setting learning rate to 0.00025736266425992817.\n","Epoch 55/100\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 78ms/step - head_1_accuracy: 0.9072 - head_1_loss: 0.2524 - head_2_accuracy: 0.9074 - head_2_loss: 0.2524 - head_3_accuracy: 0.9073 - head_3_loss: 0.2524 - head_4_accuracy: 0.9077 - head_4_loss: 0.2521 - head_5_accuracy: 0.9074 - head_5_loss: 0.2524 - head_6_accuracy: 0.9072 - head_6_loss: 0.2524 - head_7_accuracy: 0.9068 - head_7_loss: 0.2524 - loss: 1.7665\n","Epoch 55: val_head_1_accuracy did not improve from 0.98041\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m27s\u001b[0m 85ms/step - head_1_accuracy: 0.9127 - head_1_loss: 0.2344 - head_2_accuracy: 0.9130 - head_2_loss: 0.2344 - head_3_accuracy: 0.9128 - head_3_loss: 0.2343 - head_4_accuracy: 0.9131 - head_4_loss: 0.2342 - head_5_accuracy: 0.9128 - head_5_loss: 0.2343 - head_6_accuracy: 0.9131 - head_6_loss: 0.2343 - head_7_accuracy: 0.9124 - head_7_loss: 0.2344 - loss: 1.6405 - val_head_1_accuracy: 0.9450 - val_head_1_loss: 0.1506 - val_head_2_accuracy: 0.9449 - val_head_2_loss: 0.1505 - val_head_3_accuracy: 0.9451 - val_head_3_loss: 0.1505 - val_head_4_accuracy: 0.9452 - val_head_4_loss: 0.1503 - val_head_5_accuracy: 0.9449 - val_head_5_loss: 0.1505 - val_head_6_accuracy: 0.9445 - val_head_6_loss: 0.1506 - val_head_7_accuracy: 0.9456 - val_head_7_loss: 0.1502 - val_loss: 1.0573 - learning_rate: 2.5736e-04\n","\n","Epoch 56: LearningRateScheduler setting learning rate to 0.00024796422910880423.\n","Epoch 56/100\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 78ms/step - head_1_accuracy: 0.9313 - head_1_loss: 0.1887 - head_2_accuracy: 0.9312 - head_2_loss: 0.1888 - head_3_accuracy: 0.9310 - head_3_loss: 0.1886 - head_4_accuracy: 0.9313 - head_4_loss: 0.1886 - head_5_accuracy: 0.9309 - head_5_loss: 0.1888 - head_6_accuracy: 0.9313 - head_6_loss: 0.1888 - head_7_accuracy: 0.9313 - head_7_loss: 0.1886 - loss: 1.3209\n","Epoch 56: val_head_1_accuracy did not improve from 0.98041\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m27s\u001b[0m 85ms/step - head_1_accuracy: 0.9321 - head_1_loss: 0.1873 - head_2_accuracy: 0.9319 - head_2_loss: 0.1874 - head_3_accuracy: 0.9318 - head_3_loss: 0.1873 - head_4_accuracy: 0.9319 - head_4_loss: 0.1873 - head_5_accuracy: 0.9318 - head_5_loss: 0.1874 - head_6_accuracy: 0.9320 - head_6_loss: 0.1874 - head_7_accuracy: 0.9317 - head_7_loss: 0.1873 - loss: 1.3117 - val_head_1_accuracy: 0.9575 - val_head_1_loss: 0.1094 - val_head_2_accuracy: 0.9575 - val_head_2_loss: 0.1095 - val_head_3_accuracy: 0.9575 - val_head_3_loss: 0.1095 - val_head_4_accuracy: 0.9576 - val_head_4_loss: 0.1093 - val_head_5_accuracy: 0.9579 - val_head_5_loss: 0.1095 - val_head_6_accuracy: 0.9578 - val_head_6_loss: 0.1094 - val_head_7_accuracy: 0.9578 - val_head_7_loss: 0.1094 - val_loss: 0.7660 - learning_rate: 2.4796e-04\n","\n","Epoch 57: LearningRateScheduler setting learning rate to 0.0002386182399138461.\n","Epoch 57/100\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 79ms/step - head_1_accuracy: 0.9412 - head_1_loss: 0.1611 - head_2_accuracy: 0.9407 - head_2_loss: 0.1614 - head_3_accuracy: 0.9410 - head_3_loss: 0.1613 - head_4_accuracy: 0.9409 - head_4_loss: 0.1610 - head_5_accuracy: 0.9409 - head_5_loss: 0.1612 - head_6_accuracy: 0.9409 - head_6_loss: 0.1613 - head_7_accuracy: 0.9410 - head_7_loss: 0.1612 - loss: 1.1284\n","Epoch 57: val_head_1_accuracy did not improve from 0.98041\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m27s\u001b[0m 85ms/step - head_1_accuracy: 0.9422 - head_1_loss: 0.1576 - head_2_accuracy: 0.9422 - head_2_loss: 0.1579 - head_3_accuracy: 0.9422 - head_3_loss: 0.1577 - head_4_accuracy: 0.9425 - head_4_loss: 0.1576 - head_5_accuracy: 0.9423 - head_5_loss: 0.1577 - head_6_accuracy: 0.9424 - head_6_loss: 0.1577 - head_7_accuracy: 0.9423 - head_7_loss: 0.1577 - loss: 1.1044 - val_head_1_accuracy: 0.9619 - val_head_1_loss: 0.0973 - val_head_2_accuracy: 0.9616 - val_head_2_loss: 0.0972 - val_head_3_accuracy: 0.9616 - val_head_3_loss: 0.0973 - val_head_4_accuracy: 0.9617 - val_head_4_loss: 0.0972 - val_head_5_accuracy: 0.9619 - val_head_5_loss: 0.0971 - val_head_6_accuracy: 0.9618 - val_head_6_loss: 0.0973 - val_head_7_accuracy: 0.9620 - val_head_7_loss: 0.0972 - val_loss: 0.6820 - learning_rate: 2.3862e-04\n","\n","Epoch 58: LearningRateScheduler setting learning rate to 0.0002293341072939474.\n","Epoch 58/100\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 79ms/step - head_1_accuracy: 0.9486 - head_1_loss: 0.1398 - head_2_accuracy: 0.9488 - head_2_loss: 0.1399 - head_3_accuracy: 0.9491 - head_3_loss: 0.1399 - head_4_accuracy: 0.9482 - head_4_loss: 0.1397 - head_5_accuracy: 0.9486 - head_5_loss: 0.1398 - head_6_accuracy: 0.9493 - head_6_loss: 0.1398 - head_7_accuracy: 0.9490 - head_7_loss: 0.1398 - loss: 0.9786\n","Epoch 58: val_head_1_accuracy did not improve from 0.98041\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m27s\u001b[0m 85ms/step - head_1_accuracy: 0.9494 - head_1_loss: 0.1369 - head_2_accuracy: 0.9496 - head_2_loss: 0.1368 - head_3_accuracy: 0.9496 - head_3_loss: 0.1368 - head_4_accuracy: 0.9494 - head_4_loss: 0.1367 - head_5_accuracy: 0.9494 - head_5_loss: 0.1368 - head_6_accuracy: 0.9498 - head_6_loss: 0.1367 - head_7_accuracy: 0.9498 - head_7_loss: 0.1368 - loss: 0.9577 - val_head_1_accuracy: 0.9656 - val_head_1_loss: 0.0906 - val_head_2_accuracy: 0.9656 - val_head_2_loss: 0.0905 - val_head_3_accuracy: 0.9656 - val_head_3_loss: 0.0907 - val_head_4_accuracy: 0.9654 - val_head_4_loss: 0.0905 - val_head_5_accuracy: 0.9655 - val_head_5_loss: 0.0905 - val_head_6_accuracy: 0.9655 - val_head_6_loss: 0.0903 - val_head_7_accuracy: 0.9656 - val_head_7_loss: 0.0904 - val_loss: 0.6340 - learning_rate: 2.2933e-04\n","\n","Epoch 59: LearningRateScheduler setting learning rate to 0.00022012117958367398.\n","Epoch 59/100\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 78ms/step - head_1_accuracy: 0.9514 - head_1_loss: 0.1314 - head_2_accuracy: 0.9517 - head_2_loss: 0.1314 - head_3_accuracy: 0.9518 - head_3_loss: 0.1315 - head_4_accuracy: 0.9514 - head_4_loss: 0.1314 - head_5_accuracy: 0.9516 - head_5_loss: 0.1314 - head_6_accuracy: 0.9518 - head_6_loss: 0.1314 - head_7_accuracy: 0.9514 - head_7_loss: 0.1314 - loss: 0.9199\n","Epoch 59: val_head_1_accuracy did not improve from 0.98041\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m27s\u001b[0m 85ms/step - head_1_accuracy: 0.9511 - head_1_loss: 0.1308 - head_2_accuracy: 0.9513 - head_2_loss: 0.1308 - head_3_accuracy: 0.9514 - head_3_loss: 0.1309 - head_4_accuracy: 0.9512 - head_4_loss: 0.1308 - head_5_accuracy: 0.9512 - head_5_loss: 0.1309 - head_6_accuracy: 0.9514 - head_6_loss: 0.1308 - head_7_accuracy: 0.9511 - head_7_loss: 0.1309 - loss: 0.9161 - val_head_1_accuracy: 0.9637 - val_head_1_loss: 0.0935 - val_head_2_accuracy: 0.9639 - val_head_2_loss: 0.0936 - val_head_3_accuracy: 0.9641 - val_head_3_loss: 0.0935 - val_head_4_accuracy: 0.9639 - val_head_4_loss: 0.0936 - val_head_5_accuracy: 0.9636 - val_head_5_loss: 0.0936 - val_head_6_accuracy: 0.9638 - val_head_6_loss: 0.0935 - val_head_7_accuracy: 0.9638 - val_head_7_loss: 0.0934 - val_loss: 0.6552 - learning_rate: 2.2012e-04\n","\n","Epoch 60: LearningRateScheduler setting learning rate to 0.00021098873342028398.\n","Epoch 60/100\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 79ms/step - head_1_accuracy: 0.9515 - head_1_loss: 0.1323 - head_2_accuracy: 0.9520 - head_2_loss: 0.1320 - head_3_accuracy: 0.9520 - head_3_loss: 0.1320 - head_4_accuracy: 0.9520 - head_4_loss: 0.1320 - head_5_accuracy: 0.9516 - head_5_loss: 0.1320 - head_6_accuracy: 0.9518 - head_6_loss: 0.1319 - head_7_accuracy: 0.9520 - head_7_loss: 0.1320 - loss: 0.9242\n","Epoch 60: val_head_1_accuracy did not improve from 0.98041\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m27s\u001b[0m 86ms/step - head_1_accuracy: 0.9540 - head_1_loss: 0.1252 - head_2_accuracy: 0.9542 - head_2_loss: 0.1249 - head_3_accuracy: 0.9540 - head_3_loss: 0.1249 - head_4_accuracy: 0.9541 - head_4_loss: 0.1250 - head_5_accuracy: 0.9542 - head_5_loss: 0.1250 - head_6_accuracy: 0.9543 - head_6_loss: 0.1248 - head_7_accuracy: 0.9543 - head_7_loss: 0.1250 - loss: 0.8751 - val_head_1_accuracy: 0.9694 - val_head_1_loss: 0.0804 - val_head_2_accuracy: 0.9695 - val_head_2_loss: 0.0802 - val_head_3_accuracy: 0.9693 - val_head_3_loss: 0.0802 - val_head_4_accuracy: 0.9694 - val_head_4_loss: 0.0802 - val_head_5_accuracy: 0.9691 - val_head_5_loss: 0.0803 - val_head_6_accuracy: 0.9690 - val_head_6_loss: 0.0803 - val_head_7_accuracy: 0.9689 - val_head_7_loss: 0.0803 - val_loss: 0.5615 - learning_rate: 2.1099e-04\n","\n","Epoch 61: LearningRateScheduler setting learning rate to 0.00020194596440293946.\n","Epoch 61/100\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 79ms/step - head_1_accuracy: 0.9597 - head_1_loss: 0.1106 - head_2_accuracy: 0.9596 - head_2_loss: 0.1106 - head_3_accuracy: 0.9597 - head_3_loss: 0.1106 - head_4_accuracy: 0.9596 - head_4_loss: 0.1107 - head_5_accuracy: 0.9595 - head_5_loss: 0.1106 - head_6_accuracy: 0.9595 - head_6_loss: 0.1108 - head_7_accuracy: 0.9598 - head_7_loss: 0.1108 - loss: 0.7747\n","Epoch 61: val_head_1_accuracy did not improve from 0.98041\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m27s\u001b[0m 86ms/step - head_1_accuracy: 0.9602 - head_1_loss: 0.1088 - head_2_accuracy: 0.9602 - head_2_loss: 0.1088 - head_3_accuracy: 0.9601 - head_3_loss: 0.1088 - head_4_accuracy: 0.9602 - head_4_loss: 0.1088 - head_5_accuracy: 0.9602 - head_5_loss: 0.1088 - head_6_accuracy: 0.9602 - head_6_loss: 0.1088 - head_7_accuracy: 0.9604 - head_7_loss: 0.1089 - loss: 0.7619 - val_head_1_accuracy: 0.9704 - val_head_1_loss: 0.0773 - val_head_2_accuracy: 0.9707 - val_head_2_loss: 0.0773 - val_head_3_accuracy: 0.9705 - val_head_3_loss: 0.0771 - val_head_4_accuracy: 0.9708 - val_head_4_loss: 0.0772 - val_head_5_accuracy: 0.9703 - val_head_5_loss: 0.0772 - val_head_6_accuracy: 0.9705 - val_head_6_loss: 0.0773 - val_head_7_accuracy: 0.9705 - val_head_7_loss: 0.0773 - val_loss: 0.5416 - learning_rate: 2.0195e-04\n","\n","Epoch 62: LearningRateScheduler setting learning rate to 0.00019300197783351806.\n","Epoch 62/100\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 79ms/step - head_1_accuracy: 0.9616 - head_1_loss: 0.1053 - head_2_accuracy: 0.9615 - head_2_loss: 0.1051 - head_3_accuracy: 0.9613 - head_3_loss: 0.1052 - head_4_accuracy: 0.9613 - head_4_loss: 0.1056 - head_5_accuracy: 0.9618 - head_5_loss: 0.1052 - head_6_accuracy: 0.9620 - head_6_loss: 0.1051 - head_7_accuracy: 0.9617 - head_7_loss: 0.1053 - loss: 0.7367\n","Epoch 62: val_head_1_accuracy did not improve from 0.98041\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m27s\u001b[0m 86ms/step - head_1_accuracy: 0.9611 - head_1_loss: 0.1062 - head_2_accuracy: 0.9614 - head_2_loss: 0.1062 - head_3_accuracy: 0.9612 - head_3_loss: 0.1062 - head_4_accuracy: 0.9612 - head_4_loss: 0.1064 - head_5_accuracy: 0.9614 - head_5_loss: 0.1063 - head_6_accuracy: 0.9615 - head_6_loss: 0.1062 - head_7_accuracy: 0.9614 - head_7_loss: 0.1063 - loss: 0.7440 - val_head_1_accuracy: 0.9745 - val_head_1_loss: 0.0696 - val_head_2_accuracy: 0.9748 - val_head_2_loss: 0.0694 - val_head_3_accuracy: 0.9746 - val_head_3_loss: 0.0695 - val_head_4_accuracy: 0.9748 - val_head_4_loss: 0.0695 - val_head_5_accuracy: 0.9742 - val_head_5_loss: 0.0696 - val_head_6_accuracy: 0.9747 - val_head_6_loss: 0.0693 - val_head_7_accuracy: 0.9745 - val_head_7_loss: 0.0695 - val_loss: 0.4851 - learning_rate: 1.9300e-04\n","\n","Epoch 63: LearningRateScheduler setting learning rate to 0.00018416577954834612.