"]},{"cell_type":"code","execution_count":1,"metadata":{"colab":{"base_uri":"https://localhost:8080/"},"executionInfo":{"elapsed":9329,"status":"ok","timestamp":1764238677292,"user":{"displayName":"Ping Ping","userId":"16193029904393259603"},"user_tz":-420},"id":"yiE0unDzFPTR","outputId":"4271cb96-e1cc-4c1b-9c85-045ef94b0760"},"outputs":[{"output_type":"stream","name":"stdout","text":["Compute dtype: float32\n","Variable dtype: float32\n","GPU Available: [PhysicalDevice(name='/physical_device:GPU:0', device_type='GPU')]\n","GPU detected! Training will be fast.\n"]}],"source":["# Google Colab specific imports\n","from google.colab import drive\n","# Core libraries\n","import numpy as np\n","import pandas as pd\n","import os\n","import time\n","import shutil\n","\n","# Suppress TensorFlow warnings\n","os.environ['TF_CPP_MIN_LOG_LEVEL'] = '2'\n","\n","# Visualization\n","import matplotlib.pyplot as plt\n","import cv2\n","from PIL import Image\n","import seaborn as sns\n","sns.set_style('darkgrid')\n","\n","# Machine Learning\n","from sklearn.metrics import confusion_matrix, classification_report\n","from sklearn.model_selection import train_test_split\n","\n","# TensorFlow and Keras\n","import tensorflow as tf\n","from tensorflow import keras\n","\n","# Enable mixed precision for faster training on GPU\n","from tensorflow.keras import mixed_precision\n","policy = mixed_precision.Policy('float32')\n","mixed_precision.set_global_policy(policy)\n","print('Compute dtype: %s' % policy.compute_dtype)\n","print('Variable dtype: %s' % policy.variable_dtype)\n","# Keras modules\n","from tensorflow.keras.preprocessing.image import ImageDataGenerator\n","from tensorflow.keras.layers import Dense, Activation, Dropout, Conv2D, MaxPooling2D, BatchNormalization\n","from tensorflow.keras.optimizers import Adam, Adamax\n","from tensorflow.keras.metrics import categorical_crossentropy\n","from tensorflow.keras import regularizers\n","from tensorflow.keras.models import Model\n","\n","# Check GPU availability\n","print(\"GPU Available:\", tf.config.list_physical_devices('GPU'))\n","if tf.config.list_physical_devices('GPU'):\n"," print(\"GPU detected! Training will be fast.\")\n","else:\n"," print(\"No GPU detected. Training will be slow.\")"]},{"cell_type":"code","execution_count":2,"metadata":{"colab":{"base_uri":"https://localhost:8080/"},"executionInfo":{"elapsed":4388,"status":"ok","timestamp":1764238681678,"user":{"displayName":"Ping Ping","userId":"16193029904393259603"},"user_tz":-420},"id":"G_mHIM1kR-jX","outputId":"ac511106-a60f-48f3-aadc-2ffaee48eb6c"},"outputs":[{"output_type":"stream","name":"stderr","text":["/usr/local/lib/python3.12/dist-packages/albumentations/augmentations/transforms.py:1952: UserWarning: blur_limit and sigma_limit minimum value can not be both equal to 0. blur_limit minimum value changed to 3.\n"," warnings.warn(\n"]}],"source":["# Install Albumentations (run once in the notebook environment)\n","# If running locally, you can comment out the pip install line and install via your environment\n","%pip install albumentations==1.1.0 -q\n","\n","import albumentations as A\n","import numpy as np\n","\n","# Define richer training augmentations to better simulate real-world clutter\n","train_augment = A.Compose([\n"," A.RandomRotate90(p=0.5),\n"," A.HorizontalFlip(p=0.5),\n"," A.ShiftScaleRotate(shift_limit=0.08, scale_limit=0.08, rotate_limit=20, p=0.7),\n"," A.RandomBrightnessContrast(brightness_limit=0.2, contrast_limit=0.2, p=0.7),\n"," A.GaussianBlur(blur_limit=3, p=0.2),\n"," A.GaussNoise(var_limit=(10.0, 50.0), p=0.2),\n"," A.GridDropout(ratio=0.2, p=0.3),\n"," A.CoarseDropout(max_holes=1, max_height=50, max_width=50, p=0.3),\n","], p=1.0)\n","\n","# Preprocessing function for ImageDataGenerator: apply Albumentations then EfficientNet preprocess\n","def albumentations_preprocess(image):\n"," # ImageDataGenerator passes images as float32 arrays in [0,255]\n"," img_uint8 = image.astype('uint8')\n"," augmented = train_augment(image=img_uint8)['image']\n"," # Convert to float32 and apply EfficientNet preprocessing\n"," return tf.keras.applications.efficientnet.preprocess_input(augmented.astype('float32'))\n","\n"]},{"cell_type":"markdown","metadata":{"id":"bUgE4lf7FPTS"},"source":["### The size of the images are very large. If you try to use the full image size training will take a very long time. Therefore you\n","### want to reduce the image size. However if you include resizing of images as part of training it requires each image to be repeatedly\n","### resized in each epoch. This also make the training times to be large. It is more efficient to resize all the images FIRST and save them.\n","### Then use these images for training."]},{"cell_type":"markdown","metadata":{"id":"gtjuYUnyFPTS"},"source":["\n","#
Resize all images and save to the working directory
"]},{"cell_type":"code","execution_count":3,"metadata":{"colab":{"base_uri":"https://localhost:8080/"},"id":"iy31fzChFPTT","executionInfo":{"status":"ok","timestamp":1764239060177,"user_tz":-420,"elapsed":378495,"user":{"displayName":"Ping Ping","userId":"16193029904393259603"}},"outputId":"ced3cfbd-dec0-47df-8aa2-1ad7a649aefc"},"outputs":[{"output_type":"stream","name":"stdout","text":["Drive already mounted at /content/drive; to attempt to forcibly remount, call drive.mount(\"/content/drive\", force_remount=True).\n","processing Alternaria Leaf Spot class\n","processing Bacterial Soft Rot class\n","processing Black Rot class\n","processing Downy Mildew class\n","processing Healthy class\n","resizing elapsed time was 0.0 hours, 6.0 minutes, 15.52 seconds)\n"]}],"source":["# Mount Google Drive\n","drive.mount('/content/drive')\n","start_time=time.time()\n","sdir = '/content/drive/MyDrive/dataset_train'\n","working_dir = '/content/cauliflower_training/'\n","img_height = 300\n","img_width = 300\n","\n","# Create working directory first\n","os.makedirs(working_dir, exist_ok=True)\n","\n","save_path = os.path.join(working_dir, 'cali')\n","\n","if os.path.isdir(save_path):\n"," shutil.rmtree(save_path) # if the directory exists remove it\n","\n","os.makedirs(save_path, exist_ok=True) # Changed from os.mkdir() to os.makedirs()\n","\n","classlist = sorted(os.listdir(sdir))\n","for klass in classlist:\n"," classpath = os.path.join(sdir, klass)\n"," if not os.path.isdir(classpath): # skip if not a directory\n"," continue\n"," print('processing ', klass, ' class')\n"," class_dest_path = os.path.join(save_path, klass)\n"," os.makedirs(class_dest_path, exist_ok=True) # Changed here too\n"," flist = os.listdir(classpath)\n"," for f in flist:\n"," fpath = os.path.join(classpath,f)\n"," f_dest_path = os.path.join(class_dest_path, f)\n"," try: # make sure images are correct image files\n"," # Use PIL to ensure consistent RGB handling across readers\n"," with Image.open(fpath) as img:\n"," img = img.convert('RGB')\n"," # Use modern resampling API when available\n"," try:\n"," resample_method = Image.Resampling.LANCZOS\n"," except AttributeError:\n"," resample_method = Image.LANCZOS if hasattr(Image, 'LANCZOS') else Image.NEAREST\n"," img = img.resize((img_height, img_width), resample_method)\n"," img.save(f_dest_path)\n"," except Exception as e:\n"," print('Image ', fpath, ' is an invalid image file:', e)\n","\n","tr_duration = time.time() - start_time\n","hours = tr_duration // 3600\n","minutes = (tr_duration - (hours * 3600)) // 60\n","seconds = tr_duration - ((hours * 3600) + (minutes * 60))\n","msg = f'resizing elapsed time was {str(hours)} hours, {minutes:4.1f} minutes, {seconds:4.2f} seconds)'\n","print(msg, flush=True)"]},{"cell_type":"code","execution_count":4,"metadata":{"colab":{"base_uri":"https://localhost:8080/"},"id":"gHHGvBF8Z1WV","executionInfo":{"status":"ok","timestamp":1764239109276,"user_tz":-420,"elapsed":49101,"user":{"displayName":"Ping Ping","userId":"16193029904393259603"}},"outputId":"d6b2a60e-0045-4b2d-d381-874c649f17db"},"outputs":[{"output_type":"stream","name":"stdout","text":["Moved 0 files to quarantine: /content/cauliflower_training/quarantine\n"]}],"source":["from PIL import Image\n","import shutil\n","import os\n","\n","sdir = '/content/drive/MyDrive/dataset_train'\n","quarantine = os.path.join(working_dir, 'quarantine')\n","os.makedirs(quarantine, exist_ok=True)\n","\n","moved = 0\n","for root, _, files in os.walk(sdir):\n"," for f in files:\n"," path = os.path.join(root, f)\n"," # skip if already in quarantine\n"," try:\n"," if os.path.commonpath([quarantine, path]) == quarantine:\n"," continue\n"," except ValueError:\n"," # fallback when commonpath can't compare (different path types/drives)\n"," abs_quarantine = os.path.abspath(quarantine)\n"," abs_path = os.path.abspath(path)\n"," if abs_path == abs_quarantine or abs_path.startswith(abs_quarantine + os.sep):\n"," continue\n"," try:\n"," with Image.open(path) as im:\n"," im.verify() # will raise if file is truncated or not an image\n"," except Exception:\n"," rel = os.path.relpath(path, sdir)\n"," dest_path = os.path.join(quarantine, rel)\n"," os.makedirs(os.path.dirname(dest_path), exist_ok=True)\n"," try:\n"," shutil.move(path, dest_path)\n"," moved += 1\n"," except Exception as e:\n"," print('Failed to move', path, '->', dest_path, ':', e)\n","\n","print(f'Moved {moved} files to quarantine: {quarantine}')"]},{"cell_type":"markdown","metadata":{"id":"G4PQ20znFPTT"},"source":["\n","#