\n","Epoch 63/100\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 79ms/step - head_1_accuracy: 0.9614 - head_1_loss: 0.1074 - head_2_accuracy: 0.9610 - head_2_loss: 0.1073 - head_3_accuracy: 0.9611 - head_3_loss: 0.1072 - head_4_accuracy: 0.9613 - head_4_loss: 0.1075 - head_5_accuracy: 0.9612 - head_5_loss: 0.1074 - head_6_accuracy: 0.9614 - head_6_loss: 0.1074 - head_7_accuracy: 0.9612 - head_7_loss: 0.1073 - loss: 0.7516\n","Epoch 63: val_head_1_accuracy did not improve from 0.98041\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m27s\u001b[0m 85ms/step - head_1_accuracy: 0.9607 - head_1_loss: 0.1082 - head_2_accuracy: 0.9605 - head_2_loss: 0.1082 - head_3_accuracy: 0.9604 - head_3_loss: 0.1081 - head_4_accuracy: 0.9606 - head_4_loss: 0.1082 - head_5_accuracy: 0.9606 - head_5_loss: 0.1082 - head_6_accuracy: 0.9607 - head_6_loss: 0.1082 - head_7_accuracy: 0.9605 - head_7_loss: 0.1081 - loss: 0.7575 - val_head_1_accuracy: 0.9768 - val_head_1_loss: 0.0589 - val_head_2_accuracy: 0.9768 - val_head_2_loss: 0.0589 - val_head_3_accuracy: 0.9770 - val_head_3_loss: 0.0588 - val_head_4_accuracy: 0.9768 - val_head_4_loss: 0.0590 - val_head_5_accuracy: 0.9765 - val_head_5_loss: 0.0589 - val_head_6_accuracy: 0.9771 - val_head_6_loss: 0.0588 - val_head_7_accuracy: 0.9765 - val_head_7_loss: 0.0590 - val_loss: 0.4121 - learning_rate: 1.8417e-04\n","\n","Epoch 64: LearningRateScheduler setting learning rate to 0.00017544626685008413.\n","Epoch 64/100\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 79ms/step - head_1_accuracy: 0.9641 - head_1_loss: 0.0960 - head_2_accuracy: 0.9645 - head_2_loss: 0.0957 - head_3_accuracy: 0.9644 - head_3_loss: 0.0959 - head_4_accuracy: 0.9645 - head_4_loss: 0.0960 - head_5_accuracy: 0.9643 - head_5_loss: 0.0958 - head_6_accuracy: 0.9644 - head_6_loss: 0.0960 - head_7_accuracy: 0.9642 - head_7_loss: 0.0961 - loss: 0.6715\n","Epoch 64: val_head_1_accuracy did not improve from 0.98041\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m27s\u001b[0m 86ms/step - head_1_accuracy: 0.9649 - head_1_loss: 0.0948 - head_2_accuracy: 0.9651 - head_2_loss: 0.0946 - head_3_accuracy: 0.9650 - head_3_loss: 0.0947 - head_4_accuracy: 0.9652 - head_4_loss: 0.0948 - head_5_accuracy: 0.9650 - head_5_loss: 0.0947 - head_6_accuracy: 0.9651 - head_6_loss: 0.0947 - head_7_accuracy: 0.9648 - head_7_loss: 0.0948 - loss: 0.6626 - val_head_1_accuracy: 0.9708 - val_head_1_loss: 0.0694 - val_head_2_accuracy: 0.9711 - val_head_2_loss: 0.0690 - val_head_3_accuracy: 0.9710 - val_head_3_loss: 0.0692 - val_head_4_accuracy: 0.9711 - val_head_4_loss: 0.0691 - val_head_5_accuracy: 0.9708 - val_head_5_loss: 0.0693 - val_head_6_accuracy: 0.9711 - val_head_6_loss: 0.0691 - val_head_7_accuracy: 0.9712 - val_head_7_loss: 0.0690 - val_loss: 0.4813 - learning_rate: 1.7545e-04\n","\n","Epoch 65: LearningRateScheduler setting learning rate to 0.00016685221954889804.\n","Epoch 65/100\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 79ms/step - head_1_accuracy: 0.9661 - head_1_loss: 0.0956 - head_2_accuracy: 0.9658 - head_2_loss: 0.0955 - head_3_accuracy: 0.9659 - head_3_loss: 0.0955 - head_4_accuracy: 0.9656 - head_4_loss: 0.0955 - head_5_accuracy: 0.9656 - head_5_loss: 0.0957 - head_6_accuracy: 0.9658 - head_6_loss: 0.0953 - head_7_accuracy: 0.9652 - head_7_loss: 0.0954 - loss: 0.6686\n","Epoch 65: val_head_1_accuracy did not improve from 0.98041\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m27s\u001b[0m 85ms/step - head_1_accuracy: 0.9659 - head_1_loss: 0.0937 - head_2_accuracy: 0.9659 - head_2_loss: 0.0936 - head_3_accuracy: 0.9660 - head_3_loss: 0.0936 - head_4_accuracy: 0.9656 - head_4_loss: 0.0936 - head_5_accuracy: 0.9657 - head_5_loss: 0.0936 - head_6_accuracy: 0.9657 - head_6_loss: 0.0935 - head_7_accuracy: 0.9655 - head_7_loss: 0.0936 - loss: 0.6557 - val_head_1_accuracy: 0.9797 - val_head_1_loss: 0.0518 - val_head_2_accuracy: 0.9797 - val_head_2_loss: 0.0517 - val_head_3_accuracy: 0.9797 - val_head_3_loss: 0.0517 - val_head_4_accuracy: 0.9798 - val_head_4_loss: 0.0518 - val_head_5_accuracy: 0.9797 - val_head_5_loss: 0.0517 - val_head_6_accuracy: 0.9797 - val_head_6_loss: 0.0516 - val_head_7_accuracy: 0.9798 - val_head_7_loss: 0.0518 - val_loss: 0.3602 - learning_rate: 1.6685e-04\n","\n","Epoch 66: LearningRateScheduler setting learning rate to 0.00015839229112193386.\n","Epoch 66/100\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 79ms/step - head_1_accuracy: 0.9692 - head_1_loss: 0.0863 - head_2_accuracy: 0.9692 - head_2_loss: 0.0861 - head_3_accuracy: 0.9692 - head_3_loss: 0.0860 - head_4_accuracy: 0.9690 - head_4_loss: 0.0862 - head_5_accuracy: 0.9693 - head_5_loss: 0.0862 - head_6_accuracy: 0.9690 - head_6_loss: 0.0861 - head_7_accuracy: 0.9689 - head_7_loss: 0.0862 - loss: 0.6032\n","Epoch 66: val_head_1_accuracy did not improve from 0.98041\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m27s\u001b[0m 85ms/step - head_1_accuracy: 0.9675 - head_1_loss: 0.0893 - head_2_accuracy: 0.9674 - head_2_loss: 0.0892 - head_3_accuracy: 0.9676 - head_3_loss: 0.0891 - head_4_accuracy: 0.9674 - head_4_loss: 0.0893 - head_5_accuracy: 0.9676 - head_5_loss: 0.0893 - head_6_accuracy: 0.9674 - head_6_loss: 0.0892 - head_7_accuracy: 0.9672 - head_7_loss: 0.0893 - loss: 0.6250 - val_head_1_accuracy: 0.9787 - val_head_1_loss: 0.0540 - val_head_2_accuracy: 0.9785 - val_head_2_loss: 0.0540 - val_head_3_accuracy: 0.9786 - val_head_3_loss: 0.0539 - val_head_4_accuracy: 0.9786 - val_head_4_loss: 0.0540 - val_head_5_accuracy: 0.9786 - val_head_5_loss: 0.0540 - val_head_6_accuracy: 0.9786 - val_head_6_loss: 0.0538 - val_head_7_accuracy: 0.9786 - val_head_7_loss: 0.0539 - val_loss: 0.3741 - learning_rate: 1.5839e-04\n","\n","Epoch 67: LearningRateScheduler setting learning rate to 0.00015007500000000006.\n","Epoch 67/100\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 78ms/step - head_1_accuracy: 0.9698 - head_1_loss: 0.0792 - head_2_accuracy: 0.9701 - head_2_loss: 0.0791 - head_3_accuracy: 0.9700 - head_3_loss: 0.0791 - head_4_accuracy: 0.9704 - head_4_loss: 0.0791 - head_5_accuracy: 0.9700 - head_5_loss: 0.0792 - head_6_accuracy: 0.9699 - head_6_loss: 0.0791 - head_7_accuracy: 0.9701 - head_7_loss: 0.0791 - loss: 0.5538\n","Epoch 67: val_head_1_accuracy did not improve from 0.98041\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m27s\u001b[0m 85ms/step - head_1_accuracy: 0.9682 - head_1_loss: 0.0858 - head_2_accuracy: 0.9683 - head_2_loss: 0.0858 - head_3_accuracy: 0.9682 - head_3_loss: 0.0858 - head_4_accuracy: 0.9684 - head_4_loss: 0.0859 - head_5_accuracy: 0.9684 - head_5_loss: 0.0859 - head_6_accuracy: 0.9682 - head_6_loss: 0.0858 - head_7_accuracy: 0.9683 - head_7_loss: 0.0858 - loss: 0.6008 - val_head_1_accuracy: 0.9803 - val_head_1_loss: 0.0508 - val_head_2_accuracy: 0.9803 - val_head_2_loss: 0.0507 - val_head_3_accuracy: 0.9807 - val_head_3_loss: 0.0506 - val_head_4_accuracy: 0.9803 - val_head_4_loss: 0.0508 - val_head_5_accuracy: 0.9806 - val_head_5_loss: 0.0507 - val_head_6_accuracy: 0.9805 - val_head_6_loss: 0.0507 - val_head_7_accuracy: 0.9802 - val_head_7_loss: 0.0508 - val_loss: 0.3528 - learning_rate: 1.5007e-04\n","\n","Epoch 68: LearningRateScheduler setting learning rate to 0.00014190872099022978.\n","Epoch 68/100\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 79ms/step - head_1_accuracy: 0.9713 - head_1_loss: 0.0784 - head_2_accuracy: 0.9713 - head_2_loss: 0.0783 - head_3_accuracy: 0.9715 - head_3_loss: 0.0784 - head_4_accuracy: 0.9710 - head_4_loss: 0.0784 - head_5_accuracy: 0.9713 - head_5_loss: 0.0784 - head_6_accuracy: 0.9716 - head_6_loss: 0.0783 - head_7_accuracy: 0.9711 - head_7_loss: 0.0783 - loss: 0.5485\n","Epoch 68: val_head_1_accuracy improved from 0.98041 to 0.98367, saving model to /kaggle/working/results/models/ViT_UNet_MultiHead_best.keras\n","\n","Epoch 68: finished saving model to /kaggle/working/results/models/ViT_UNet_MultiHead_best.keras\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m29s\u001b[0m 91ms/step - head_1_accuracy: 0.9710 - head_1_loss: 0.0793 - head_2_accuracy: 0.9711 - head_2_loss: 0.0791 - head_3_accuracy: 0.9712 - head_3_loss: 0.0792 - head_4_accuracy: 0.9709 - head_4_loss: 0.0793 - head_5_accuracy: 0.9710 - head_5_loss: 0.0792 - head_6_accuracy: 0.9712 - head_6_loss: 0.0793 - head_7_accuracy: 0.9710 - head_7_loss: 0.0792 - loss: 0.5547 - val_head_1_accuracy: 0.9837 - val_head_1_loss: 0.0471 - val_head_2_accuracy: 0.9837 - val_head_2_loss: 0.0470 - val_head_3_accuracy: 0.9835 - val_head_3_loss: 0.0469 - val_head_4_accuracy: 0.9835 - val_head_4_loss: 0.0470 - val_head_5_accuracy: 0.9835 - val_head_5_loss: 0.0471 - val_head_6_accuracy: 0.9836 - val_head_6_loss: 0.0470 - val_head_7_accuracy: 0.9837 - val_head_7_loss: 0.0471 - val_loss: 0.3257 - learning_rate: 1.4191e-04\n","\n","Epoch 69: LearningRateScheduler setting learning rate to 0.00013390167684336024.\n","Epoch 69/100\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 79ms/step - head_1_accuracy: 0.9717 - head_1_loss: 0.0772 - head_2_accuracy: 0.9715 - head_2_loss: 0.0772 - head_3_accuracy: 0.9716 - head_3_loss: 0.0774 - head_4_accuracy: 0.9717 - head_4_loss: 0.0774 - head_5_accuracy: 0.9716 - head_5_loss: 0.0772 - head_6_accuracy: 0.9719 - head_6_loss: 0.0773 - head_7_accuracy: 0.9714 - head_7_loss: 0.0772 - loss: 0.5409\n","Epoch 69: val_head_1_accuracy improved from 0.98367 to 0.98479, saving model to /kaggle/working/results/models/ViT_UNet_MultiHead_best.keras\n","\n","Epoch 69: finished saving model to /kaggle/working/results/models/ViT_UNet_MultiHead_best.keras\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m29s\u001b[0m 91ms/step - head_1_accuracy: 0.9725 - head_1_loss: 0.0742 - head_2_accuracy: 0.9723 - head_2_loss: 0.0743 - head_3_accuracy: 0.9725 - head_3_loss: 0.0743 - head_4_accuracy: 0.9726 - head_4_loss: 0.0742 - head_5_accuracy: 0.9723 - head_5_loss: 0.0743 - head_6_accuracy: 0.9727 - head_6_loss: 0.0742 - head_7_accuracy: 0.9724 - head_7_loss: 0.0742 - loss: 0.5197 - val_head_1_accuracy: 0.9848 - val_head_1_loss: 0.0428 - val_head_2_accuracy: 0.9846 - val_head_2_loss: 0.0427 - val_head_3_accuracy: 0.9848 - val_head_3_loss: 0.0428 - val_head_4_accuracy: 0.9848 - val_head_4_loss: 0.0428 - val_head_5_accuracy: 0.9847 - val_head_5_loss: 0.0429 - val_head_6_accuracy: 0.9850 - val_head_6_loss: 0.0428 - val_head_7_accuracy: 0.9847 - val_head_7_loss: 0.0429 - val_loss: 0.2958 - learning_rate: 1.3390e-04\n","\n","Epoch 70: LearningRateScheduler setting learning rate to 0.00012606192997411915.\n","Epoch 70/100\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 79ms/step - head_1_accuracy: 0.9740 - head_1_loss: 0.0725 - head_2_accuracy: 0.9741 - head_2_loss: 0.0725 - head_3_accuracy: 0.9742 - head_3_loss: 0.0725 - head_4_accuracy: 0.9740 - head_4_loss: 0.0726 - head_5_accuracy: 0.9741 - head_5_loss: 0.0726 - head_6_accuracy: 0.9740 - head_6_loss: 0.0726 - head_7_accuracy: 0.9740 - head_7_loss: 0.0726 - loss: 0.5079\n","Epoch 70: val_head_1_accuracy improved from 0.98479 to 0.98627, saving model to /kaggle/working/results/models/ViT_UNet_MultiHead_best.keras\n","\n","Epoch 70: finished saving model to /kaggle/working/results/models/ViT_UNet_MultiHead_best.keras\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m29s\u001b[0m 91ms/step - head_1_accuracy: 0.9738 - head_1_loss: 0.0721 - head_2_accuracy: 0.9739 - head_2_loss: 0.0722 - head_3_accuracy: 0.9739 - head_3_loss: 0.0721 - head_4_accuracy: 0.9737 - head_4_loss: 0.0722 - head_5_accuracy: 0.9740 - head_5_loss: 0.0722 - head_6_accuracy: 0.9739 - head_6_loss: 0.0721 - head_7_accuracy: 0.9738 - head_7_loss: 0.0722 - loss: 0.5054 - val_head_1_accuracy: 0.9863 - val_head_1_loss: 0.0361 - val_head_2_accuracy: 0.9862 - val_head_2_loss: 0.0361 - val_head_3_accuracy: 0.9863 - val_head_3_loss: 0.0360 - val_head_4_accuracy: 0.9861 - val_head_4_loss: 0.0362 - val_head_5_accuracy: 0.9861 - val_head_5_loss: 0.0361 - val_head_6_accuracy: 0.9862 - val_head_6_loss: 0.0361 - val_head_7_accuracy: 0.9862 - val_head_7_loss: 0.0361 - val_loss: 0.2525 - learning_rate: 1.2606e-04\n","\n","Epoch 71: LearningRateScheduler setting learning rate to 0.00011839737434305693.\n","Epoch 71/100\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 79ms/step - head_1_accuracy: 0.9747 - head_1_loss: 0.0700 - head_2_accuracy: 0.9747 - head_2_loss: 0.0702 - head_3_accuracy: 0.9746 - head_3_loss: 0.0701 - head_4_accuracy: 0.9747 - head_4_loss: 0.0702 - head_5_accuracy: 0.9746 - head_5_loss: 0.0701 - head_6_accuracy: 0.9745 - head_6_loss: 0.0703 - head_7_accuracy: 0.9746 - head_7_loss: 0.0701 - loss: 0.4910\n","Epoch 71: val_head_1_accuracy did not improve from 0.98627\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m27s\u001b[0m 86ms/step - head_1_accuracy: 0.9739 - head_1_loss: 0.0716 - head_2_accuracy: 0.9738 - head_2_loss: 0.0717 - head_3_accuracy: 0.9737 - head_3_loss: 0.0718 - head_4_accuracy: 0.9737 - head_4_loss: 0.0717 - head_5_accuracy: 0.9737 - head_5_loss: 0.0717 - head_6_accuracy: 0.9738 - head_6_loss: 0.0718 - head_7_accuracy: 0.9738 - head_7_loss: 0.0717 - loss: 0.5022 - val_head_1_accuracy: 0.9834 - val_head_1_loss: 0.0451 - val_head_2_accuracy: 0.9833 - val_head_2_loss: 0.0451 - val_head_3_accuracy: 0.9835 - val_head_3_loss: 0.0450 - val_head_4_accuracy: 0.9833 - val_head_4_loss: 0.0451 - val_head_5_accuracy: 0.9831 - val_head_5_loss: 0.0452 - val_head_6_accuracy: 0.9834 - val_head_6_loss: 0.0451 - val_head_7_accuracy: 0.9833 - val_head_7_loss: 0.0451 - val_loss: 0.3117 - learning_rate: 1.1840e-04\n","\n","Epoch 72: LearningRateScheduler setting learning rate to 0.00011091572750799748.\n","Epoch 72/100\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 79ms/step - head_1_accuracy: 0.9770 - head_1_loss: 0.0660 - head_2_accuracy: 0.9771 - head_2_loss: 0.0658 - head_3_accuracy: 0.9772 - head_3_loss: 0.0658 - head_4_accuracy: 0.9772 - head_4_loss: 0.0661 - head_5_accuracy: 0.9770 - head_5_loss: 0.0659 - head_6_accuracy: 0.9767 - head_6_loss: 0.0659 - head_7_accuracy: 0.9769 - head_7_loss: 0.0659 - loss: 0.4614\n","Epoch 72: val_head_1_accuracy did not improve from 0.98627\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m27s\u001b[0m 85ms/step - head_1_accuracy: 0.9763 - head_1_loss: 0.0645 - head_2_accuracy: 0.9764 - head_2_loss: 0.0646 - head_3_accuracy: 0.9765 - head_3_loss: 0.0645 - head_4_accuracy: 0.9765 - head_4_loss: 0.0647 - head_5_accuracy: 0.9764 - head_5_loss: 0.0646 - head_6_accuracy: 0.9762 - head_6_loss: 0.0646 - head_7_accuracy: 0.9763 - head_7_loss: 0.0645 - loss: 0.4521 - val_head_1_accuracy: 0.9849 - val_head_1_loss: 0.0362 - val_head_2_accuracy: 0.9849 - val_head_2_loss: 0.0362 - val_head_3_accuracy: 0.9849 - val_head_3_loss: 0.0362 - val_head_4_accuracy: 0.9849 - val_head_4_loss: 0.0363 - val_head_5_accuracy: 0.9851 - val_head_5_loss: 0.0362 - val_head_6_accuracy: 0.9850 - val_head_6_loss: 0.0362 - val_head_7_accuracy: 0.9849 - val_head_7_loss: 0.0362 - val_loss: 0.2524 - learning_rate: 1.1092e-04\n","\n","Epoch 73: LearningRateScheduler setting learning rate to 0.00010362452285311177.\n","Epoch 73/100\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 79ms/step - head_1_accuracy: 0.9777 - head_1_loss: 0.0626 - head_2_accuracy: 0.9777 - head_2_loss: 0.0626 - head_3_accuracy: 0.9779 - head_3_loss: 0.0625 - head_4_accuracy: 0.9778 - head_4_loss: 0.0626 - head_5_accuracy: 0.9776 - head_5_loss: 0.0626 - head_6_accuracy: 0.9776 - head_6_loss: 0.0626 - head_7_accuracy: 0.9775 - head_7_loss: 0.0625 - loss: 0.4380\n","Epoch 73: val_head_1_accuracy improved from 0.98627 to 0.98634, saving model to /kaggle/working/results/models/ViT_UNet_MultiHead_best.keras\n","\n","Epoch 73: finished saving model to /kaggle/working/results/models/ViT_UNet_MultiHead_best.keras\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m28s\u001b[0m 90ms/step - head_1_accuracy: 0.9774 - head_1_loss: 0.0626 - head_2_accuracy: 0.9776 - head_2_loss: 0.0625 - head_3_accuracy: 0.9776 - head_3_loss: 0.0625 - head_4_accuracy: 0.9776 - head_4_loss: 0.0625 - head_5_accuracy: 0.9774 - head_5_loss: 0.0626 - head_6_accuracy: 0.9775 - head_6_loss: 0.0625 - head_7_accuracy: 0.9774 - head_7_loss: 0.0625 - loss: 0.4380 - val_head_1_accuracy: 0.9863 - val_head_1_loss: 0.0347 - val_head_2_accuracy: 0.9863 - val_head_2_loss: 0.0347 - val_head_3_accuracy: 0.9862 - val_head_3_loss: 0.0347 - val_head_4_accuracy: 0.9858 - val_head_4_loss: 0.0347 - val_head_5_accuracy: 0.9863 - val_head_5_loss: 0.0347 - val_head_6_accuracy: 0.9861 - val_head_6_loss: 0.0346 - val_head_7_accuracy: 0.9860 - val_head_7_loss: 0.0348 - val_loss: 0.2428 - learning_rate: 1.0362e-04\n","\n","Epoch 74: LearningRateScheduler setting learning rate to 9.653110200343824e-05.\n","Epoch 74/100\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 79ms/step - head_1_accuracy: 0.9805 - head_1_loss: 0.0553 - head_2_accuracy: 0.9805 - head_2_loss: 0.0552 - head_3_accuracy: 0.9805 - head_3_loss: 0.0552 - head_4_accuracy: 0.9807 - head_4_loss: 0.0553 - head_5_accuracy: 0.9808 - head_5_loss: 0.0552 - head_6_accuracy: 0.9805 - head_6_loss: 0.0552 - head_7_accuracy: 0.9807 - head_7_loss: 0.0553 - loss: 0.3867\n","Epoch 74: val_head_1_accuracy did not improve from 0.98634\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m27s\u001b[0m 85ms/step - head_1_accuracy: 0.9787 - head_1_loss: 0.0602 - head_2_accuracy: 0.9788 - head_2_loss: 0.0601 - head_3_accuracy: 0.9789 - head_3_loss: 0.0601 - head_4_accuracy: 0.9789 - head_4_loss: 0.0601 - head_5_accuracy: 0.9789 - head_5_loss: 0.0601 - head_6_accuracy: 0.9788 - head_6_loss: 0.0601 - head_7_accuracy: 0.9788 - head_7_loss: 0.0602 - loss: 0.4210 - val_head_1_accuracy: 0.9845 - val_head_1_loss: 0.0383 - val_head_2_accuracy: 0.9844 - val_head_2_loss: 0.0383 - val_head_3_accuracy: 0.9846 - val_head_3_loss: 0.0383 - val_head_4_accuracy: 0.9845 - val_head_4_loss: 0.0383 - val_head_5_accuracy: 0.9846 - val_head_5_loss: 0.0383 - val_head_6_accuracy: 0.9845 - val_head_6_loss: 0.0382 - val_head_7_accuracy: 0.9847 - val_head_7_loss: 0.0383 - val_loss: 0.2686 - learning_rate: 9.6531e-05\n","\n","Epoch 75: LearningRateScheduler setting learning rate to 8.96426074324889e-05.\n","Epoch 75/100\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 78ms/step - head_1_accuracy: 0.9787 - head_1_loss: 0.0589 - head_2_accuracy: 0.9786 - head_2_loss: 0.0590 - head_3_accuracy: 0.9786 - head_3_loss: 0.0590 - head_4_accuracy: 0.9788 - head_4_loss: 0.0591 - head_5_accuracy: 0.9788 - head_5_loss: 0.0591 - head_6_accuracy: 0.9787 - head_6_loss: 0.0590 - head_7_accuracy: 0.9786 - head_7_loss: 0.0588 - loss: 0.4130\n","Epoch 75: val_head_1_accuracy improved from 0.98634 to 0.98850, saving model to /kaggle/working/results/models/ViT_UNet_MultiHead_best.keras\n","\n","Epoch 75: finished saving model to /kaggle/working/results/models/ViT_UNet_MultiHead_best.keras\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m28s\u001b[0m 90ms/step - head_1_accuracy: 0.9784 - head_1_loss: 0.0589 - head_2_accuracy: 0.9785 - head_2_loss: 0.0588 - head_3_accuracy: 0.9785 - head_3_loss: 0.0588 - head_4_accuracy: 0.9785 - head_4_loss: 0.0588 - head_5_accuracy: 0.9785 - head_5_loss: 0.0590 - head_6_accuracy: 0.9785 - head_6_loss: 0.0588 - head_7_accuracy: 0.9785 - head_7_loss: 0.0588 - loss: 0.4120 - val_head_1_accuracy: 0.9885 - val_head_1_loss: 0.0303 - val_head_2_accuracy: 0.9883 - val_head_2_loss: 0.0302 - val_head_3_accuracy: 0.9883 - val_head_3_loss: 0.0302 - val_head_4_accuracy: 0.9883 - val_head_4_loss: 0.0303 - val_head_5_accuracy: 0.9883 - val_head_5_loss: 0.0303 - val_head_6_accuracy: 0.9883 - val_head_6_loss: 0.0302 - val_head_7_accuracy: 0.9883 - val_head_7_loss: 0.0303 - val_loss: 0.2092 - learning_rate: 8.9643e-05\n","\n","Epoch 76: LearningRateScheduler setting learning rate to 8.29659752703842e-05.\n","Epoch 76/100\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 78ms/step - head_1_accuracy: 0.9806 - head_1_loss: 0.0557 - head_2_accuracy: 0.9808 - head_2_loss: 0.0557 - head_3_accuracy: 0.9803 - head_3_loss: 0.0558 - head_4_accuracy: 0.9804 - head_4_loss: 0.0558 - head_5_accuracy: 0.9805 - head_5_loss: 0.0557 - head_6_accuracy: 0.9805 - head_6_loss: 0.0558 - head_7_accuracy: 0.9801 - head_7_loss: 0.0558 - loss: 0.3903\n","Epoch 76: val_head_1_accuracy improved from 0.98850 to 0.98976, saving model to /kaggle/working/results/models/ViT_UNet_MultiHead_best.keras\n","\n","Epoch 76: finished saving model to /kaggle/working/results/models/ViT_UNet_MultiHead_best.keras\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m28s\u001b[0m 90ms/step - head_1_accuracy: 0.9813 - head_1_loss: 0.0542 - head_2_accuracy: 0.9813 - head_2_loss: 0.0543 - head_3_accuracy: 0.9810 - head_3_loss: 0.0543 - head_4_accuracy: 0.9813 - head_4_loss: 0.0543 - head_5_accuracy: 0.9813 - head_5_loss: 0.0543 - head_6_accuracy: 0.9812 - head_6_loss: 0.0543 - head_7_accuracy: 0.9812 - head_7_loss: 0.0543 - loss: 0.3801 - val_head_1_accuracy: 0.9898 - val_head_1_loss: 0.0279 - val_head_2_accuracy: 0.9898 - val_head_2_loss: 0.0278 - val_head_3_accuracy: 0.9899 - val_head_3_loss: 0.0278 - val_head_4_accuracy: 0.9898 - val_head_4_loss: 0.0279 - val_head_5_accuracy: 0.9898 - val_head_5_loss: 0.0279 - val_head_6_accuracy: 0.9898 - val_head_6_loss: 0.0278 - val_head_7_accuracy: 0.9898 - val_head_7_loss: 0.0279 - val_loss: 0.1944 - learning_rate: 8.2966e-05\n","\n","Epoch 77: LearningRateScheduler setting learning rate to 7.650792831975739e-05.