Read in resized images and create a dataframe of image paths and class labels
"]},{"cell_type":"code","execution_count":5,"metadata":{"colab":{"base_uri":"https://localhost:8080/"},"id":"LlmyAWPxFPTT","executionInfo":{"status":"ok","timestamp":1764239109438,"user_tz":-420,"elapsed":164,"user":{"displayName":"Ping Ping","userId":"16193029904393259603"}},"outputId":"0efb3563-609a-49fa-a0de-19666247af07"},"outputs":[{"output_type":"stream","name":"stdout","text":["train_df lenght: 6203 test_df length: 776 valid_df length: 775\n","The number of classes in the dataset is: 5\n"," CLASS IMAGE COUNT \n"," Alternaria Leaf Spot 1206 \n"," Bacterial Soft Rot 1121 \n"," Black Rot 627 \n"," Downy Mildew 721 \n"," Healthy 2528 \n","Healthy has the most images= 2528 Black Rot has the least images= 627\n","average height= 300 average width= 300 aspect ratio= 1.0\n"]}],"source":["def make_df(save_path):\n"," classlist=os.listdir(save_path)\n"," filepaths=[]\n"," labels=[]\n"," for klass in classlist:\n"," classpath=os.path.join(save_path, klass)\n"," flist=os.listdir(classpath)\n"," for f in flist:\n"," fpath=os.path.join(classpath,f)\n"," filepaths.append(fpath)\n"," labels.append(klass)\n"," Fseries=pd.Series(filepaths, name='filepaths')\n"," Lseries=pd.Series(labels, name='labels')\n"," df=pd.concat([Fseries, Lseries], axis=1)\n"," train_df, dummy_df=train_test_split(df, train_size=.8, shuffle=True, random_state=123, stratify=df['labels'])\n"," valid_df, test_df=train_test_split(dummy_df,train_size=.5, shuffle=True, random_state=123, stratify=dummy_df['labels'])\n"," print('train_df lenght: ', len(train_df), ' test_df length: ', len(test_df), ' valid_df length: ', len(valid_df))\n"," # get the number of classes and the images count for each class in train_df\n"," classes=sorted(list(train_df['labels'].unique()))\n"," class_count = len(classes)\n"," print('The number of classes in the dataset is: ', class_count)\n"," groups=train_df.groupby('labels')\n"," print('{0:^30s} {1:^13s}'.format('CLASS', 'IMAGE COUNT'))\n"," countlist=[]\n"," classlist=[]\n"," for label in sorted(list(train_df['labels'].unique())):\n"," group=groups.get_group(label)\n"," countlist.append(len(group))\n"," classlist.append(label)\n"," print('{0:^30s} {1:^13s}'.format(label, str(len(group))))\n","\n"," # get the classes with the minimum and maximum number of train images\n"," max_value=np.max(countlist)\n"," max_index=countlist.index(max_value)\n"," max_class=classlist[max_index]\n"," min_value=np.min(countlist)\n"," min_index=countlist.index(min_value)\n"," min_class=classlist[min_index]\n"," print(max_class, ' has the most images= ',max_value, ' ', min_class, ' has the least images= ', min_value)\n"," # lets get the average height and width of a sample of the train images\n"," ht=0\n"," wt=0\n"," # select 100 random samples of train_df\n"," train_df_sample=train_df.sample(n=100, random_state=123,axis=0)\n"," for i in range (len(train_df_sample)):\n"," fpath=train_df_sample['filepaths'].iloc[i]\n"," img=plt.imread(fpath)\n"," shape=img.shape\n"," ht += shape[0]\n"," wt += shape[1]\n"," print('average height= ', ht//100, ' average width= ', wt//100, 'aspect ratio= ', ht/wt)\n"," return train_df, test_df, valid_df, classes\n","train_df, test_df, valid_df, classes = make_df(save_path)\n","class_count=len(classes)"]},{"cell_type":"markdown","metadata":{"id":"tMvLY60VR-jZ"},"source":["\n","#
Balancing Dataset
"]},{"cell_type":"code","execution_count":6,"metadata":{"colab":{"base_uri":"https://localhost:8080/"},"id":"evEW_PwrkXFq","executionInfo":{"status":"ok","timestamp":1764239109478,"user_tz":-420,"elapsed":17,"user":{"displayName":"Ping Ping","userId":"16193029904393259603"}},"outputId":"59251fee-6eac-4562-e55b-2e884011ae17"},"outputs":[{"output_type":"stream","name":"stdout","text":["================================================================================\n","BALANCING TRAINING DATASET\n","================================================================================\n","Initial dataframe length: 6203\n","\n","=== Initial Class Distribution ===\n","labels\n","Alternaria Leaf Spot 1206\n","Bacterial Soft Rot 1121\n","Black Rot 627\n","Downy Mildew 721\n","Healthy 2528\n","Name: count, dtype: int64\n","\n","=== Step 1: Downsampling classes with > 600 samples ===\n","✓ Downsampled Alternaria Leaf Spot from 1206 to 600\n","✓ Downsampled Bacterial Soft Rot from 1121 to 600\n","✓ Downsampled Black Rot from 627 to 600\n","✓ Downsampled Downy Mildew from 721 to 600\n","✓ Downsampled Healthy from 2528 to 600\n","\n","After downsampling: 3000 total samples\n","\n","=== Step 2: Augmenting classes with < 600 samples ===\n","\n","Total augmented images created: 0\n","\n","=== Final Balanced Distribution ===\n","labels\n","Alternaria Leaf Spot 600\n","Bacterial Soft Rot 600\n","Black Rot 600\n","Downy Mildew 600\n","Healthy 600\n","Name: count, dtype: int64\n","\n","Final dataframe length: 3000\n","✓ SUCCESS: All classes perfectly balanced at 600 samples!\n","\n","✓ Saved balanced dataset to: ./train_balanced.csv\n"]}],"source":["# IMPROVED BALANCE FUNCTION - Now handles both upsampling AND downsampling\n","def balance(df, n, working_dir, img_size):\n"," \"\"\"\n"," Balance dataset to exactly n samples per class by:\n"," 1. Downsampling majority classes to n\n"," 2. Augmenting minority classes up to n\n"," \"\"\"\n"," df = df.copy()\n"," print(f'Initial dataframe length: {len(df)}')\n"," print('\\n=== Initial Class Distribution ===')\n"," initial_dist = df['labels'].value_counts().sort_index()\n"," print(initial_dist)\n","\n"," # STEP 1: DOWNSAMPLE majority classes first\n"," print(f'\\n=== Step 1: Downsampling classes with > {n} samples ===')\n"," balanced_frames = []\n"," for label in sorted(df['labels'].unique()):\n"," class_df = df[df['labels'] == label]\n"," count = len(class_df)\n","\n"," if count > n:\n"," class_df = class_df.sample(n=n, random_state=123)\n"," print(f'✓ Downsampled {label:30s} from {count:4d} to {n}')\n"," else:\n"," print(f' Kept {label:30s} at {count:4d} samples (will augment later)')\n","\n"," balanced_frames.append(class_df)\n","\n"," df = pd.concat(balanced_frames, axis=0).reset_index(drop=True)\n"," print(f'\\nAfter downsampling: {len(df)} total samples')\n","\n"," # STEP 2: AUGMENT minority classes\n"," print(f'\\n=== Step 2: Augmenting classes with < {n} samples ===')\n"," aug_dir = os.path.join(working_dir, 'aug')\n"," if os.path.isdir(aug_dir):\n"," shutil.rmtree(aug_dir)\n"," os.mkdir(aug_dir)\n","\n"," for label in df['labels'].unique():\n"," dir_path = os.path.join(aug_dir, label)\n"," os.mkdir(dir_path)\n","\n"," total_augmented = 0\n"," gen = ImageDataGenerator(\n"," horizontal_flip=True,\n"," rotation_range=20,\n"," width_shift_range=.2,\n"," height_shift_range=.2,\n"," zoom_range=.2\n"," )\n","\n"," groups = df.groupby('labels')\n"," for label in sorted(df['labels'].unique()):\n"," group = groups.get_group(label)\n"," sample_count = len(group)\n","\n"," if sample_count < n:\n"," delta = n - sample_count\n"," target_dir = os.path.join(aug_dir, label)\n"," print(f'✓ Creating {delta:3d} augmented images for {label:30s}')\n","\n"," aug_gen = gen.flow_from_dataframe(\n"," group,\n"," x_col='filepaths',\n"," y_col=None,\n"," target_size=img_size,\n"," class_mode=None,\n"," batch_size=1,\n"," shuffle=False,\n"," save_to_dir=target_dir,\n"," save_prefix='aug-',\n"," color_mode='rgb',\n"," save_format='jpg'\n"," )\n","\n"," aug_img_count = 0\n"," while aug_img_count < delta:\n"," images = next(aug_gen)\n"," aug_img_count += len(images)\n","\n"," total_augmented += aug_img_count\n","\n"," print(f'\\nTotal augmented images created: {total_augmented}')\n","\n"," # STEP 3: Merge augmented images\n"," aug_fpaths = []\n"," aug_labels = []\n"," classlist = os.listdir(aug_dir)\n"," for klass in classlist:\n"," classpath = os.path.join(aug_dir, klass)\n"," flist = os.listdir(classpath)\n"," for f in flist:\n"," fpath = os.path.join(classpath, f)\n"," aug_fpaths.append(fpath)\n"," aug_labels.append(klass)\n","\n"," if aug_fpaths:\n"," Fseries = pd.Series(aug_fpaths, name='filepaths')\n"," Lseries = pd.Series(aug_labels, name='labels')\n"," aug_df = pd.concat([Fseries, Lseries], axis=1)\n"," df = pd.concat([df, aug_df], axis=0).reset_index(drop=True)\n","\n"," print(f'\\n=== Final Balanced Distribution ===')\n"," final_dist = df['labels'].value_counts().sort_index()\n"," print(final_dist)\n"," print(f'\\nFinal dataframe length: {len(df)}')\n","\n"," # Verify perfect balance\n"," if final_dist.nunique() == 1 and final_dist.iloc[0] == n:\n"," print(f'✓ SUCCESS: All classes perfectly balanced at {n} samples!')