\n","Epoch 77/100\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 78ms/step - head_1_accuracy: 0.9812 - head_1_loss: 0.0527 - head_2_accuracy: 0.9810 - head_2_loss: 0.0528 - head_3_accuracy: 0.9814 - head_3_loss: 0.0528 - head_4_accuracy: 0.9811 - head_4_loss: 0.0529 - head_5_accuracy: 0.9812 - head_5_loss: 0.0528 - head_6_accuracy: 0.9813 - head_6_loss: 0.0527 - head_7_accuracy: 0.9813 - head_7_loss: 0.0528 - loss: 0.3695\n","Epoch 77: val_head_1_accuracy did not improve from 0.98976\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m27s\u001b[0m 84ms/step - head_1_accuracy: 0.9812 - head_1_loss: 0.0537 - head_2_accuracy: 0.9810 - head_2_loss: 0.0537 - head_3_accuracy: 0.9813 - head_3_loss: 0.0538 - head_4_accuracy: 0.9811 - head_4_loss: 0.0538 - head_5_accuracy: 0.9812 - head_5_loss: 0.0538 - head_6_accuracy: 0.9812 - head_6_loss: 0.0537 - head_7_accuracy: 0.9810 - head_7_loss: 0.0538 - loss: 0.3763 - val_head_1_accuracy: 0.9891 - val_head_1_loss: 0.0295 - val_head_2_accuracy: 0.9889 - val_head_2_loss: 0.0295 - val_head_3_accuracy: 0.9892 - val_head_3_loss: 0.0295 - val_head_4_accuracy: 0.9890 - val_head_4_loss: 0.0295 - val_head_5_accuracy: 0.9891 - val_head_5_loss: 0.0295 - val_head_6_accuracy: 0.9892 - val_head_6_loss: 0.0295 - val_head_7_accuracy: 0.9892 - val_head_7_loss: 0.0296 - val_loss: 0.2072 - learning_rate: 7.6508e-05\n","\n","Epoch 78: LearningRateScheduler setting learning rate to 7.027496928646257e-05.\n","Epoch 78/100\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 79ms/step - head_1_accuracy: 0.9825 - head_1_loss: 0.0497 - head_2_accuracy: 0.9825 - head_2_loss: 0.0497 - head_3_accuracy: 0.9824 - head_3_loss: 0.0497 - head_4_accuracy: 0.9825 - head_4_loss: 0.0498 - head_5_accuracy: 0.9825 - head_5_loss: 0.0497 - head_6_accuracy: 0.9826 - head_6_loss: 0.0497 - head_7_accuracy: 0.9825 - head_7_loss: 0.0497 - loss: 0.3480\n","Epoch 78: val_head_1_accuracy did not improve from 0.98976\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m27s\u001b[0m 85ms/step - head_1_accuracy: 0.9815 - head_1_loss: 0.0519 - head_2_accuracy: 0.9814 - head_2_loss: 0.0519 - head_3_accuracy: 0.9815 - head_3_loss: 0.0519 - head_4_accuracy: 0.9816 - head_4_loss: 0.0520 - head_5_accuracy: 0.9814 - head_5_loss: 0.0519 - head_6_accuracy: 0.9816 - head_6_loss: 0.0519 - head_7_accuracy: 0.9814 - head_7_loss: 0.0519 - loss: 0.3636 - val_head_1_accuracy: 0.9876 - val_head_1_loss: 0.0337 - val_head_2_accuracy: 0.9876 - val_head_2_loss: 0.0337 - val_head_3_accuracy: 0.9877 - val_head_3_loss: 0.0337 - val_head_4_accuracy: 0.9877 - val_head_4_loss: 0.0338 - val_head_5_accuracy: 0.9876 - val_head_5_loss: 0.0337 - val_head_6_accuracy: 0.9878 - val_head_6_loss: 0.0338 - val_head_7_accuracy: 0.9876 - val_head_7_loss: 0.0338 - val_loss: 0.2367 - learning_rate: 7.0275e-05\n","\n","Epoch 79: LearningRateScheduler setting learning rate to 6.427337423190096e-05.\n","Epoch 79/100\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 78ms/step - head_1_accuracy: 0.9821 - head_1_loss: 0.0505 - head_2_accuracy: 0.9818 - head_2_loss: 0.0505 - head_3_accuracy: 0.9818 - head_3_loss: 0.0508 - head_4_accuracy: 0.9822 - head_4_loss: 0.0506 - head_5_accuracy: 0.9821 - head_5_loss: 0.0505 - head_6_accuracy: 0.9823 - head_6_loss: 0.0505 - head_7_accuracy: 0.9820 - head_7_loss: 0.0505 - loss: 0.3538\n","Epoch 79: val_head_1_accuracy improved from 0.98976 to 0.98998, saving model to /kaggle/working/results/models/ViT_UNet_MultiHead_best.keras\n","\n","Epoch 79: finished saving model to /kaggle/working/results/models/ViT_UNet_MultiHead_best.keras\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m28s\u001b[0m 90ms/step - head_1_accuracy: 0.9821 - head_1_loss: 0.0506 - head_2_accuracy: 0.9818 - head_2_loss: 0.0505 - head_3_accuracy: 0.9820 - head_3_loss: 0.0506 - head_4_accuracy: 0.9821 - head_4_loss: 0.0506 - head_5_accuracy: 0.9821 - head_5_loss: 0.0506 - head_6_accuracy: 0.9821 - head_6_loss: 0.0505 - head_7_accuracy: 0.9820 - head_7_loss: 0.0506 - loss: 0.3542 - val_head_1_accuracy: 0.9900 - val_head_1_loss: 0.0296 - val_head_2_accuracy: 0.9900 - val_head_2_loss: 0.0296 - val_head_3_accuracy: 0.9900 - val_head_3_loss: 0.0296 - val_head_4_accuracy: 0.9899 - val_head_4_loss: 0.0297 - val_head_5_accuracy: 0.9899 - val_head_5_loss: 0.0296 - val_head_6_accuracy: 0.9898 - val_head_6_loss: 0.0296 - val_head_7_accuracy: 0.9899 - val_head_7_loss: 0.0297 - val_loss: 0.2054 - learning_rate: 6.4273e-05\n","\n","Epoch 80: LearningRateScheduler setting learning rate to 5.850918625355897e-05.\n","Epoch 80/100\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 78ms/step - head_1_accuracy: 0.9809 - head_1_loss: 0.0524 - head_2_accuracy: 0.9812 - head_2_loss: 0.0524 - head_3_accuracy: 0.9811 - head_3_loss: 0.0524 - head_4_accuracy: 0.9811 - head_4_loss: 0.0524 - head_5_accuracy: 0.9813 - head_5_loss: 0.0524 - head_6_accuracy: 0.9813 - head_6_loss: 0.0524 - head_7_accuracy: 0.9812 - head_7_loss: 0.0524 - loss: 0.3667\n","Epoch 80: val_head_1_accuracy did not improve from 0.98998\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m27s\u001b[0m 85ms/step - head_1_accuracy: 0.9809 - head_1_loss: 0.0531 - head_2_accuracy: 0.9810 - head_2_loss: 0.0531 - head_3_accuracy: 0.9809 - head_3_loss: 0.0531 - head_4_accuracy: 0.9810 - head_4_loss: 0.0531 - head_5_accuracy: 0.9811 - head_5_loss: 0.0532 - head_6_accuracy: 0.9811 - head_6_loss: 0.0531 - head_7_accuracy: 0.9810 - head_7_loss: 0.0531 - loss: 0.3720 - val_head_1_accuracy: 0.9885 - val_head_1_loss: 0.0327 - val_head_2_accuracy: 0.9884 - val_head_2_loss: 0.0327 - val_head_3_accuracy: 0.9885 - val_head_3_loss: 0.0326 - val_head_4_accuracy: 0.9883 - val_head_4_loss: 0.0328 - val_head_5_accuracy: 0.9883 - val_head_5_loss: 0.0327 - val_head_6_accuracy: 0.9883 - val_head_6_loss: 0.0328 - val_head_7_accuracy: 0.9884 - val_head_7_loss: 0.0327 - val_loss: 0.2302 - learning_rate: 5.8509e-05\n","\n","Epoch 81: LearningRateScheduler setting learning rate to 5.298820940012169e-05.\n","Epoch 81/100\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 79ms/step - head_1_accuracy: 0.9834 - head_1_loss: 0.0467 - head_2_accuracy: 0.9836 - head_2_loss: 0.0465 - head_3_accuracy: 0.9835 - head_3_loss: 0.0465 - head_4_accuracy: 0.9833 - head_4_loss: 0.0466 - head_5_accuracy: 0.9831 - head_5_loss: 0.0467 - head_6_accuracy: 0.9833 - head_6_loss: 0.0466 - head_7_accuracy: 0.9836 - head_7_loss: 0.0465 - loss: 0.3261\n","Epoch 81: val_head_1_accuracy improved from 0.98998 to 0.99147, saving model to /kaggle/working/results/models/ViT_UNet_MultiHead_best.keras\n","\n","Epoch 81: finished saving model to /kaggle/working/results/models/ViT_UNet_MultiHead_best.keras\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m28s\u001b[0m 90ms/step - head_1_accuracy: 0.9835 - head_1_loss: 0.0467 - head_2_accuracy: 0.9836 - head_2_loss: 0.0466 - head_3_accuracy: 0.9836 - head_3_loss: 0.0466 - head_4_accuracy: 0.9835 - head_4_loss: 0.0467 - head_5_accuracy: 0.9834 - head_5_loss: 0.0467 - head_6_accuracy: 0.9834 - head_6_loss: 0.0466 - head_7_accuracy: 0.9836 - head_7_loss: 0.0466 - loss: 0.3266 - val_head_1_accuracy: 0.9915 - val_head_1_loss: 0.0241 - val_head_2_accuracy: 0.9915 - val_head_2_loss: 0.0241 - val_head_3_accuracy: 0.9916 - val_head_3_loss: 0.0241 - val_head_4_accuracy: 0.9915 - val_head_4_loss: 0.0241 - val_head_5_accuracy: 0.9916 - val_head_5_loss: 0.0242 - val_head_6_accuracy: 0.9915 - val_head_6_loss: 0.0241 - val_head_7_accuracy: 0.9916 - val_head_7_loss: 0.0242 - val_loss: 0.1692 - learning_rate: 5.2988e-05\n","\n","Epoch 82: LearningRateScheduler setting learning rate to 4.7716002827287234e-05.\n","Epoch 82/100\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 79ms/step - head_1_accuracy: 0.9853 - head_1_loss: 0.0441 - head_2_accuracy: 0.9852 - head_2_loss: 0.0441 - head_3_accuracy: 0.9851 - head_3_loss: 0.0441 - head_4_accuracy: 0.9852 - head_4_loss: 0.0443 - head_5_accuracy: 0.9851 - head_5_loss: 0.0442 - head_6_accuracy: 0.9851 - head_6_loss: 0.0440 - head_7_accuracy: 0.9850 - head_7_loss: 0.0440 - loss: 0.3088\n","Epoch 82: val_head_1_accuracy improved from 0.99147 to 0.99198, saving model to /kaggle/working/results/models/ViT_UNet_MultiHead_best.keras\n","\n","Epoch 82: finished saving model to /kaggle/working/results/models/ViT_UNet_MultiHead_best.keras\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m28s\u001b[0m 90ms/step - head_1_accuracy: 0.9847 - head_1_loss: 0.0440 - head_2_accuracy: 0.9847 - head_2_loss: 0.0440 - head_3_accuracy: 0.9846 - head_3_loss: 0.0440 - head_4_accuracy: 0.9846 - head_4_loss: 0.0441 - head_5_accuracy: 0.9846 - head_5_loss: 0.0440 - head_6_accuracy: 0.9847 - head_6_loss: 0.0439 - head_7_accuracy: 0.9846 - head_7_loss: 0.0440 - loss: 0.3082 - val_head_1_accuracy: 0.9920 - val_head_1_loss: 0.0214 - val_head_2_accuracy: 0.9920 - val_head_2_loss: 0.0214 - val_head_3_accuracy: 0.9920 - val_head_3_loss: 0.0214 - val_head_4_accuracy: 0.9921 - val_head_4_loss: 0.0215 - val_head_5_accuracy: 0.9920 - val_head_5_loss: 0.0214 - val_head_6_accuracy: 0.9920 - val_head_6_loss: 0.0214 - val_head_7_accuracy: 0.9921 - val_head_7_loss: 0.0215 - val_loss: 0.1503 - learning_rate: 4.7716e-05\n","\n","Epoch 83: LearningRateScheduler setting learning rate to 4.26978752001687e-05.\n","Epoch 83/100\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 79ms/step - head_1_accuracy: 0.9845 - head_1_loss: 0.0446 - head_2_accuracy: 0.9843 - head_2_loss: 0.0448 - head_3_accuracy: 0.9843 - head_3_loss: 0.0448 - head_4_accuracy: 0.9845 - head_4_loss: 0.0447 - head_5_accuracy: 0.9846 - head_5_loss: 0.0447 - head_6_accuracy: 0.9843 - head_6_loss: 0.0448 - head_7_accuracy: 0.9843 - head_7_loss: 0.0447 - loss: 0.3131\n","Epoch 83: val_head_1_accuracy improved from 0.99198 to 0.99243, saving model to /kaggle/working/results/models/ViT_UNet_MultiHead_best.keras\n","\n","Epoch 83: finished saving model to /kaggle/working/results/models/ViT_UNet_MultiHead_best.keras\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m28s\u001b[0m 90ms/step - head_1_accuracy: 0.9842 - head_1_loss: 0.0449 - head_2_accuracy: 0.9841 - head_2_loss: 0.0450 - head_3_accuracy: 0.9840 - head_3_loss: 0.0449 - head_4_accuracy: 0.9843 - head_4_loss: 0.0449 - head_5_accuracy: 0.9842 - head_5_loss: 0.0450 - head_6_accuracy: 0.9839 - head_6_loss: 0.0449 - head_7_accuracy: 0.9839 - head_7_loss: 0.0449 - loss: 0.3146 - val_head_1_accuracy: 0.9924 - val_head_1_loss: 0.0229 - val_head_2_accuracy: 0.9925 - val_head_2_loss: 0.0229 - val_head_3_accuracy: 0.9924 - val_head_3_loss: 0.0230 - val_head_4_accuracy: 0.9924 - val_head_4_loss: 0.0230 - val_head_5_accuracy: 0.9924 - val_head_5_loss: 0.0229 - val_head_6_accuracy: 0.9926 - val_head_6_loss: 0.0230 - val_head_7_accuracy: 0.9923 - val_head_7_loss: 0.0230 - val_loss: 0.1612 - learning_rate: 4.2698e-05\n","\n","Epoch 84: LearningRateScheduler setting learning rate to 3.793887934791801e-05.\n","Epoch 84/100\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 79ms/step - head_1_accuracy: 0.9843 - head_1_loss: 0.0416 - head_2_accuracy: 0.9844 - head_2_loss: 0.0416 - head_3_accuracy: 0.9843 - head_3_loss: 0.0415 - head_4_accuracy: 0.9840 - head_4_loss: 0.0417 - head_5_accuracy: 0.9846 - head_5_loss: 0.0415 - head_6_accuracy: 0.9841 - head_6_loss: 0.0416 - head_7_accuracy: 0.9847 - head_7_loss: 0.0415 - loss: 0.2912\n","Epoch 84: val_head_1_accuracy improved from 0.99243 to 0.99273, saving model to /kaggle/working/results/models/ViT_UNet_MultiHead_best.keras\n","\n","Epoch 84: finished saving model to /kaggle/working/results/models/ViT_UNet_MultiHead_best.keras\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m29s\u001b[0m 91ms/step - head_1_accuracy: 0.9842 - head_1_loss: 0.0427 - head_2_accuracy: 0.9842 - head_2_loss: 0.0427 - head_3_accuracy: 0.9842 - head_3_loss: 0.0426 - head_4_accuracy: 0.9840 - head_4_loss: 0.0427 - head_5_accuracy: 0.9843 - head_5_loss: 0.0426 - head_6_accuracy: 0.9840 - head_6_loss: 0.0427 - head_7_accuracy: 0.9843 - head_7_loss: 0.0426 - loss: 0.2988 - val_head_1_accuracy: 0.9927 - val_head_1_loss: 0.0213 - val_head_2_accuracy: 0.9927 - val_head_2_loss: 0.0213 - val_head_3_accuracy: 0.9927 - val_head_3_loss: 0.0214 - val_head_4_accuracy: 0.9927 - val_head_4_loss: 0.0214 - val_head_5_accuracy: 0.9927 - val_head_5_loss: 0.0214 - val_head_6_accuracy: 0.9927 - val_head_6_loss: 0.0213 - val_head_7_accuracy: 0.9927 - val_head_7_loss: 0.0214 - val_loss: 0.1498 - learning_rate: 3.7939e-05\n","\n","Epoch 85: LearningRateScheduler setting learning rate to 3.344380717595574e-05.