\n"," else:\n"," print('⚠ WARNING: Classes not perfectly balanced!')\n","\n"," # Save to CSV\n"," try:\n"," merged_csv = os.path.join(working_dir, 'train_balanced.csv')\n"," df.to_csv(merged_csv, index=False)\n"," print(f'\\n✓ Saved balanced dataset to: {merged_csv}')\n"," except Exception as e:\n"," print(f'Warning: could not save CSV: {e}')\n","\n"," return df\n","\n","# Apply balancing\n","n = 600 # Increased from 200 for better model training\n","working_dir = r'./'\n","img_size = (300, 300)\n","\n","print('=' * 80)\n","print('BALANCING TRAINING DATASET')\n","print('=' * 80)\n","train_df = balance(train_df, n, working_dir, img_size)"]},{"cell_type":"markdown","metadata":{"id":"jIbH3iu6R-ja"},"source":["\n","\n","#
Rebuild cleaned CSVs
"]},{"cell_type":"code","execution_count":7,"metadata":{"colab":{"base_uri":"https://localhost:8080/"},"id":"0oqcCf7skhM0","executionInfo":{"status":"ok","timestamp":1764239109501,"user_tz":-420,"elapsed":22,"user":{"displayName":"Ping Ping","userId":"16193029904393259603"}},"outputId":"2d8c7ab6-5c33-4684-855c-080aa56ab3c5"},"outputs":[{"output_type":"stream","name":"stdout","text":["Skipping CSV cleaning - ./train_augmented.csv not found yet (run balance function first)\n"]}],"source":["def write_clean_csv(input_csv, out_csv, quarantine_dir):\n"," df = pd.read_csv(input_csv)\n"," # drop any rows whose files were moved to quarantine\n"," if os.path.isdir(quarantine_dir):\n"," quarantined = set(os.listdir(quarantine_dir))\n"," df = df[~df['filepaths'].apply(lambda p: os.path.basename(p) in quarantined)].reset_index(drop=True)\n"," df.to_csv(out_csv, index=False)\n"," print('Wrote cleaned CSV:', out_csv)\n","\n","# Define quarantine_dir (needed if you ran the repair cell earlier)\n","quarantine_dir = os.path.join(working_dir, 'quarantine')\n","\n","# Only run if the input CSV exists\n","input_csv = os.path.join(working_dir, 'train_augmented.csv')\n","output_csv = os.path.join(working_dir, 'train_augmented_clean.csv')\n","\n","if os.path.exists(input_csv):\n"," write_clean_csv(input_csv, output_csv, quarantine_dir)\n","else:\n"," print(f'Skipping CSV cleaning - {input_csv} not found yet (run balance function first)')"]},{"cell_type":"markdown","metadata":{"id":"BfxKCeD1FPTT"},"source":["\n","#
Expand train_df by creating augmented images
\n"]},{"cell_type":"code","execution_count":8,"metadata":{"colab":{"base_uri":"https://localhost:8080/"},"id":"wc9r0S6DFPTU","executionInfo":{"status":"ok","timestamp":1764239109611,"user_tz":-420,"elapsed":108,"user":{"displayName":"Ping Ping","userId":"16193029904393259603"}},"outputId":"55a5b18f-5102-4846-cbd3-f87102842c20"},"outputs":[{"output_type":"stream","name":"stdout","text":["Loading pre-balanced dataset: ./train_balanced.csv\n","\n","✓ Loaded balanced dataset: 3000 samples\n","\n","Class distribution:\n","labels\n","Alternaria Leaf Spot 600\n","Bacterial Soft Rot 600\n","Black Rot 600\n","Downy Mildew 600\n","Healthy 600\n","Name: count, dtype: int64\n"]}],"source":["# Load the balanced dataset (skip if already balanced in memory)\n","balanced_csv = os.path.join(working_dir, 'train_balanced.csv')\n","\n","if os.path.exists(balanced_csv):\n"," print(f\"Loading pre-balanced dataset: {balanced_csv}\")\n"," train_df = pd.read_csv(balanced_csv)\n"," if 'filepaths' not in train_df.columns:\n"," cols = list(train_df.columns)\n"," if len(cols) >= 2:\n"," train_df = train_df.rename(columns={cols[0]: 'filepaths', cols[1]: 'labels'})\n"," train_df = train_df[['filepaths', 'labels']]\n","\n"," print(f'\\n✓ Loaded balanced dataset: {len(train_df)} samples')\n"," print('\\nClass distribution:')\n"," print(train_df['labels'].value_counts().sort_index())\n","else:\n"," print(f\"⚠ {balanced_csv} not found — ensure you ran the balance() function first\")"]},{"cell_type":"markdown","metadata":{"id":"4YWzNQRRFPTU"},"source":["\n","#
Create the train_gen, test_gen final_test_gen and valid_gen
Check generator class distribution/center>"]},{"cell_type":"code","execution_count":11,"metadata":{"colab":{"base_uri":"https://localhost:8080/"},"id":"CkcqGDPs2ecD","executionInfo":{"status":"ok","timestamp":1764239110948,"user_tz":-420,"elapsed":20,"user":{"displayName":"Ping Ping","userId":"16193029904393259603"}},"outputId":"19e3a28e-5ffe-4f68-b8b5-27e1cc95f0b3"},"outputs":[{"output_type":"stream","name":"stdout","text":["\n","=== Dataset Class Distribution ===\n","Train distribution:\n"," Alternaria Leaf Spot : 600 samples\n"," Bacterial Soft Rot : 600 samples\n"," Black Rot : 600 samples\n"," Downy Mildew : 600 samples\n"," Healthy : 600 samples\n","\n","=== Computed Class Weights ===\n"," Alternaria Leaf Spot -> Weight: 1.0000\n"," Bacterial Soft Rot -> Weight: 1.0000\n"," Black Rot -> Weight: 1.0000\n"," Downy Mildew -> Weight: 1.0000\n"," Healthy -> Weight: 1.0000\n","\n","✓ Class weights will provide additional balancing during training\n"]}],"source":["print('\\n=== Dataset Class Distribution ===')\n","print('Train distribution:')\n","train_class_dist = train_df['labels'].value_counts().sort_index()\n","for class_name, count in train_class_dist.items():\n"," print(f' {class_name:30s}: {count:4d} samples')\n","\n","# Compute class weights (even with balanced data, slight weights help)\n","from sklearn.utils.class_weight import compute_class_weight\n","\n","class_weights = compute_class_weight(\n"," class_weight='balanced',\n"," classes=np.arange(class_count),\n"," y=train_df['labels'].map(class_to_idx).values\n",")\n","\n","class_weight_dict = dict(enumerate(class_weights))\n","print(\"\\n=== Computed Class Weights ===\")\n","for idx, class_name in enumerate(class_names):\n"," print(f\" {class_name:30s} -> Weight: {class_weight_dict[idx]:.4f}\")\n","\n","print(\"\\n✓ Class weights will provide additional balancing during training\")"]},{"cell_type":"markdown","metadata":{"id":"QiJjNU33FPTU"},"source":["\n","#
Create a function to show example training images
"]},{"cell_type":"code","execution_count":12,"metadata":{"colab":{"output_embedded_package_id":"1r_4qOvQaYcFDQSYSlHj33tSHDbGxfroQ","base_uri":"https://localhost:8080/","height":943},"id":"AZJtxt_6FPTU","outputId":"82ddb25b-21d7-4c30-99ba-7e946e2d5b0e","executionInfo":{"status":"ok","timestamp":1764239118719,"user_tz":-420,"elapsed":7770,"user":{"displayName":"Ping Ping","userId":"16193029904393259603"}}},"outputs":[{"output_type":"display_data","data":{"text/plain":"Output hidden; open in https://colab.research.google.com to view."},"metadata":{}}],"source":["def show_image_samples_from_df(df, n=25):\n"," samp = df.sample(n=min(n, len(df)), random_state=123)\n"," plt.figure(figsize=(20, 20))\n"," for i, (_, row) in enumerate(samp.iterrows()):\n"," plt.subplot(5, 5, i + 1)\n"," img = Image.open(row['filepaths']).convert('RGB')\n"," img = img.resize((img_size[1], img_size[0]))\n"," plt.imshow(np.array(img) / 255.0)\n"," plt.title(row['labels'], color='blue', fontsize=14)\n"," plt.axis('off')\n"," plt.show()\n","\n","show_image_samples_from_df(train_df)"]},{"cell_type":"markdown","metadata":{"id":"Nt2bzGutFPTe"},"source":["\n","#
Create a model using transfer learning with EfficientNetB3
\n","### NOTE experts advise you make the base model initially not trainable. Then train for some number of epochs\n","### then fine tune model by making base model trainable and run more epochs\n","### I have found this to be WRONG!!!!\n","### Making the base model trainable from the outset leads to faster convegence and a lower validation loss\n","### for the same number of total epochs!"]},{"cell_type":"code","execution_count":13,"metadata":{"colab":{"base_uri":"https://localhost:8080/","height":1000},"id":"w0EIJYieFPTe","executionInfo":{"status":"ok","timestamp":1764239119506,"user_tz":-420,"elapsed":738,"user":{"displayName":"Ping Ping","userId":"16193029904393259603"}},"outputId":"00af17b9-687c-4678-9a58-5fc6876f7360"},"outputs":[{"output_type":"stream","name":"stdout","text":["Building EfficientNetB3 model...\n","\n","✓ Model compiled with learning rate: 0.0005\n"," Total parameters: 11,709,236\n"]},{"output_type":"display_data","data":{"text/plain":["\u001b[1mModel: \"functional\"\u001b[0m\n"],"text/html":["
\n"]},"metadata":{}}],"source":["img_shape = (img_size[0], img_size[1], 3)\n","model_name = 'EfficientNetB3'\n","\n","print(f'Building {model_name} model...')\n","\n","# Create base model\n","base_model = tf.keras.applications.efficientnet.EfficientNetB3(\n"," include_top=False,\n"," weights=\"imagenet\",\n"," input_shape=img_shape,\n"," pooling='max'\n",")\n","\n","base_model.trainable = True\n","\n","# Build model with MODERATE regularization\n","x = base_model.output\n","x = BatchNormalization(axis=-1, momentum=0.99, epsilon=0.001)(x)\n","\n","# Add intermediate layer for better feature learning\n","x = Dense(512, activation='relu',\n"," kernel_regularizer=regularizers.l2(0.001))(x)\n","x = Dropout(rate=.45, seed=123)(x)\n","\n","x = Dense(256, activation='relu',\n"," kernel_regularizer=regularizers.l2(0.001))(x)\n","x = Dropout(rate=.35, seed=123)(x)\n","\n","# Output layer MUST be float32 for mixed precision\n","output = Dense(class_count, activation='softmax', dtype='float32', name='predictions')(x)\n","\n","model = Model(inputs=base_model.input, outputs=output)\n","\n","# Compile model with lower initial learning rate\n","lr = 0.0005 # Reduced from 0.001 for more stable training\n","model.compile(\n"," Adamax(learning_rate=lr),\n"," loss='categorical_crossentropy',\n"," metrics=['accuracy']\n",")\n","\n","print(f'\\n✓ Model compiled with learning rate: {lr}')\n","print(f' Total parameters: {model.count_params():,}')\n","model.summary()"]},{"cell_type":"markdown","metadata":{"id":"hx32vSChFPTe"},"source":["\n","#
Create a custom Keras callback to continue and optionally set LR or halt training
\n","The LR_ASK callback is a convenient callback that allows you to continue training for ask_epoch more epochs or to halt training. \n","If you elect to continue training for more epochs you are given the option to retain the current learning rate (LR) or to \n","enter a new value for the learning rate. The form of use is: \n","ask=LR_ASK(model,epochs, ask_epoch) where: \n","* model is a string which is the name of your compiled model\n","* epochs is an integer which is the number of epochs to run specified in model.fit\n","* ask_epoch is an integer. If ask_epoch is set to a value say 5 then the model will train for 5 epochs. \n"," then the user is ask to enter H to halt training, or enter an inter value. For example if you enter 4 \n"," training will continue for 4 more epochs to epoch 9 then you will be queried again. Once you enter an \n"," integer value you are prompted to press ENTER to continue training using the current learning rate \n"," or to enter a new value for the learning rate. \n"," \n"," At the end of training the model weights are set to the weights for the epoch that achieved the lowest validation loss"]},{"cell_type":"code","execution_count":14,"metadata":{"id":"4C0xLA05FPTe","executionInfo":{"status":"ok","timestamp":1764239119515,"user_tz":-420,"elapsed":4,"user":{"displayName":"Ping Ping","userId":"16193029904393259603"}}},"outputs":[],"source":["class LR_ASK(keras.callbacks.Callback):\n"," def __init__ (self, model, epochs, ask_epoch): # initialization of the callback\n"," super(LR_ASK, self).__init__()\n"," self.user_model=model # Changed from self.model to self.user_model\n"," self.ask_epoch=ask_epoch\n"," self.epochs=epochs\n"," self.ask=True # if True query the user on a specified epoch\n"," self.lowest_vloss=np.inf\n"," self.best_weights=self.user_model.get_weights() # Changed to self.user_model\n"," self.best_epoch=1\n","\n","\n"," def on_train_begin(self, logs=None): # this runs on the beginning of training\n"," if self.ask_epoch == 0:\n"," print('you set ask_epoch = 0, ask_epoch will be set to 1', flush=True)\n"," self.ask_epoch=1\n"," if self.ask_epoch >= self.epochs: # you are running for epochs but ask_epoch>epochs\n"," print('ask_epoch >= epochs, will train for ', epochs, ' epochs', flush=True)\n"," self.ask=False # do not query the user\n"," if self.epochs == 1:\n"," self.ask=False # running only for 1 epoch so do not query user\n"," else:\n"," print('Training will proceed until epoch', ask_epoch,' then you will be asked to')\n"," print(' enter H to halt training or enter an integer for how many more epochs to run then be asked again')\n"," self.start_time= time.time() # set the time at which training started\n","\n"," def on_train_end(self, logs=None): # runs at the end of training\n"," print('loading model with weights from epoch ', self.best_epoch)\n"," self.user_model.set_weights(self.best_weights) # Changed to self.user_model\n"," tr_duration=time.time() - self.start_time # determine how long the training cycle lasted\n"," hours = tr_duration // 3600\n"," minutes = (tr_duration - (hours * 3600)) // 60\n"," seconds = tr_duration - ((hours * 3600) + (minutes * 60))\n"," msg = f'training elapsed time was {str(hours)} hours, {minutes:4.1f} minutes, {seconds:4.2f} seconds)'\n"," print (msg, flush=True) # print out training duration time\n","\n"," def on_epoch_end(self, epoch, logs=None): # method runs on the end of each epoch\n"," v_loss=logs.get('val_loss') # get the validation loss for this epoch\n"," if v_loss< self.lowest_vloss:\n"," self.lowest_vloss=v_loss\n"," self.best_weights=self.user_model.get_weights() # Changed to self.user_model\n"," self.best_epoch=epoch + 1\n"," print (f'\\n validation loss of {v_loss:7.4f} is below lowest loss, saving weights from epoch {str(epoch + 1):3s} as best weights')\n"," else:\n"," print (f'\\n validation loss of {v_loss:7.4f} is above lowest loss of {self.lowest_vloss:7.4f} keeping weights from epoch {str(self.best_epoch)} as best weights')\n","\n"," if self.ask: # are the conditions right to query the user?\n"," if epoch + 1 ==self.ask_epoch: # is this epoch the one for quering the user?\n"," print('\\n Enter H to end training or an integer for the number of additional epochs to run then ask again')\n"," ans=input()\n","\n"," if ans == 'H' or ans =='h' or ans == '0': # quit training for these conditions\n"," print ('you entered ', ans, ' Training halted on epoch ', epoch+1, ' due to user input\\n', flush=True)\n"," self.model.stop_training = True # halt training (this self.model is the Keras callback's model)\n"," else: # user wants to continue training\n"," self.ask_epoch += int(ans)\n"," if self.ask_epoch > self.epochs:\n"," print('\\nYou specified maximum epochs of as ', self.epochs, ' cannot train for ', self.ask_epoch, flush =True)\n"," else:\n"," print ('you entered ', ans, ' Training will continue to epoch ', self.ask_epoch, flush=True)\n"," # Fixed: Changed from .lr to .learning_rate\n"," lr=float(tf.keras.backend.get_value(self.user_model.optimizer.learning_rate))\n"," print(f'current LR is {lr:7.5f} hit enter to keep this LR or enter a new LR')\n"," ans=input(' ')\n"," if ans =='':\n"," print (f'keeping current LR of {lr:7.5f}')\n"," else:\n"," new_lr=float(ans)\n"," # Fixed: Changed from .lr to .learning_rate\n"," tf.keras.backend.set_value(self.user_model.optimizer.learning_rate, new_lr)\n"," print(' changing LR to ', ans)"]},{"cell_type":"markdown","metadata":{"id":"3uWPcdEcFPTe"},"source":["\n","#