\n","Epoch 85/100\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 79ms/step - head_1_accuracy: 0.9864 - head_1_loss: 0.0382 - head_2_accuracy: 0.9862 - head_2_loss: 0.0382 - head_3_accuracy: 0.9863 - head_3_loss: 0.0382 - head_4_accuracy: 0.9861 - head_4_loss: 0.0382 - head_5_accuracy: 0.9863 - head_5_loss: 0.0382 - head_6_accuracy: 0.9861 - head_6_loss: 0.0381 - head_7_accuracy: 0.9865 - head_7_loss: 0.0382 - loss: 0.2673\n","Epoch 85: val_head_1_accuracy did not improve from 0.99273\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m27s\u001b[0m 85ms/step - head_1_accuracy: 0.9857 - head_1_loss: 0.0401 - head_2_accuracy: 0.9854 - head_2_loss: 0.0401 - head_3_accuracy: 0.9856 - head_3_loss: 0.0401 - head_4_accuracy: 0.9854 - head_4_loss: 0.0401 - head_5_accuracy: 0.9855 - head_5_loss: 0.0401 - head_6_accuracy: 0.9855 - head_6_loss: 0.0400 - head_7_accuracy: 0.9856 - head_7_loss: 0.0400 - loss: 0.2805 - val_head_1_accuracy: 0.9920 - val_head_1_loss: 0.0209 - val_head_2_accuracy: 0.9919 - val_head_2_loss: 0.0209 - val_head_3_accuracy: 0.9920 - val_head_3_loss: 0.0209 - val_head_4_accuracy: 0.9921 - val_head_4_loss: 0.0210 - val_head_5_accuracy: 0.9920 - val_head_5_loss: 0.0209 - val_head_6_accuracy: 0.9918 - val_head_6_loss: 0.0210 - val_head_7_accuracy: 0.9918 - val_head_7_loss: 0.0210 - val_loss: 0.1472 - learning_rate: 3.3444e-05\n","\n","Epoch 86: LearningRateScheduler setting learning rate to 2.9217184840928013e-05.\n","Epoch 86/100\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 79ms/step - head_1_accuracy: 0.9869 - head_1_loss: 0.0386 - head_2_accuracy: 0.9871 - head_2_loss: 0.0384 - head_3_accuracy: 0.9870 - head_3_loss: 0.0385 - head_4_accuracy: 0.9870 - head_4_loss: 0.0386 - head_5_accuracy: 0.9871 - head_5_loss: 0.0384 - head_6_accuracy: 0.9871 - head_6_loss: 0.0385 - head_7_accuracy: 0.9870 - head_7_loss: 0.0384 - loss: 0.2695\n","Epoch 86: val_head_1_accuracy improved from 0.99273 to 0.99362, saving model to /kaggle/working/results/models/ViT_UNet_MultiHead_best.keras\n","\n","Epoch 86: finished saving model to /kaggle/working/results/models/ViT_UNet_MultiHead_best.keras\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m28s\u001b[0m 90ms/step - head_1_accuracy: 0.9870 - head_1_loss: 0.0388 - head_2_accuracy: 0.9872 - head_2_loss: 0.0387 - head_3_accuracy: 0.9871 - head_3_loss: 0.0388 - head_4_accuracy: 0.9871 - head_4_loss: 0.0388 - head_5_accuracy: 0.9872 - head_5_loss: 0.0387 - head_6_accuracy: 0.9872 - head_6_loss: 0.0388 - head_7_accuracy: 0.9871 - head_7_loss: 0.0388 - loss: 0.2716 - val_head_1_accuracy: 0.9936 - val_head_1_loss: 0.0189 - val_head_2_accuracy: 0.9935 - val_head_2_loss: 0.0189 - val_head_3_accuracy: 0.9935 - val_head_3_loss: 0.0189 - val_head_4_accuracy: 0.9935 - val_head_4_loss: 0.0190 - val_head_5_accuracy: 0.9934 - val_head_5_loss: 0.0189 - val_head_6_accuracy: 0.9936 - val_head_6_loss: 0.0189 - val_head_7_accuracy: 0.9935 - val_head_7_loss: 0.0189 - val_loss: 0.1330 - learning_rate: 2.9217e-05\n","\n","Epoch 87: LearningRateScheduler setting learning rate to 2.526326819325076e-05.\n","Epoch 87/100\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 79ms/step - head_1_accuracy: 0.9864 - head_1_loss: 0.0404 - head_2_accuracy: 0.9861 - head_2_loss: 0.0404 - head_3_accuracy: 0.9862 - head_3_loss: 0.0404 - head_4_accuracy: 0.9862 - head_4_loss: 0.0403 - head_5_accuracy: 0.9862 - head_5_loss: 0.0403 - head_6_accuracy: 0.9862 - head_6_loss: 0.0403 - head_7_accuracy: 0.9861 - head_7_loss: 0.0403 - loss: 0.2824\n","Epoch 87: val_head_1_accuracy did not improve from 0.99362\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m27s\u001b[0m 85ms/step - head_1_accuracy: 0.9860 - head_1_loss: 0.0408 - head_2_accuracy: 0.9858 - head_2_loss: 0.0408 - head_3_accuracy: 0.9859 - head_3_loss: 0.0408 - head_4_accuracy: 0.9858 - head_4_loss: 0.0408 - head_5_accuracy: 0.9858 - head_5_loss: 0.0408 - head_6_accuracy: 0.9859 - head_6_loss: 0.0408 - head_7_accuracy: 0.9858 - head_7_loss: 0.0407 - loss: 0.2856 - val_head_1_accuracy: 0.9924 - val_head_1_loss: 0.0209 - val_head_2_accuracy: 0.9922 - val_head_2_loss: 0.0209 - val_head_3_accuracy: 0.9923 - val_head_3_loss: 0.0209 - val_head_4_accuracy: 0.9922 - val_head_4_loss: 0.0209 - val_head_5_accuracy: 0.9923 - val_head_5_loss: 0.0209 - val_head_6_accuracy: 0.9923 - val_head_6_loss: 0.0210 - val_head_7_accuracy: 0.9922 - val_head_7_loss: 0.0210 - val_loss: 0.1473 - learning_rate: 2.5263e-05\n","\n","Epoch 88: LearningRateScheduler setting learning rate to 2.1586038491829064e-05.\n","Epoch 88/100\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 79ms/step - head_1_accuracy: 0.9879 - head_1_loss: 0.0368 - head_2_accuracy: 0.9879 - head_2_loss: 0.0368 - head_3_accuracy: 0.9879 - head_3_loss: 0.0367 - head_4_accuracy: 0.9880 - head_4_loss: 0.0368 - head_5_accuracy: 0.9879 - head_5_loss: 0.0367 - head_6_accuracy: 0.9879 - head_6_loss: 0.0368 - head_7_accuracy: 0.9880 - head_7_loss: 0.0368 - loss: 0.2574\n","Epoch 88: val_head_1_accuracy did not improve from 0.99362\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m27s\u001b[0m 85ms/step - head_1_accuracy: 0.9866 - head_1_loss: 0.0387 - head_2_accuracy: 0.9867 - head_2_loss: 0.0386 - head_3_accuracy: 0.9866 - head_3_loss: 0.0387 - head_4_accuracy: 0.9867 - head_4_loss: 0.0387 - head_5_accuracy: 0.9867 - head_5_loss: 0.0387 - head_6_accuracy: 0.9867 - head_6_loss: 0.0387 - head_7_accuracy: 0.9868 - head_7_loss: 0.0387 - loss: 0.2709 - val_head_1_accuracy: 0.9931 - val_head_1_loss: 0.0194 - val_head_2_accuracy: 0.9931 - val_head_2_loss: 0.0194 - val_head_3_accuracy: 0.9931 - val_head_3_loss: 0.0194 - val_head_4_accuracy: 0.9930 - val_head_4_loss: 0.0195 - val_head_5_accuracy: 0.9932 - val_head_5_loss: 0.0194 - val_head_6_accuracy: 0.9930 - val_head_6_loss: 0.0194 - val_head_7_accuracy: 0.9931 - val_head_7_loss: 0.0195 - val_loss: 0.1367 - learning_rate: 2.1586e-05\n","\n","Epoch 89: LearningRateScheduler setting learning rate to 1.8189198395266834e-05.\n","Epoch 89/100\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 79ms/step - head_1_accuracy: 0.9876 - head_1_loss: 0.0363 - head_2_accuracy: 0.9875 - head_2_loss: 0.0363 - head_3_accuracy: 0.9875 - head_3_loss: 0.0363 - head_4_accuracy: 0.9874 - head_4_loss: 0.0363 - head_5_accuracy: 0.9876 - head_5_loss: 0.0363 - head_6_accuracy: 0.9874 - head_6_loss: 0.0363 - head_7_accuracy: 0.9878 - head_7_loss: 0.0363 - loss: 0.2541\n","Epoch 89: val_head_1_accuracy improved from 0.99362 to 0.99384, saving model to /kaggle/working/results/models/ViT_UNet_MultiHead_best.keras\n","\n","Epoch 89: finished saving model to /kaggle/working/results/models/ViT_UNet_MultiHead_best.keras\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m28s\u001b[0m 90ms/step - head_1_accuracy: 0.9874 - head_1_loss: 0.0370 - head_2_accuracy: 0.9873 - head_2_loss: 0.0369 - head_3_accuracy: 0.9872 - head_3_loss: 0.0369 - head_4_accuracy: 0.9873 - head_4_loss: 0.0369 - head_5_accuracy: 0.9874 - head_5_loss: 0.0369 - head_6_accuracy: 0.9873 - head_6_loss: 0.0369 - head_7_accuracy: 0.9875 - head_7_loss: 0.0369 - loss: 0.2586 - val_head_1_accuracy: 0.9938 - val_head_1_loss: 0.0188 - val_head_2_accuracy: 0.9937 - val_head_2_loss: 0.0189 - val_head_3_accuracy: 0.9938 - val_head_3_loss: 0.0189 - val_head_4_accuracy: 0.9939 - val_head_4_loss: 0.0189 - val_head_5_accuracy: 0.9938 - val_head_5_loss: 0.0189 - val_head_6_accuracy: 0.9938 - val_head_6_loss: 0.0189 - val_head_7_accuracy: 0.9938 - val_head_7_loss: 0.0189 - val_loss: 0.1326 - learning_rate: 1.8189e-05\n","\n","Epoch 90: LearningRateScheduler setting learning rate to 1.5076168233603417e-05.\n","Epoch 90/100\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 79ms/step - head_1_accuracy: 0.9868 - head_1_loss: 0.0377 - head_2_accuracy: 0.9864 - head_2_loss: 0.0378 - head_3_accuracy: 0.9867 - head_3_loss: 0.0377 - head_4_accuracy: 0.9866 - head_4_loss: 0.0376 - head_5_accuracy: 0.9864 - head_5_loss: 0.0377 - head_6_accuracy: 0.9866 - head_6_loss: 0.0377 - head_7_accuracy: 0.9863 - head_7_loss: 0.0377 - loss: 0.2639\n","Epoch 90: val_head_1_accuracy did not improve from 0.99384\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m27s\u001b[0m 85ms/step - head_1_accuracy: 0.9876 - head_1_loss: 0.0363 - head_2_accuracy: 0.9874 - head_2_loss: 0.0363 - head_3_accuracy: 0.9876 - head_3_loss: 0.0362 - head_4_accuracy: 0.9874 - head_4_loss: 0.0362 - head_5_accuracy: 0.9875 - head_5_loss: 0.0363 - head_6_accuracy: 0.9875 - head_6_loss: 0.0362 - head_7_accuracy: 0.9873 - head_7_loss: 0.0362 - loss: 0.2539 - val_head_1_accuracy: 0.9937 - val_head_1_loss: 0.0179 - val_head_2_accuracy: 0.9937 - val_head_2_loss: 0.0179 - val_head_3_accuracy: 0.9938 - val_head_3_loss: 0.0179 - val_head_4_accuracy: 0.9938 - val_head_4_loss: 0.0179 - val_head_5_accuracy: 0.9938 - val_head_5_loss: 0.0179 - val_head_6_accuracy: 0.9937 - val_head_6_loss: 0.0179 - val_head_7_accuracy: 0.9937 - val_head_7_loss: 0.0179 - val_loss: 0.1260 - learning_rate: 1.5076e-05\n","\n","Epoch 91: LearningRateScheduler setting learning rate to 1.2250082564331516e-05.\n","Epoch 91/100\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 79ms/step - head_1_accuracy: 0.9886 - head_1_loss: 0.0337 - head_2_accuracy: 0.9889 - head_2_loss: 0.0336 - head_3_accuracy: 0.9889 - head_3_loss: 0.0336 - head_4_accuracy: 0.9890 - head_4_loss: 0.0338 - head_5_accuracy: 0.9889 - head_5_loss: 0.0337 - head_6_accuracy: 0.9888 - head_6_loss: 0.0337 - head_7_accuracy: 0.9887 - head_7_loss: 0.0336 - loss: 0.2356\n","Epoch 91: val_head_1_accuracy did not improve from 0.99384\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m27s\u001b[0m 86ms/step - head_1_accuracy: 0.9880 - head_1_loss: 0.0342 - head_2_accuracy: 0.9879 - head_2_loss: 0.0341 - head_3_accuracy: 0.9880 - head_3_loss: 0.0341 - head_4_accuracy: 0.9882 - head_4_loss: 0.0341 - head_5_accuracy: 0.9881 - head_5_loss: 0.0341 - head_6_accuracy: 0.9880 - head_6_loss: 0.0341 - head_7_accuracy: 0.9879 - head_7_loss: 0.0341 - loss: 0.2386 - val_head_1_accuracy: 0.9935 - val_head_1_loss: 0.0180 - val_head_2_accuracy: 0.9934 - val_head_2_loss: 0.0180 - val_head_3_accuracy: 0.9935 - val_head_3_loss: 0.0180 - val_head_4_accuracy: 0.9935 - val_head_4_loss: 0.0181 - val_head_5_accuracy: 0.9935 - val_head_5_loss: 0.0180 - val_head_6_accuracy: 0.9933 - val_head_6_loss: 0.0181 - val_head_7_accuracy: 0.9935 - val_head_7_loss: 0.0181 - val_loss: 0.1270 - learning_rate: 1.2250e-05\n","\n","Epoch 92: LearningRateScheduler setting learning rate to 9.713787016162943e-06.\n","Epoch 92/100\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 79ms/step - head_1_accuracy: 0.9871 - head_1_loss: 0.0378 - head_2_accuracy: 0.9871 - head_2_loss: 0.0378 - head_3_accuracy: 0.9871 - head_3_loss: 0.0379 - head_4_accuracy: 0.9873 - head_4_loss: 0.0380 - head_5_accuracy: 0.9871 - head_5_loss: 0.0379 - head_6_accuracy: 0.9873 - head_6_loss: 0.0379 - head_7_accuracy: 0.9869 - head_7_loss: 0.0379 - loss: 0.2653\n","Epoch 92: val_head_1_accuracy improved from 0.99384 to 0.99451, saving model to /kaggle/working/results/models/ViT_UNet_MultiHead_best.keras\n","\n","Epoch 92: finished saving model to /kaggle/working/results/models/ViT_UNet_MultiHead_best.keras\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m28s\u001b[0m 90ms/step - head_1_accuracy: 0.9874 - head_1_loss: 0.0362 - head_2_accuracy: 0.9876 - head_2_loss: 0.0362 - head_3_accuracy: 0.9875 - head_3_loss: 0.0362 - head_4_accuracy: 0.9876 - head_4_loss: 0.0363 - head_5_accuracy: 0.9875 - head_5_loss: 0.0362 - head_6_accuracy: 0.9875 - head_6_loss: 0.0362 - head_7_accuracy: 0.9874 - head_7_loss: 0.0362 - loss: 0.2536 - val_head_1_accuracy: 0.9945 - val_head_1_loss: 0.0169 - val_head_2_accuracy: 0.9945 - val_head_2_loss: 0.0170 - val_head_3_accuracy: 0.9944 - val_head_3_loss: 0.0169 - val_head_4_accuracy: 0.9944 - val_head_4_loss: 0.0170 - val_head_5_accuracy: 0.9944 - val_head_5_loss: 0.0170 - val_head_6_accuracy: 0.9945 - val_head_6_loss: 0.0170 - val_head_7_accuracy: 0.9944 - val_head_7_loss: 0.0170 - val_loss: 0.1194 - learning_rate: 9.7138e-06\n","\n","Epoch 93: LearningRateScheduler setting learning rate to 7.469835423722156e-06.\n","Epoch 93/100\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 79ms/step - head_1_accuracy: 0.9891 - head_1_loss: 0.0333 - head_2_accuracy: 0.9891 - head_2_loss: 0.0334 - head_3_accuracy: 0.9889 - head_3_loss: 0.0336 - head_4_accuracy: 0.9889 - head_4_loss: 0.0337 - head_5_accuracy: 0.9890 - head_5_loss: 0.0334 - head_6_accuracy: 0.9892 - head_6_loss: 0.0337 - head_7_accuracy: 0.9892 - head_7_loss: 0.0333 - loss: 0.2343\n","Epoch 93: val_head_1_accuracy did not improve from 0.99451\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m27s\u001b[0m 86ms/step - head_1_accuracy: 0.9886 - head_1_loss: 0.0333 - head_2_accuracy: 0.9887 - head_2_loss: 0.0333 - head_3_accuracy: 0.9886 - head_3_loss: 0.0334 - head_4_accuracy: 0.9885 - head_4_loss: 0.0334 - head_5_accuracy: 0.9887 - head_5_loss: 0.0333 - head_6_accuracy: 0.9887 - head_6_loss: 0.0334 - head_7_accuracy: 0.9887 - head_7_loss: 0.0332 - loss: 0.2335 - val_head_1_accuracy: 0.9941 - val_head_1_loss: 0.0172 - val_head_2_accuracy: 0.9941 - val_head_2_loss: 0.0172 - val_head_3_accuracy: 0.9941 - val_head_3_loss: 0.0172 - val_head_4_accuracy: 0.9940 - val_head_4_loss: 0.0172 - val_head_5_accuracy: 0.9941 - val_head_5_loss: 0.0172 - val_head_6_accuracy: 0.9942 - val_head_6_loss: 0.0172 - val_head_7_accuracy: 0.9941 - val_head_7_loss: 0.0172 - val_loss: 0.1210 - learning_rate: 7.4698e-06\n","\n","Epoch 94: LearningRateScheduler setting learning rate to 5.520487256051157e-06.\n","Epoch 94/100\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 79ms/step - head_1_accuracy: 0.9888 - head_1_loss: 0.0344 - head_2_accuracy: 0.9890 - head_2_loss: 0.0344 - head_3_accuracy: 0.9889 - head_3_loss: 0.0344 - head_4_accuracy: 0.9886 - head_4_loss: 0.0345 - head_5_accuracy: 0.9887 - head_5_loss: 0.0345 - head_6_accuracy: 0.9889 - head_6_loss: 0.0344 - head_7_accuracy: 0.9889 - head_7_loss: 0.0344 - loss: 0.2409\n","Epoch 94: val_head_1_accuracy did not improve from 0.99451\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m27s\u001b[0m 86ms/step - head_1_accuracy: 0.9885 - head_1_loss: 0.0337 - head_2_accuracy: 0.9886 - head_2_loss: 0.0337 - head_3_accuracy: 0.9885 - head_3_loss: 0.0337 - head_4_accuracy: 0.9883 - head_4_loss: 0.0338 - head_5_accuracy: 0.9884 - head_5_loss: 0.0338 - head_6_accuracy: 0.9886 - head_6_loss: 0.0337 - head_7_accuracy: 0.9886 - head_7_loss: 0.0337 - loss: 0.2361 - val_head_1_accuracy: 0.9942 - val_head_1_loss: 0.0171 - val_head_2_accuracy: 0.9941 - val_head_2_loss: 0.0171 - val_head_3_accuracy: 0.9941 - val_head_3_loss: 0.0171 - val_head_4_accuracy: 0.9941 - val_head_4_loss: 0.0171 - val_head_5_accuracy: 0.9941 - val_head_5_loss: 0.0171 - val_head_6_accuracy: 0.9942 - val_head_6_loss: 0.0171 - val_head_7_accuracy: 0.9942 - val_head_7_loss: 0.0171 - val_loss: 0.1205 - learning_rate: 5.5205e-06\n","\n","Epoch 95: LearningRateScheduler setting learning rate to 3.867705341515533e-06.\n","Epoch 95/100\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 79ms/step - head_1_accuracy: 0.9888 - head_1_loss: 0.0318 - head_2_accuracy: 0.9887 - head_2_loss: 0.0318 - head_3_accuracy: 0.9887 - head_3_loss: 0.0317 - head_4_accuracy: 0.9888 - head_4_loss: 0.0316 - head_5_accuracy: 0.9886 - head_5_loss: 0.0319 - head_6_accuracy: 0.9887 - head_6_loss: 0.0317 - head_7_accuracy: 0.9887 - head_7_loss: 0.0318 - loss: 0.2222\n","Epoch 95: val_head_1_accuracy improved from 0.99451 to 0.99466, saving model to /kaggle/working/results/models/ViT_UNet_MultiHead_best.keras\n","\n","Epoch 95: finished saving model to /kaggle/working/results/models/ViT_UNet_MultiHead_best.keras\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m29s\u001b[0m 90ms/step - head_1_accuracy: 0.9885 - head_1_loss: 0.0330 - head_2_accuracy: 0.9884 - head_2_loss: 0.0330 - head_3_accuracy: 0.9885 - head_3_loss: 0.0329 - head_4_accuracy: 0.9885 - head_4_loss: 0.0329 - head_5_accuracy: 0.9883 - head_5_loss: 0.0332 - head_6_accuracy: 0.9884 - head_6_loss: 0.0330 - head_7_accuracy: 0.9884 - head_7_loss: 0.0331 - loss: 0.2313 - val_head_1_accuracy: 0.9947 - val_head_1_loss: 0.0166 - val_head_2_accuracy: 0.9946 - val_head_2_loss: 0.0166 - val_head_3_accuracy: 0.9946 - val_head_3_loss: 0.0166 - val_head_4_accuracy: 0.9946 - val_head_4_loss: 0.0166 - val_head_5_accuracy: 0.9946 - val_head_5_loss: 0.0166 - val_head_6_accuracy: 0.9946 - val_head_6_loss: 0.0166 - val_head_7_accuracy: 0.9946 - val_head_7_loss: 0.0166 - val_loss: 0.1167 - learning_rate: 3.8677e-06\n","\n","Epoch 96: LearningRateScheduler setting learning rate to 2.5131538914029374e-06.\n","Epoch 96/100\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 79ms/step - head_1_accuracy: 0.9893 - head_1_loss: 0.0320 - head_2_accuracy: 0.9893 - head_2_loss: 0.0322 - head_3_accuracy: 0.9892 - head_3_loss: 0.0321 - head_4_accuracy: 0.9893 - head_4_loss: 0.0322 - head_5_accuracy: 0.9891 - head_5_loss: 0.0322 - head_6_accuracy: 0.9892 - head_6_loss: 0.0322 - head_7_accuracy: 0.9891 - head_7_loss: 0.0321 - loss: 0.2251\n","Epoch 96: val_head_1_accuracy did not improve from 0.99466\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m27s\u001b[0m 86ms/step - head_1_accuracy: 0.9889 - head_1_loss: 0.0318 - head_2_accuracy: 0.9890 - head_2_loss: 0.0318 - head_3_accuracy: 0.9889 - head_3_loss: 0.0318 - head_4_accuracy: 0.9890 - head_4_loss: 0.0318 - head_5_accuracy: 0.9889 - head_5_loss: 0.0319 - head_6_accuracy: 0.9889 - head_6_loss: 0.0319 - head_7_accuracy: 0.9889 - head_7_loss: 0.0318 - loss: 0.2228 - val_head_1_accuracy: 0.9945 - val_head_1_loss: 0.0168 - val_head_2_accuracy: 0.9945 - val_head_2_loss: 0.0168 - val_head_3_accuracy: 0.9945 - val_head_3_loss: 0.0168 - val_head_4_accuracy: 0.9945 - val_head_4_loss: 0.0168 - val_head_5_accuracy: 0.9945 - val_head_5_loss: 0.0168 - val_head_6_accuracy: 0.9945 - val_head_6_loss: 0.0168 - val_head_7_accuracy: 0.9946 - val_head_7_loss: 0.0168 - val_loss: 0.1184 - learning_rate: 2.5132e-06\n","\n","Epoch 97: LearningRateScheduler setting learning rate to 1.4581968242032775e-06.\n","Epoch 97/100\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 79ms/step - head_1_accuracy: 0.9883 - head_1_loss: 0.0315 - head_2_accuracy: 0.9882 - head_2_loss: 0.0317 - head_3_accuracy: 0.9880 - head_3_loss: 0.0316 - head_4_accuracy: 0.9882 - head_4_loss: 0.0319 - head_5_accuracy: 0.9883 - head_5_loss: 0.0317 - head_6_accuracy: 0.9882 - head_6_loss: 0.0316 - head_7_accuracy: 0.9883 - head_7_loss: 0.0316 - loss: 0.2216\n","Epoch 97: val_head_1_accuracy did not improve from 0.99466\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m27s\u001b[0m 85ms/step - head_1_accuracy: 0.9884 - head_1_loss: 0.0327 - head_2_accuracy: 0.9883 - head_2_loss: 0.0327 - head_3_accuracy: 0.9881 - head_3_loss: 0.0327 - head_4_accuracy: 0.9883 - head_4_loss: 0.0329 - head_5_accuracy: 0.9883 - head_5_loss: 0.0328 - head_6_accuracy: 0.9883 - head_6_loss: 0.0327 - head_7_accuracy: 0.9882 - head_7_loss: 0.0327 - loss: 0.2294 - val_head_1_accuracy: 0.9945 - val_head_1_loss: 0.0164 - val_head_2_accuracy: 0.9946 - val_head_2_loss: 0.0164 - val_head_3_accuracy: 0.9944 - val_head_3_loss: 0.0164 - val_head_4_accuracy: 0.9946 - val_head_4_loss: 0.0164 - val_head_5_accuracy: 0.9945 - val_head_5_loss: 0.0164 - val_head_6_accuracy: 0.9946 - val_head_6_loss: 0.0164 - val_head_7_accuracy: 0.9945 - val_head_7_loss: 0.0164 - val_loss: 0.1153 - learning_rate: 1.4582e-06\n","\n","Epoch 98: LearningRateScheduler setting learning rate to 7.038963922582963e-07.\n","Epoch 98/100\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 79ms/step - head_1_accuracy: 0.9872 - head_1_loss: 0.0349 - head_2_accuracy: 0.9875 - head_2_loss: 0.0347 - head_3_accuracy: 0.9874 - head_3_loss: 0.0346 - head_4_accuracy: 0.9872 - head_4_loss: 0.0353 - head_5_accuracy: 0.9872 - head_5_loss: 0.0347 - head_6_accuracy: 0.9874 - head_6_loss: 0.0347 - head_7_accuracy: 0.9874 - head_7_loss: 0.0347 - loss: 0.2436\n","Epoch 98: val_head_1_accuracy did not improve from 0.99466\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m27s\u001b[0m 85ms/step - head_1_accuracy: 0.9879 - head_1_loss: 0.0338 - head_2_accuracy: 0.9880 - head_2_loss: 0.0335 - head_3_accuracy: 0.9879 - head_3_loss: 0.0335 - head_4_accuracy: 0.9878 - head_4_loss: 0.0337 - head_5_accuracy: 0.9877 - head_5_loss: 0.0336 - head_6_accuracy: 0.9879 - head_6_loss: 0.0336 - head_7_accuracy: 0.9878 - head_7_loss: 0.0336 - loss: 0.2356 - val_head_1_accuracy: 0.9946 - val_head_1_loss: 0.0164 - val_head_2_accuracy: 0.9945 - val_head_2_loss: 0.0164 - val_head_3_accuracy: 0.9946 - val_head_3_loss: 0.0164 - val_head_4_accuracy: 0.9946 - val_head_4_loss: 0.0164 - val_head_5_accuracy: 0.9946 - val_head_5_loss: 0.0164 - val_head_6_accuracy: 0.9946 - val_head_6_loss: 0.0164 - val_head_7_accuracy: 0.9946 - val_head_7_loss: 0.0164 - val_loss: 0.1153 - learning_rate: 7.0390e-07\n","\n","Epoch 99: LearningRateScheduler setting learning rate to 2.510121121636345e-07.\n","Epoch 99/100\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 78ms/step - head_1_accuracy: 0.9889 - head_1_loss: 0.0331 - head_2_accuracy: 0.9888 - head_2_loss: 0.0331 - head_3_accuracy: 0.9889 - head_3_loss: 0.0331 - head_4_accuracy: 0.9887 - head_4_loss: 0.0332 - head_5_accuracy: 0.9889 - head_5_loss: 0.0330 - head_6_accuracy: 0.9888 - head_6_loss: 0.0331 - head_7_accuracy: 0.9887 - head_7_loss: 0.0331 - loss: 0.2317\n","Epoch 99: val_head_1_accuracy did not improve from 0.99466\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m27s\u001b[0m 85ms/step - head_1_accuracy: 0.9888 - head_1_loss: 0.0331 - head_2_accuracy: 0.9887 - head_2_loss: 0.0331 - head_3_accuracy: 0.9888 - head_3_loss: 0.0330 - head_4_accuracy: 0.9888 - head_4_loss: 0.0331 - head_5_accuracy: 0.9889 - head_5_loss: 0.0330 - head_6_accuracy: 0.9888 - head_6_loss: 0.0331 - head_7_accuracy: 0.9887 - head_7_loss: 0.0330 - loss: 0.2316 - val_head_1_accuracy: 0.9944 - val_head_1_loss: 0.0164 - val_head_2_accuracy: 0.9946 - val_head_2_loss: 0.0164 - val_head_3_accuracy: 0.9945 - val_head_3_loss: 0.0164 - val_head_4_accuracy: 0.9944 - val_head_4_loss: 0.0164 - val_head_5_accuracy: 0.9944 - val_head_5_loss: 0.0164 - val_head_6_accuracy: 0.9945 - val_head_6_loss: 0.0164 - val_head_7_accuracy: 0.9945 - val_head_7_loss: 0.0164 - val_loss: 0.1153 - learning_rate: 2.5101e-07\n","\n","Epoch 100: LearningRateScheduler setting learning rate to 1e-07.