Instantiate custom callback"]},{"cell_type":"code","execution_count":15,"metadata":{"colab":{"base_uri":"https://localhost:8080/"},"id":"FAGhAW_HcKYy","executionInfo":{"status":"ok","timestamp":1764239119526,"user_tz":-420,"elapsed":6,"user":{"displayName":"Ping Ping","userId":"16193029904393259603"}},"outputId":"d2be9a46-f8d6-4976-f9e5-325bb2de401f"},"outputs":[{"output_type":"stream","name":"stdout","text":["Rebuilding tf.data datasets with robust PIL loaders...\n"," Train steps (batches): 24\n"," Valid steps (batches): 7\n"," Test steps (batches): 7\n"]}],"source":["os.makedirs(quarantine_dir, exist_ok=True)\n","\n","def _decode_and_preprocess_train(path, label):\n"," p = path.numpy().decode('utf-8')\n"," try:\n"," with Image.open(p) as im:\n"," im = im.convert('RGB')\n"," im = im.resize((img_size[1], img_size[0]))\n"," img = np.array(im, dtype=np.uint8)\n"," except Exception as e:\n"," # Move problematic file to quarantine to avoid repeated failures\n"," try:\n"," dest = os.path.join(quarantine_dir, os.path.basename(p))\n"," shutil.move(p, dest)\n"," print('Quarantined bad file:', p)\n"," except Exception:\n"," pass\n"," img = np.zeros((img_size[0], img_size[1], 3), dtype=np.uint8)\n"," img_pp = albumentations_preprocess(img)\n"," label_ohe = tf.one_hot(int(label.numpy()), depth=class_count).numpy()\n"," return img_pp, label_ohe\n","\n","\n","def _decode_and_preprocess_eval(path, label):\n"," p = path.numpy().decode('utf-8')\n"," try:\n"," with Image.open(p) as im:\n"," im = im.convert('RGB')\n"," im = im.resize((img_size[1], img_size[0]))\n"," img = np.array(im, dtype=np.float32)\n"," except Exception:\n"," img = np.zeros((img_size[0], img_size[1], 3), dtype=np.float32)\n"," img_pp = tf.keras.applications.efficientnet.preprocess_input(img)\n"," label_ohe = tf.one_hot(int(label.numpy()), depth=class_count).numpy()\n"," return img_pp, label_ohe\n","\n","# Rebuild TF datasets using the shape-aware wrappers defined earlier\n","print('Rebuilding tf.data datasets with robust PIL loaders...')\n","train_paths = train_df['filepaths'].astype(str).values\n","train_labels = train_df['labels'].map(class_to_idx).values.astype('int32')\n","valid_paths = valid_df['filepaths'].astype(str).values\n","valid_labels = valid_df['labels'].map(class_to_idx).values.astype('int32')\n","test_paths = test_df['filepaths'].astype(str).values\n","test_labels = test_df['labels'].map(class_to_idx).values.astype('int32')\n","\n","train_ds = tf.data.Dataset.from_tensor_slices((train_paths, train_labels))\n","train_ds = train_ds.shuffle(buffer_size=1000, seed=123)\n","train_ds = train_ds.map(_py_train, num_parallel_calls=AUTOTUNE)\n","train_ds = train_ds.batch(batch_size).prefetch(AUTOTUNE)\n","\n","valid_ds = tf.data.Dataset.from_tensor_slices((valid_paths, valid_labels))\n","valid_ds = valid_ds.map(_py_eval, num_parallel_calls=AUTOTUNE)\n","valid_ds = valid_ds.batch(batch_size).prefetch(AUTOTUNE)\n","\n","test_ds = tf.data.Dataset.from_tensor_slices((test_paths, test_labels))\n","test_ds = test_ds.map(_py_eval, num_parallel_calls=AUTOTUNE)\n","test_ds = test_ds.batch(batch_size).prefetch(AUTOTUNE)\n","\n","# Recompute steps\n","train_steps = int(np.ceil(len(train_df) / batch_size))\n","valid_steps = int(np.ceil(len(valid_df) / batch_size))\n","test_steps = int(np.ceil(len(test_df) / batch_size))\n","\n","print(f' Train steps (batches): {train_steps}')\n","print(f' Valid steps (batches): {valid_steps}')\n","print(f' Test steps (batches): {test_steps}')"]},{"cell_type":"code","execution_count":16,"metadata":{"colab":{"base_uri":"https://localhost:8080/"},"id":"CQAQ0uoYFPTe","executionInfo":{"status":"ok","timestamp":1764239119533,"user_tz":-420,"elapsed":6,"user":{"displayName":"Ping Ping","userId":"16193029904393259603"}},"outputId":"0de90800-93c1-47e1-d09c-cafbb65e7b9e"},"outputs":[{"output_type":"stream","name":"stdout","text":["✓ Callbacks configured:\n"]}],"source":["from tensorflow.keras.callbacks import ModelCheckpoint, EarlyStopping, ReduceLROnPlateau\n","\n","epochs = 150 # Increased for better convergence\n","\n","# Save best model to Google Drive\n","checkpoint = ModelCheckpoint(\n"," '/content/drive/MyDrive/best_cauliflower_model.h5',\n"," monitor='val_loss',\n"," save_best_only=True,\n"," verbose=1,\n"," mode='min'\n",")\n","\n","# Reduce learning rate when validation loss plateaus\n","reduce_lr = ReduceLROnPlateau(\n"," monitor='val_loss',\n"," factor=0.5, # Reduce LR by half\n"," patience=4, # Wait 3 epochs before reducing\n"," min_lr=1e-7,\n"," verbose=1,\n"," mode='min'\n",")\n","\n","# Stop early if no improvement\n","early_stop = EarlyStopping(\n"," monitor='val_loss',\n"," patience=12, # Increased patience for better training\n"," restore_best_weights=True,\n"," verbose=1,\n"," mode='min'\n",")\n","\n","callbacks = [checkpoint, reduce_lr, early_stop]\n","print('✓ Callbacks configured:')"]},{"cell_type":"markdown","metadata":{"id":"pOalrkwLFPTe"},"source":["\n","#
Train the model\n","### Note unlike how you are told it is BETTER to make the base model trainable from the outset\n","### It will converge faster and have a lower validation losss"]},{"cell_type":"code","execution_count":17,"metadata":{"colab":{"base_uri":"https://localhost:8080/"},"id":"xob3iOCiFPTf","executionInfo":{"status":"ok","timestamp":1764242563227,"user_tz":-420,"elapsed":3443693,"user":{"displayName":"Ping Ping","userId":"16193029904393259603"}},"outputId":"4f987898-ff8b-4215-d314-5462403ab4c1"},"outputs":[{"output_type":"stream","name":"stdout","text":["STARTING TRAINING\n","================================================================================\n","Epochs: 150\n","Batch size: 128\n","Steps per epoch: 24\n","Validation steps: 7\n","================================================================================\n","Epoch 1/150\n","\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 7s/step - accuracy: 0.4333 - loss: 3.0678 \n","Epoch 1: val_loss improved from inf to 2.99970, saving model to /content/drive/MyDrive/best_cauliflower_model.h5\n"]},{"output_type":"stream","name":"stderr","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"]},{"output_type":"stream","name":"stdout","text":["\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m440s\u001b[0m 9s/step - accuracy: 0.4319 - loss: 3.0780 - val_accuracy: 0.5071 - val_loss: 2.9997 - learning_rate: 5.0000e-04\n","Epoch 2/150\n","\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 651ms/step - accuracy: 0.3664 - loss: 2.8209\n","Epoch 2: val_loss did not improve from 2.99970\n","\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m18s\u001b[0m 758ms/step - accuracy: 0.3709 - loss: 2.8078 - val_accuracy: 0.4426 - val_loss: 3.0399 - learning_rate: 5.0000e-04\n","Epoch 3/150\n","\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 644ms/step - accuracy: 0.5944 - loss: 2.1612\n","Epoch 3: val_loss improved from 2.99970 to 2.40780, saving model to /content/drive/MyDrive/best_cauliflower_model.h5\n"]},{"output_type":"stream","name":"stderr","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"]},{"output_type":"stream","name":"stdout","text":["\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m21s\u001b[0m 892ms/step - accuracy: 0.5964 - loss: 2.1575 - val_accuracy: 0.5974 - val_loss: 2.4078 - learning_rate: 5.0000e-04\n","Epoch 4/150\n","\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 642ms/step - accuracy: 0.6577 - loss: 1.9481\n","Epoch 4: val_loss improved from 2.40780 to 1.80148, saving model to /content/drive/MyDrive/best_cauliflower_model.h5\n"]},{"output_type":"stream","name":"stderr","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"]},{"output_type":"stream","name":"stdout","text":["\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m20s\u001b[0m 819ms/step - accuracy: 0.6594 - loss: 1.9447 - val_accuracy: 0.7432 - val_loss: 1.8015 - learning_rate: 5.0000e-04\n","Epoch 5/150\n","\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 645ms/step - accuracy: 0.7470 - loss: 1.7310\n","Epoch 5: val_loss improved from 1.80148 to 1.40463, saving model to /content/drive/MyDrive/best_cauliflower_model.h5\n"]},{"output_type":"stream","name":"stderr","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"]},{"output_type":"stream","name":"stdout","text":["\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m21s\u001b[0m 892ms/step - accuracy: 0.7481 - loss: 1.7291 - val_accuracy: 0.8787 - val_loss: 1.4046 - learning_rate: 5.0000e-04\n","Epoch 6/150\n","\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 639ms/step - accuracy: 0.7524 - loss: 1.7050\n","Epoch 6: val_loss improved from 1.40463 to 1.27108, saving model to /content/drive/MyDrive/best_cauliflower_model.h5\n"]},{"output_type":"stream","name":"stderr","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"]},{"output_type":"stream","name":"stdout","text":["\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m21s\u001b[0m 887ms/step - accuracy: 0.7540 - loss: 1.7018 - val_accuracy: 0.9316 - val_loss: 1.2711 - learning_rate: 5.0000e-04\n","Epoch 7/150\n","\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 645ms/step - accuracy: 0.8180 - loss: 1.5672\n","Epoch 7: val_loss improved from 