\n","Epoch 100/100\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 79ms/step - head_1_accuracy: 0.9894 - head_1_loss: 0.0308 - head_2_accuracy: 0.9892 - head_2_loss: 0.0306 - head_3_accuracy: 0.9893 - head_3_loss: 0.0307 - head_4_accuracy: 0.9893 - head_4_loss: 0.0307 - head_5_accuracy: 0.9893 - head_5_loss: 0.0307 - head_6_accuracy: 0.9893 - head_6_loss: 0.0307 - head_7_accuracy: 0.9892 - head_7_loss: 0.0307 - loss: 0.2149\n","Epoch 100: val_head_1_accuracy did not improve from 0.99466\n","\u001b[1m316/316\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m27s\u001b[0m 85ms/step - head_1_accuracy: 0.9891 - head_1_loss: 0.0326 - head_2_accuracy: 0.9889 - head_2_loss: 0.0325 - head_3_accuracy: 0.9890 - head_3_loss: 0.0326 - head_4_accuracy: 0.9889 - head_4_loss: 0.0326 - head_5_accuracy: 0.9891 - head_5_loss: 0.0326 - head_6_accuracy: 0.9891 - head_6_loss: 0.0326 - head_7_accuracy: 0.9890 - head_7_loss: 0.0325 - loss: 0.2282 - val_head_1_accuracy: 0.9945 - val_head_1_loss: 0.0164 - val_head_2_accuracy: 0.9945 - val_head_2_loss: 0.0164 - val_head_3_accuracy: 0.9946 - val_head_3_loss: 0.0164 - val_head_4_accuracy: 0.9946 - val_head_4_loss: 0.0164 - val_head_5_accuracy: 0.9946 - val_head_5_loss: 0.0164 - val_head_6_accuracy: 0.9946 - val_head_6_loss: 0.0164 - val_head_7_accuracy: 0.9946 - val_head_7_loss: 0.0164 - val_loss: 0.1152 - learning_rate: 1.0000e-07\n"]},{"name":"stderr","output_type":"stream","text":["/usr/local/lib/python3.12/dist-packages/keras/src/layers/layer.py:424: UserWarning: `build()` was called on layer 'vi_t__patch_encoder', however the layer does not have a `build()` method implemented and it looks like it has unbuilt state. This will cause the layer to be marked as built, despite not being actually built, which may cause failures down the line. Make sure to implement a proper `build()` method.\n"," warnings.warn(\n","/usr/local/lib/python3.12/dist-packages/keras/src/layers/layer.py:424: UserWarning: `build()` was called on layer 'vi_t__class__token__norm', however the layer does not have a `build()` method implemented and it looks like it has unbuilt state. This will cause the layer to be marked as built, despite not being actually built, which may cause failures down the line. Make sure to implement a proper `build()` method.\n"," warnings.warn(\n","WARNING:absl:You are saving your model as an HDF5 file via `model.save()` or `keras.saving.save_model(model)`. This file format is considered legacy. We recommend using instead the native Keras format, e.g. `model.save('my_model.keras')` or `keras.saving.save_model(model, 'my_model.keras')`. \n"]},{"name":"stdout","output_type":"stream","text":["Saved .h5 models: ViT_UNet_MultiHead_best.h5, ViT_UNet_MultiHead_final.h5\n"]},{"data":{"image/png":"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\n","text/plain":["
"]},"metadata":{},"output_type":"display_data"},{"name":"stdout","output_type":"stream","text":["Saved training plots directly to /kaggle/working/results\n","Saved training workbook directly to /kaggle/working/results/model_training_multihead_results.xlsx\n","Saved model registry directly to /kaggle/working/results/model_registry_multihead.json\n"]},{"data":{"text/html":["
\n","\n","\n"," \n"," \n"," \n"," \n"," \n"," \n"," \n"," \n"," \n"," \n"," \n"," \n"," \n"," \n"," \n"," \n"," \n"," \n"," \n"," \n"," \n"," \n"," \n"," \n"," \n"," \n"," \n"," \n"," \n"," \n"," \n"," \n"," \n"," \n"," \n"," \n"," \n"," \n"," \n"," \n"," \n"," \n"," \n"," \n"," \n"," \n"," \n"," \n"," \n"," \n"," \n"," \n"," \n"," \n"," \n"," \n"," \n"," \n"," \n"," \n","
model_namebest_model_pathfinal_model_pathbest_h5_pathfinal_h5_pathtest_accuracycohen_kappatraining_time_sectotal_paramstrainable_paramsnon_trainable_params
0AlexNet_CNN_MultiHead/kaggle/working/results/models/AlexNet_CNN_Mul.../kaggle/working/results/models/AlexNet_CNN_Mul.../kaggle/working/results/models/AlexNet_CNN_Mul.../kaggle/working/results/models/AlexNet_CNN_Mul...0.9994800.9993741263.71070334061592340615920
1GFNet_MultiHead/kaggle/working/results/models/GFNet_MultiHead.../kaggle/working/results/models/GFNet_MultiHead.../kaggle/working/results/models/GFNet_MultiHead.../kaggle/working/results/models/GFNet_MultiHead...1.0000001.0000003140.46096525580304255803040
2ViT_UNet_MultiHead/kaggle/working/results/models/ViT_UNet_MultiH.../kaggle/working/results/models/ViT_UNet_MultiH.../kaggle/working/results/models/ViT_UNet_MultiH.../kaggle/working/results/models/ViT_UNet_MultiH...0.9945820.9934732901.37374316316944163169440
\n","
"],"text/plain":[" model_name best_model_path \\\n","0 AlexNet_CNN_MultiHead /kaggle/working/results/models/AlexNet_CNN_Mul... \n","1 GFNet_MultiHead /kaggle/working/results/models/GFNet_MultiHead... \n","2 ViT_UNet_MultiHead /kaggle/working/results/models/ViT_UNet_MultiH... \n","\n"," final_model_path \\\n","0 /kaggle/working/results/models/AlexNet_CNN_Mul... \n","1 /kaggle/working/results/models/GFNet_MultiHead... \n","2 /kaggle/working/results/models/ViT_UNet_MultiH... \n","\n"," best_h5_path \\\n","0 /kaggle/working/results/models/AlexNet_CNN_Mul... \n","1 /kaggle/working/results/models/GFNet_MultiHead... \n","2 /kaggle/working/results/models/ViT_UNet_MultiH... \n","\n"," final_h5_path test_accuracy \\\n","0 /kaggle/working/results/models/AlexNet_CNN_Mul... 0.999480 \n","1 /kaggle/working/results/models/GFNet_MultiHead... 1.000000 \n","2 /kaggle/working/results/models/ViT_UNet_MultiH... 0.994582 \n","\n"," cohen_kappa training_time_sec total_params trainable_params \\\n","0 0.999374 1263.710703 34061592 34061592 \n","1 1.000000 3140.460965 25580304 25580304 \n","2 0.993473 2901.373743 16316944 16316944 \n","\n"," non_trainable_params \n","0 0 \n","1 0 \n","2 0 "]},"execution_count":19,"metadata":{},"output_type":"execute_result"},{"data":{"image/png":"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\n","text/plain":["
"]},"metadata":{},"output_type":"display_data"}],"source":["training_wb = ensure_workbook(TRAINING_WORKBOOK)\n","\n","# Clear existing per-model sheets (keep Summary) to avoid stale data\n","if 'Summary' in training_wb.sheetnames and len(training_wb.sheetnames) > 1:\n"," for sheet in training_wb.sheetnames[1:]:\n"," del training_wb[sheet]\n","\n","summary_rows = []\n","registry = {}\n","\n","for model_name, spec in MODEL_SPECS.items():\n"," print('\\n' + '=' * 30 + f' Training {model_name} ' + '=' * 30)\n"," summary = run_training(model_name, spec, training_wb)\n"," summary_rows.append(summary)\n"," registry[model_name] = {\n"," 'best_model_path': summary['best_model_path'],\n"," 'final_model_path': summary['final_model_path'],\n"," 'best_h5_path': summary.get('best_h5_path', str(summary['best_model_path']).replace('.keras', '.h5')),\n"," 'final_h5_path': summary.get('final_h5_path', str(summary['final_model_path']).replace('.keras', '.h5')),\n"," 'custom_objects': sorted(CUSTOM_OBJECTS.keys()),\n"," 'patch_size': P_S,\n"," 'num_classes': num_classes,\n"," 'bands': B,\n"," }\n","\n","# Write consolidated summary sheet\n","summary_df = pd.DataFrame(summary_rows)\n","summary_ws = training_wb['Summary']\n","summary_ws.delete_rows(1, summary_ws.max_row)\n","for r_idx, row in enumerate(dataframe_to_rows(summary_df, index=False, header=True), start=1):\n"," for c_idx, val in enumerate(row, start=1):\n"," summary_ws.cell(row=r_idx, column=c_idx, value=val)\n","autosize_columns(summary_ws)\n","\n","# Save workbook and registry directly to /kaggle/working/results/\n","training_wb.save(TRAINING_WORKBOOK)\n","MODEL_REGISTRY_PATH.write_text(json.dumps(registry, indent=2))\n","\n","# Mirror into SAVED_MODEL_DIR if different\n","if SAVED_MODEL_DIR != OUTPUT_ROOT:\n"," try:\n"," shutil.copy2(TRAINING_WORKBOOK, SAVED_MODEL_DIR / TRAINING_WORKBOOK.name)\n"," shutil.copy2(MODEL_REGISTRY_PATH, SAVED_MODEL_DIR / MODEL_REGISTRY_PATH.name)\n"," except Exception as e:\n"," print(f\"Notice: Could not mirror registry/workbook to models dir: {e}\")\n","\n","print('Saved training workbook directly to', TRAINING_WORKBOOK)\n","print('Saved model registry directly to', MODEL_REGISTRY_PATH)\n","summary_df\n"]},{"cell_type":"markdown","id":"100-smoke-hdr","metadata":{"papermill":{"duration":4.862604,"end_time":"2026-09-12T03:55:00.014229+00:00","exception":false,"start_time":"2026-09-12T03:54:55.151625+00:00","status":"completed"},"tags":[]},"source":["# 10.0 — Smoke Check\n","\n","Reload each saved best model from the registry and assert that it returns exactly\n","7 output heads on a 4-sample mini-batch."]},{"cell_type":"code","execution_count":20,"id":"100-smoke-code","metadata":{"execution":{"iopub.execute_input":"2026-09-12T03:55:09.628762Z","iopub.status.busy":"2026-09-12T03:55:09.627942Z","iopub.status.idle":"2026-09-12T03:55:48.190421Z","shell.execute_reply":"2026-09-12T03:55:48.189646Z"},"papermill":{"duration":43.414795,"end_time":"2026-09-12T03:55:48.192347+00:00","exception":false,"start_time":"2026-09-12T03:55:04.777552+00:00","status":"completed"},"tags":[]},"outputs":[{"name":"stdout","output_type":"stream","text":["\n","Smoke check: AlexNet_CNN_MultiHead\n","Best .keras heads: 7 | sample shape: (4, 8)\n"]},{"name":"stderr","output_type":"stream","text":["WARNING:absl:Compiled the loaded model, but the compiled metrics have yet to be built. `model.compile_metrics` will be empty until you train or evaluate the model.\n"]},{"name":"stdout","output_type":"stream","text":["Best .h5 heads: 7 | sample shape: (4, 8)\n","\n","Smoke check: GFNet_MultiHead\n"]},{"name":"stderr","output_type":"stream","text":["/usr/local/lib/python3.12/dist-packages/keras/src/layers/layer.py:424: UserWarning: `build()` was called on layer 'gf__patch_encoder', however the layer does not have a `build()` method implemented and it looks like it has unbuilt state. This will cause the layer to be marked as built, despite not being actually built, which may cause failures down the line. Make sure to implement a proper `build()` method.\n"," warnings.warn(\n","/usr/local/lib/python3.12/dist-packages/keras/src/layers/layer.py:424: UserWarning: `build()` was called on layer 'gf__block', however the layer does not have a `build()` method implemented and it looks like it has unbuilt state. This will cause the layer to be marked as built, despite not being actually built, which may cause failures down the line. Make sure to implement a proper `build()` method.\n"," warnings.warn(\n","/usr/local/lib/python3.12/dist-packages/keras/src/layers/layer.py:424: UserWarning: `build()` was called on layer 'gf__block_1', however the layer does not have a `build()` method implemented and it looks like it has unbuilt state. This will cause the layer to be marked as built, despite not being actually built, which may cause failures down the line. Make sure to implement a proper `build()` method.\n"," warnings.warn(\n","/usr/local/lib/python3.12/dist-packages/keras/src/layers/layer.py:424: UserWarning: `build()` was called on layer 'gf__block_2', however the layer does not have a `build()` method implemented and it looks like it has unbuilt state. This will cause the layer to be marked as built, despite not being actually built, which may cause failures down the line. Make sure to implement a proper `build()` method.\n"," warnings.warn(\n","/usr/local/lib/python3.12/dist-packages/keras/src/layers/layer.py:424: UserWarning: `build()` was called on layer 'gf__block_3', however the layer does not have a `build()` method implemented and it looks like it has unbuilt state. This will cause the layer to be marked as built, despite not being actually built, which may cause failures down the line. Make sure to implement a proper `build()` method.\n"," warnings.warn(\n","/usr/local/lib/python3.12/dist-packages/keras/src/layers/layer.py:424: UserWarning: `build()` was called on layer 'gf__block_4', however the layer does not have a `build()` method implemented and it looks like it has unbuilt state. This will cause the layer to be marked as built, despite not being actually built, which may cause failures down the line. Make sure to implement a proper `build()` method.\n"," warnings.warn(\n","/usr/local/lib/python3.12/dist-packages/keras/src/layers/layer.py:424: UserWarning: `build()` was called on layer 'gf__block_5', however the layer does not have a `build()` method implemented and it looks like it has unbuilt state. This will cause the layer to be marked as built, despite not being actually built, which may cause failures down the line. Make sure to implement a proper `build()` method.\n"," warnings.warn(\n","/usr/local/lib/python3.12/dist-packages/keras/src/layers/layer.py:424: UserWarning: `build()` was called on layer 'gf__block_6', however the layer does not have a `build()` method implemented and it looks like it has unbuilt state. This will cause the layer to be marked as built, despite not being actually built, which may cause failures down the line. Make sure to implement a proper `build()` method.\n"," warnings.warn(\n","/usr/local/lib/python3.12/dist-packages/keras/src/layers/layer.py:424: UserWarning: `build()` was called on layer 'gf__block_7', however the layer does not have a `build()` method implemented and it looks like it has unbuilt state. This will cause the layer to be marked as built, despite not being actually built, which may cause failures down the line. Make sure to implement a proper `build()` method.\n"," warnings.warn(\n","/usr/local/lib/python3.12/dist-packages/keras/src/layers/layer.py:424: UserWarning: `build()` was called on layer 'gf__block_8', however the layer does not have a `build()` method implemented and it looks like it has unbuilt state. This will cause the layer to be marked as built, despite not being actually built, which may cause failures down the line. Make sure to implement a proper `build()` method.\n"," warnings.warn(\n","/usr/local/lib/python3.12/dist-packages/keras/src/layers/layer.py:424: UserWarning: `build()` was called on layer 'gf__block_9', however the layer does not have a `build()` method implemented and it looks like it has unbuilt state. This will cause the layer to be marked as built, despite not being actually built, which may cause failures down the line. Make sure to implement a proper `build()` method.\n"," warnings.warn(\n","/usr/local/lib/python3.12/dist-packages/keras/src/layers/layer.py:424: UserWarning: `build()` was called on layer 'gf__block_10', however the layer does not have a `build()` method implemented and it looks like it has unbuilt state. This will cause the layer to be marked as built, despite not being actually built, which may cause failures down the line. Make sure to implement a proper `build()` method.\n"," warnings.warn(\n","/usr/local/lib/python3.12/dist-packages/keras/src/layers/layer.py:424: UserWarning: `build()` was called on layer 'gf__block_11', however the layer does not have a `build()` method implemented and it looks like it has unbuilt state. This will cause the layer to be marked as built, despite not being actually built, which may cause failures down the line. Make sure to implement a proper `build()` method.\n"," warnings.warn(\n"]},{"name":"stdout","output_type":"stream","text":["WARNING:tensorflow:5 out of the last 425 calls to .one_step_on_data_distributed at 0x787f35608220> triggered tf.function retracing. Tracing is expensive and the excessive number of tracings could be due to (1) creating @tf.function repeatedly in a loop, (2) passing tensors with different shapes, (3) passing Python objects instead of tensors. For (1), please define your @tf.function outside of the loop. For (2), @tf.function has reduce_retracing=True option that can avoid unnecessary retracing. For (3), please refer to https://www.tensorflow.org/guide/function#controlling_retracing and https://www.tensorflow.org/api_docs/python/tf/function for more details.\n"]},{"name":"stderr","output_type":"stream","text":["WARNING:tensorflow:5 out of the last 425 calls to .one_step_on_data_distributed at 0x787f35608220> triggered tf.function retracing. Tracing is expensive and the excessive number of tracings could be due to (1) creating @tf.function repeatedly in a loop, (2) passing tensors with different shapes, (3) passing Python objects instead of tensors. For (1), please define your @tf.function outside of the loop. For (2), @tf.function has reduce_retracing=True option that can avoid unnecessary retracing. For (3), please refer to https://www.tensorflow.org/guide/function#controlling_retracing and https://www.tensorflow.org/api_docs/python/tf/function for more details.\n"]},{"name":"stdout","output_type":"stream","text":["Best .keras heads: 7 | sample shape: (4, 8)\n"]},{"name":"stderr","output_type":"stream","text":["WARNING:absl:Compiled the loaded model, but the compiled metrics have yet to be built. `model.compile_metrics` will be empty until you train or evaluate the model.\n"]},{"name":"stdout","output_type":"stream","text":["WARNING:tensorflow:6 out of the last 426 calls to .one_step_on_data_distributed at 0x7880f0c60e00> triggered tf.function retracing. Tracing is expensive and the excessive number of tracings could be due to (1) creating @tf.function repeatedly in a loop, (2) passing tensors with different shapes, (3) passing Python objects instead of tensors. For (1), please define your @tf.function outside of the loop. For (2), @tf.function has reduce_retracing=True option that can avoid unnecessary retracing. For (3), please refer to https://www.tensorflow.org/guide/function#controlling_retracing and https://www.tensorflow.org/api_docs/python/tf/function for more details.\n"]},{"name":"stderr","output_type":"stream","text":["WARNING:tensorflow:6 out of the last 426 calls to .one_step_on_data_distributed at 0x7880f0c60e00> triggered tf.function retracing. Tracing is expensive and the excessive number of tracings could be due to (1) creating @tf.function repeatedly in a loop, (2) passing tensors with different shapes, (3) passing Python objects instead of tensors. For (1), please define your @tf.function outside of the loop. For (2), @tf.function has reduce_retracing=True option that can avoid unnecessary retracing. For (3), please refer to https://www.tensorflow.org/guide/function#controlling_retracing and https://www.tensorflow.org/api_docs/python/tf/function for more details.\n"]},{"name":"stdout","output_type":"stream","text":["Best .h5 heads: 7 | sample shape: (4, 8)\n","\n","Smoke check: ViT_UNet_MultiHead\n"]},{"name":"stderr","output_type":"stream","text":["/usr/local/lib/python3.12/dist-packages/keras/src/layers/layer.py:424: UserWarning: `build()` was called on layer 'vi_t__patch_encoder', however the layer does not have a `build()` method implemented and it looks like it has unbuilt state. This will cause the layer to be marked as built, despite not being actually built, which may cause failures down the line. Make sure to implement a proper `build()` method.\n"," warnings.warn(\n","/usr/local/lib/python3.12/dist-packages/keras/src/layers/layer.py:424: UserWarning: `build()` was called on layer 'vi_t__class__token__norm', however the layer does not have a `build()` method implemented and it looks like it has unbuilt state. This will cause the layer to be marked as built, despite not being actually built, which may cause failures down the line. Make sure to implement a proper `build()` method.\n"," warnings.warn(\n"]},{"name":"stdout","output_type":"stream","text":["Best .keras heads: 7 | sample shape: (4, 8)\n"]},{"name":"stderr","output_type":"stream","text":["WARNING:absl:Compiled the loaded model, but the compiled metrics have yet to be built. `model.compile_metrics` will be empty until you train or evaluate the model.\n"]},{"name":"stdout","output_type":"stream","text":["Best .h5 heads: 7 | sample shape: (4, 8)\n"]}],"source":["for model_name, info in json.loads(MODEL_REGISTRY_PATH.read_text()).items():\n"," print('\\nSmoke check:', model_name)\n"," best_path = Path(info['best_model_path'])\n"," if not best_path.exists():\n"," best_path = OUTPUT_ROOT / best_path.name\n"," model = load_model(str(best_path), custom_objects=CUSTOM_OBJECTS, safe_mode=False)\n"," outputs = model.predict(x_test[:4], verbose=0)\n"," assert isinstance(outputs, list) and len(outputs) == 7, (\n"," f\"Expected 7 heads, got {len(outputs) if isinstance(outputs, list) else 1}\"\n"," )\n"," print('Best .keras heads:', len(outputs), '| sample shape:', outputs[0].shape)\n","\n"," # Verify .h5 model\n"," h5_path = Path(info.get('best_h5_path', str(best_path).replace('.keras', '.h5')))\n"," if not h5_path.exists():\n"," h5_path = OUTPUT_ROOT / h5_path.name\n"," if h5_path.exists():\n"," try:\n"," h5_model = load_model(str(h5_path), custom_objects=CUSTOM_OBJECTS, safe_mode=False)\n"," h5_outs = h5_model.predict(x_test[:4], verbose=0)\n"," print('Best .h5 heads:', len(h5_outs), '| sample shape:', h5_outs[0].shape)\n"," except Exception as e:\n"," print(f'Note: .h5 verification encountered: {e}')\n"]}],"metadata":{"accelerator":"GPU","colab":{"gpuType":"T4","provenance":[]},"kernelspec":{"display_name":"Python 3","language":"python","name":"python3"},"language_info":{"codemirror_mode":{"name":"ipython","version":3},"file_extension":".py","mimetype":"text/x-python","name":"python","nbconvert_exporter":"python","pygments_lexer":"ipython3","version":"3.12.13"},"papermill":{"default_parameters":{},"duration":7477.992965,"end_time":"2026-09-12T03:55:56.796187+00:00","environment_variables":{},"exception":null,"input_path":"__notebook__.ipynb","output_path":"__notebook__.ipynb","parameters":{},"start_time":"2026-09-12T01:51:18.803222+00:00","version":"2.7.0"},"kaggle":{"accelerator":"gpu","dataSources":[{"sourceType":"datasetVersion","sourceId":19526106}],"isInternetEnabled":true,"language":"python","sourceType":"notebook","isGpuEnabled":false}},"nbformat":4,"nbformat_minor":5}