1.27108 to 1.18150, saving model to /content/drive/MyDrive/best_cauliflower_model.h5\n"]},{"output_type":"stream","name":"stderr","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"]},{"output_type":"stream","name":"stdout","text":["\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m20s\u001b[0m 822ms/step - accuracy: 0.8194 - loss: 1.5638 - val_accuracy: 0.9626 - val_loss: 1.1815 - learning_rate: 5.0000e-04\n","Epoch 8/150\n","\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 643ms/step - accuracy: 0.8405 - loss: 1.5180\n","Epoch 8: val_loss improved from 1.18150 to 1.16558, saving model to /content/drive/MyDrive/best_cauliflower_model.h5\n"]},{"output_type":"stream","name":"stderr","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"]},{"output_type":"stream","name":"stdout","text":["\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m21s\u001b[0m 896ms/step - accuracy: 0.8412 - loss: 1.5156 - val_accuracy: 0.9665 - val_loss: 1.1656 - learning_rate: 5.0000e-04\n","Epoch 9/150\n","\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 640ms/step - accuracy: 0.8709 - loss: 1.4390\n","Epoch 9: val_loss improved from 1.16558 to 1.12696, saving model to /content/drive/MyDrive/best_cauliflower_model.h5\n"]},{"output_type":"stream","name":"stderr","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"]},{"output_type":"stream","name":"stdout","text":["\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m21s\u001b[0m 879ms/step - accuracy: 0.8719 - loss: 1.4365 - val_accuracy: 0.9794 - val_loss: 1.1270 - learning_rate: 5.0000e-04\n","Epoch 10/150\n","\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 657ms/step - accuracy: 0.8940 - loss: 1.3669\n","Epoch 10: val_loss improved from 1.12696 to 1.11380, saving model to /content/drive/MyDrive/best_cauliflower_model.h5\n"]},{"output_type":"stream","name":"stderr","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"]},{"output_type":"stream","name":"stdout","text":["\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m20s\u001b[0m 834ms/step - accuracy: 0.8945 - loss: 1.3657 - val_accuracy: 0.9845 - val_loss: 1.1138 - learning_rate: 5.0000e-04\n","Epoch 11/150\n","\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 643ms/step - accuracy: 0.9018 - loss: 1.3218\n","Epoch 11: val_loss improved from 1.11380 to 1.10739, saving model to /content/drive/MyDrive/best_cauliflower_model.h5\n"]},{"output_type":"stream","name":"stderr","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"]},{"output_type":"stream","name":"stdout","text":["\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m21s\u001b[0m 893ms/step - accuracy: 0.9025 - loss: 1.3205 - val_accuracy: 0.9819 - val_loss: 1.1074 - learning_rate: 5.0000e-04\n","Epoch 12/150\n","\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 638ms/step - accuracy: 0.9264 - loss: 1.2684\n","Epoch 12: val_loss improved from 1.10739 to 1.09452, saving model to /content/drive/MyDrive/best_cauliflower_model.h5\n"]},{"output_type":"stream","name":"stderr","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"]},{"output_type":"stream","name":"stdout","text":["\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m20s\u001b[0m 826ms/step - accuracy: 0.9268 - loss: 1.2677 - val_accuracy: 0.9884 - val_loss: 1.0945 - learning_rate: 5.0000e-04\n","Epoch 13/150\n","\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 645ms/step - accuracy: 0.9310 - loss: 1.2420\n","Epoch 13: val_loss improved from 1.09452 to 1.09037, saving model to /content/drive/MyDrive/best_cauliflower_model.h5\n"]},{"output_type":"stream","name":"stderr","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"]},{"output_type":"stream","name":"stdout","text":["\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m20s\u001b[0m 837ms/step - accuracy: 0.9315 - loss: 1.2409 - val_accuracy: 0.9884 - val_loss: 1.0904 - learning_rate: 5.0000e-04\n","Epoch 14/150\n","\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 641ms/step - accuracy: 0.9418 - loss: 1.2119\n","Epoch 14: val_loss improved from 1.09037 to 1.08318, saving model to /content/drive/MyDrive/best_cauliflower_model.h5\n"]},{"output_type":"stream","name":"stderr","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"]},{"output_type":"stream","name":"stdout","text":["\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m21s\u001b[0m 886ms/step - accuracy: 0.9421 - loss: 1.2114 - val_accuracy: 0.9897 - val_loss: 1.0832 - learning_rate: 5.0000e-04\n","Epoch 15/150\n","\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 636ms/step - accuracy: 0.9601 - loss: 1.1633\n","Epoch 15: val_loss improved from 1.08318 to 1.07057, saving model to /content/drive/MyDrive/best_cauliflower_model.h5\n"]},{"output_type":"stream","name":"stderr","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"]},{"output_type":"stream","name":"stdout","text":["\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m19s\u001b[0m 814ms/step - accuracy: 0.9603 - loss: 1.1629 - val_accuracy: 0.9935 - val_loss: 1.0706 - learning_rate: 5.0000e-04\n","Epoch 16/150\n","\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 643ms/step - accuracy: 0.9604 - loss: 1.1637\n","Epoch 16: val_loss improved from 1.07057 to 1.06280, saving model to /content/drive/MyDrive/best_cauliflower_model.h5\n"]},{"output_type":"stream","name":"stderr","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"]},{"output_type":"stream","name":"stdout","text":["\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m21s\u001b[0m 882ms/step - accuracy: 0.9607 - loss: 1.1630 - val_accuracy: 0.9923 - val_loss: 1.0628 - learning_rate: 5.0000e-04\n","Epoch 17/150\n","\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 638ms/step - accuracy: 0.9661 - loss: 1.1505\n","Epoch 17: val_loss improved from 1.06280 to 1.06224, saving model to /content/drive/MyDrive/best_cauliflower_model.h5\n"]},{"output_type":"stream","name":"stderr","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"]},{"output_type":"stream","name":"stdout","text":["\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m21s\u001b[0m 890ms/step - accuracy: 0.9663 - loss: 1.1501 - val_accuracy: 0.9897 - val_loss: 1.0622 - learning_rate: 5.0000e-04\n","Epoch 18/150\n","\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 638ms/step - accuracy: 0.9764 - loss: 1.1157\n","Epoch 18: val_loss improved from 1.06224 to 1.05840, saving model to /content/drive/MyDrive/best_cauliflower_model.h5\n"]},{"output_type":"stream","name":"stderr","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"]},{"output_type":"stream","name":"stdout","text":["\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m21s\u001b[0m 888ms/step - accuracy: 0.9764 - loss: 1.1156 - val_accuracy: 0.9910 - val_loss: 1.0584 - learning_rate: 5.0000e-04\n","Epoch 19/150\n","\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 649ms/step - accuracy: 0.9729 - loss: 1.1051\n","Epoch 19: val_loss improved from 1.05840 to 1.05251, saving model to /content/drive/MyDrive/best_cauliflower_model.h5\n"]},{"output_type":"stream","name":"stderr","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"]},{"output_type":"stream","name":"stdout","text":["\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m21s\u001b[0m 890ms/step - accuracy: 0.9728 - loss: 1.1053 - val_accuracy: 0.9910 - val_loss: 1.0525 - learning_rate: 5.0000e-04\n","Epoch 20/150\n","\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 645ms/step - accuracy: 0.9612 - loss: 1.1271\n","Epoch 20: val_loss improved from 1.05251 to 1.04213, saving model to /content/drive/MyDrive/best_cauliflower_model.h5\n"]},{"output_type":"stream","name":"stderr","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"]},{"output_type":"stream","name":"stdout","text":["\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m21s\u001b[0m 881ms/step - accuracy: 0.9614 - loss: 1.1266 - val_accuracy: 0.9910 - val_loss: 1.0421 - learning_rate: 5.0000e-04\n","Epoch 21/150\n","\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 639ms/step - accuracy: 0.9748 - loss: 1.0898\n","Epoch 21: val_loss improved from 1.04213 to 1.03727, saving model to /content/drive/MyDrive/best_cauliflower_model.h5\n"]},{"output_type":"stream","name":"stderr","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"]},{"output_type":"stream","name":"stdout","text":["\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m20s\u001b[0m 820ms/step - accuracy: 0.9749 - loss: 1.0898 - val_accuracy: 0.9948 - val_loss: 1.0373 - learning_rate: 5.0000e-04\n","Epoch 22/150\n","\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 645ms/step - accuracy: 0.9801 - loss: 1.0834\n","Epoch 22: val_loss improved from 1.03727 to 1.02597, saving model to /content/drive/MyDrive/best_cauliflower_model.h5\n"]},{"output_type":"stream","name":"stderr","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"]},{"output_type":"stream","name":"stdout","text":["\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m22s\u001b[0m 921ms/step - accuracy: 0.9802 - loss: 1.0831 - val_accuracy: 0.9987 - val_loss: 1.0260 - learning_rate: 5.0000e-04\n","Epoch 23/150\n","\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 643ms/step - accuracy: 0.9727 - loss: 1.0938\n","Epoch 23: val_loss did not improve from 1.02597\n","\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m18s\u001b[0m 748ms/step - accuracy: 0.9729 - loss: 1.0932 - val_accuracy: 0.9910 - val_loss: 1.0357 - learning_rate: 5.0000e-04\n","Epoch 24/150\n","\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 637ms/step - accuracy: 0.9844 - loss: 1.0551\n","Epoch 24: val_loss improved from 1.02597 to 1.01682, saving model to /content/drive/MyDrive/best_cauliflower_model.h5\n"]},{"output_type":"stream","name":"stderr","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"]},{"output_type":"stream","name":"stdout","text":["\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m21s\u001b[0m 882ms/step - accuracy: 0.9844 - loss: 1.0550 - val_accuracy: 0.9974 - val_loss: 1.0168 - learning_rate: 5.0000e-04\n","Epoch 25/150\n","\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 650ms/step - accuracy: 0.9896 - loss: 1.0381\n","Epoch 25: val_loss improved from 1.01682 to 1.01030, saving model to /content/drive/MyDrive/best_cauliflower_model.h5\n"]},{"output_type":"stream","name":"stderr","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"]},{"output_type":"stream","name":"stdout","text":["\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m21s\u001b[0m 896ms/step - accuracy: 0.9895 - loss: 1.0383 - val_accuracy: 0.9974 - val_loss: 1.0103 - learning_rate: 5.0000e-04\n","Epoch 26/150\n","\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 647ms/step - accuracy: 0.9875 - loss: 1.0463\n","Epoch 26: val_loss improved from 1.01030 to 1.00553, saving model to /content/drive/MyDrive/best_cauliflower_model.h5\n"]},{"output_type":"stream","name":"stderr","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"]},{"output_type":"stream","name":"stdout","text":["\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m20s\u001b[0m 833ms/step - accuracy: 0.9875 - loss: 1.0460 - val_accuracy: 0.9935 - val_loss: 1.0055 - learning_rate: 5.0000e-04\n","Epoch 27/150\n","\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 640ms/step - accuracy: 0.9892 - loss: 1.0331\n","Epoch 27: val_loss improved from 1.00553 to 1.00217, saving model to /content/drive/MyDrive/best_cauliflower_model.h5\n"]},{"output_type":"stream","name":"stderr","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"]},{"output_type":"stream","name":"stdout","text":["\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m21s\u001b[0m 877ms/step - accuracy: 0.9891 - loss: 1.0334 - val_accuracy: 0.9935 - val_loss: 1.0022 - learning_rate: 5.0000e-04\n","Epoch 28/150\n","\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 643ms/step - accuracy: 0.9892 - loss: 1.0172\n","Epoch 28: val_loss improved from 1.00217 to 0.99440, saving model to /content/drive/MyDrive/best_cauliflower_model.h5\n"]},{"output_type":"stream","name":"stderr","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"]},{"output_type":"stream","name":"stdout","text":["\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m20s\u001b[0m 824ms/step - accuracy: 0.9891 - loss: 1.0172 - val_accuracy: 0.9961 - val_loss: 0.9944 - learning_rate: 5.0000e-04\n","Epoch 29/150\n","\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 643ms/step - accuracy: 0.9931 - loss: 1.0092\n","Epoch 29: val_loss improved from 0.99440 to 0.99361, saving model to /content/drive/MyDrive/best_cauliflower_model.h5\n"]},{"output_type":"stream","name":"stderr","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"]},{"output_type":"stream","name":"stdout","text":["\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m20s\u001b[0m 823ms/step - accuracy: 0.9930 - loss: 1.0093 - val_accuracy: 0.9948 - val_loss: 0.9936 - learning_rate: 5.0000e-04\n","Epoch 30/150\n","\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 643ms/step - accuracy: 0.9854 - loss: 1.0231\n","Epoch 30: val_loss improved from 0.99361 to 0.98258, saving model to /content/drive/MyDrive/best_cauliflower_model.h5\n"]},{"output_type":"stream","name":"stderr","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"]},{"output_type":"stream","name":"stdout","text":["\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m21s\u001b[0m 883ms/step - accuracy: 0.9855 - loss: 1.0226 - val_accuracy: 0.9987 - val_loss: 0.9826 - learning_rate: 5.0000e-04\n","Epoch 31/150\n","\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 648ms/step - accuracy: 0.9857 - loss: 1.0171\n","Epoch 31: val_loss improved from 0.98258 to 0.97777, saving model to /content/drive/MyDrive/best_cauliflower_model.h5\n"]},{"output_type":"stream","name":"stderr","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"]},{"output_type":"stream","name":"stdout","text":["\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m20s\u001b[0m 836ms/step - accuracy: 0.9857 - loss: 1.0168 - val_accuracy: 0.9923 - val_loss: 0.9778 - learning_rate: 5.0000e-04\n","Epoch 32/150\n","\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 639ms/step - accuracy: 0.9929 - loss: 0.9873\n","Epoch 32: val_loss did not improve from 0.97777\n","\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m18s\u001b[0m 743ms/step - accuracy: 0.9928 - loss: 0.9873 - val_accuracy: 0.9910 - val_loss: 0.9889 - learning_rate: 5.0000e-04\n","Epoch 33/150\n","\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 637ms/step - accuracy: 0.9881 - loss: 0.9826\n","Epoch 33: val_loss did not improve from 0.97777\n","\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m18s\u001b[0m 741ms/step - accuracy: 0.9881 - loss: 0.9824 - val_accuracy: 0.9897 - val_loss: 0.9798 - learning_rate: 5.0000e-04\n","Epoch 34/150\n","\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 638ms/step - accuracy: 0.9911 - loss: 0.9794\n","Epoch 34: val_loss improved from 0.97777 to 0.97111, saving model to /content/drive/MyDrive/best_cauliflower_model.h5\n"]},{"output_type":"stream","name":"stderr","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"]},{"output_type":"stream","name":"stdout","text":["\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m21s\u001b[0m 881ms/step - accuracy: 0.9911 - loss: 0.9793 - val_accuracy: 0.9923 - val_loss: 0.9711 - learning_rate: 5.0000e-04\n","Epoch 35/150\n","\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 643ms/step - accuracy: 0.9916 - loss: 0.9654\n","Epoch 35: val_loss improved from 0.97111 to 0.95788, saving model to /content/drive/MyDrive/best_cauliflower_model.h5\n"]},{"output_type":"stream","name":"stderr","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"]},{"output_type":"stream","name":"stdout","text":["\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m20s\u001b[0m 821ms/step - accuracy: 0.9916 - loss: 0.9653 - val_accuracy: 0.9948 - val_loss: 0.9579 - learning_rate: 5.0000e-04\n","Epoch 36/150\n","\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 642ms/step - accuracy: 0.9881 - loss: 0.9717\n","Epoch 36: val_loss improved from 0.95788 to 0.94274, saving model to /content/drive/MyDrive/best_cauliflower_model.h5\n"]},{"output_type":"stream","name":"stderr","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"]},{"output_type":"stream","name":"stdout","text":["\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m20s\u001b[0m 821ms/step - accuracy: 0.9881 - loss: 0.9716 - val_accuracy: 0.9961 - val_loss: 0.9427 - learning_rate: 5.0000e-04\n","Epoch 37/150\n","\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 648ms/step - accuracy: 0.9923 - loss: 0.9539\n","Epoch 37: val_loss improved from 0.94274 to 0.93951, saving model to /content/drive/MyDrive/best_cauliflower_model.h5\n"]},{"output_type":"stream","name":"stderr","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"]},{"output_type":"stream","name":"stdout","text":["\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m20s\u001b[0m 827ms/step - accuracy: 0.9923 - loss: 0.9539 - val_accuracy: 0.9935 - val_loss: 0.9395 - learning_rate: 5.0000e-04\n","Epoch 38/150\n","\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 640ms/step - accuracy: 0.9872 - loss: 0.9592\n","Epoch 38: val_loss improved from 0.93951 to 0.93673, saving model to /content/drive/MyDrive/best_cauliflower_model.h5\n"]},{"output_type":"stream","name":"stderr","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"]},{"output_type":"stream","name":"stdout","text":["\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m21s\u001b[0m 888ms/step - accuracy: 0.9873 - loss: 0.9590 - val_accuracy: 0.9935 - val_loss: 0.9367 - learning_rate: 5.0000e-04\n","Epoch 39/150\n","\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 648ms/step - accuracy: 0.9913 - loss: 0.9408\n","Epoch 39: val_loss improved from 0.93673 to 0.92579, saving model to /content/drive/MyDrive/best_cauliflower_model.h5\n"]},{"output_type":"stream","name":"stderr","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"]},{"output_type":"stream","name":"stdout","text":["\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m20s\u001b[0m 828ms/step - accuracy: 0.9913 - loss: 0.9408 - val_accuracy: 0.9961 - val_loss: 0.9258 - learning_rate: 5.0000e-04\n","Epoch 40/150\n","\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 636ms/step - accuracy: 0.9916 - loss: 0.9366\n","Epoch 40: val_loss improved from 0.92579 to 0.92142, saving model to /content/drive/MyDrive/best_cauliflower_model.h5\n"]},{"output_type":"stream","name":"stderr","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"]},{"output_type":"stream","name":"stdout","text":["\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m21s\u001b[0m 885ms/step - accuracy: 0.9915 - loss: 0.9366 - val_accuracy: 0.9935 - val_loss: 0.9214 - learning_rate: 5.0000e-04\n","Epoch 41/150\n","\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 651ms/step - accuracy: 0.9964 - loss: 0.9246\n","Epoch 41: val_loss did not improve from 0.92142\n","\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m18s\u001b[0m 753ms/step - accuracy: 0.9964 - loss: 0.9246 - val_accuracy: 0.9897 - val_loss: 0.9320 - learning_rate: 5.0000e-04\n","Epoch 42/150\n","\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 639ms/step - accuracy: 0.9924 - loss: 0.9244\n","Epoch 42: val_loss improved from 0.92142 to 0.90841, saving model to /content/drive/MyDrive/best_cauliflower_model.h5\n"]},{"output_type":"stream","name":"stderr","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"]},{"output_type":"stream","name":"stdout","text":["\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m21s\u001b[0m 893ms/step - accuracy: 0.9924 - loss: 0.9242 - val_accuracy: 0.9935 - val_loss: 0.9084 - learning_rate: 5.0000e-04\n","Epoch 43/150\n","\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 640ms/step - accuracy: 0.9947 - loss: 0.9066\n","Epoch 43: val_loss improved from 0.90841 to 0.89584, saving model to /content/drive/MyDrive/best_cauliflower_model.h5\n"]},{"output_type":"stream","name":"stderr","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"]},{"output_type":"stream","name":"stdout","text":["\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m21s\u001b[0m 875ms/step - accuracy: 0.9947 - loss: 0.9067 - val_accuracy: 0.9948 - val_loss: 0.8958 - learning_rate: 5.0000e-04\n","Epoch 44/150\n","\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 633ms/step - accuracy: 0.9957 - loss: 0.9014\n","Epoch 44: val_loss improved from 0.89584 to 0.89436, saving model to /content/drive/MyDrive/best_cauliflower_model.h5\n"]},{"output_type":"stream","name":"stderr","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"]},{"output_type":"stream","name":"stdout","text":["\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m21s\u001b[0m 868ms/step - accuracy: 0.9956 - loss: 0.9015 - val_accuracy: 0.9923 - val_loss: 0.8944 - learning_rate: 5.0000e-04\n","Epoch 45/150\n","\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 635ms/step - accuracy: 0.9925 - loss: 0.8979\n","Epoch 45: val_loss improved from 0.89436 to 0.89243, saving model to /content/drive/MyDrive/best_cauliflower_model.h5\n"]},{"output_type":"stream","name":"stderr","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"]},{"output_type":"stream","name":"stdout","text":["\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m21s\u001b[0m 873ms/step - accuracy: 0.9926 - loss: 0.8978 - val_accuracy: 0.9923 - val_loss: 0.8924 - learning_rate: 5.0000e-04\n","Epoch 46/150\n","\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 633ms/step - accuracy: 0.9959 - loss: 0.8852\n","Epoch 46: val_loss improved from 0.89243 to 0.87678, saving model to /content/drive/MyDrive/best_cauliflower_model.h5\n"]},{"output_type":"stream","name":"stderr","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"]},{"output_type":"stream","name":"stdout","text":["\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m21s\u001b[0m 869ms/step - accuracy: 0.9959 - loss: 0.8853 - val_accuracy: 0.9948 - val_loss: 0.8768 - learning_rate: 5.0000e-04\n","Epoch 47/150\n","\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 634ms/step - accuracy: 0.9940 - loss: 0.8782\n","Epoch 47: val_loss improved from 0.87678 to 0.86299, saving model to /content/drive/MyDrive/best_cauliflower_model.h5\n"]},{"output_type":"stream","name":"stderr","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"]},{"output_type":"stream","name":"stdout","text":["\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m19s\u001b[0m 810ms/step - accuracy: 0.9940 - loss: 0.8781 - val_accuracy: 0.9974 - val_loss: 0.8630 - learning_rate: 5.0000e-04\n","Epoch 48/150\n","\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 636ms/step - accuracy: 0.9957 - loss: 0.8707\n","Epoch 48: val_loss did not improve from 0.86299\n","\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m18s\u001b[0m 738ms/step - accuracy: 0.9957 - loss: 0.8706 - val_accuracy: 0.9948 - val_loss: 0.8677 - learning_rate: 5.0000e-04\n","Epoch 49/150\n","\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 631ms/step - accuracy: 0.9950 - loss: 0.8620\n","Epoch 49: val_loss improved from 0.86299 to 0.85525, saving model to /content/drive/MyDrive/best_cauliflower_model.h5\n"]},{"output_type":"stream","name":"stderr","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"]},{"output_type":"stream","name":"stdout","text":["\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m21s\u001b[0m 876ms/step - accuracy: 0.9951 - loss: 0.8619 - val_accuracy: 0.9935 - val_loss: 0.8553 - learning_rate: 5.0000e-04\n","Epoch 50/150\n","\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 635ms/step - accuracy: 0.9972 - loss: 0.8511\n","Epoch 50: val_loss improved from 0.85525 to 0.84682, saving model to /content/drive/MyDrive/best_cauliflower_model.h5\n"]},{"output_type":"stream","name":"stderr","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"]},{"output_type":"stream","name":"stdout","text":["\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m21s\u001b[0m 868ms/step - accuracy: 0.9972 - loss: 0.8510 - val_accuracy: 0.9961 - val_loss: 0.8468 - learning_rate: 5.0000e-04\n","Epoch 51/150\n","\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 635ms/step - accuracy: 0.9979 - loss: 0.8444\n","Epoch 51: val_loss improved from 0.84682 to 0.83883, saving model to /content/drive/MyDrive/best_cauliflower_model.h5\n"]},{"output_type":"stream","name":"stderr","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"]},{"output_type":"stream","name":"stdout","text":["\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m21s\u001b[0m 872ms/step - accuracy: 0.9978 - loss: 0.8443 - val_accuracy: 0.9961 - val_loss: 0.8388 - learning_rate: 5.0000e-04\n","Epoch 52/150\n","\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 634ms/step - accuracy: 0.9976 - loss: 0.8332\n","Epoch 52: val_loss improved from 0.83883 to 0.83020, saving model to /content/drive/MyDrive/best_cauliflower_model.h5\n"]},{"output_type":"stream","name":"stderr","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"]},{"output_type":"stream","name":"stdout","text":["\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m19s\u001b[0m 811ms/step - accuracy: 0.9976 - loss: 0.8332 - val_accuracy: 0.9935 - val_loss: 0.8302 - learning_rate: 5.0000e-04\n","Epoch 53/150\n","\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 639ms/step - accuracy: 0.9976 - loss: 0.8267\n","Epoch 53: val_loss improved from 0.83020 to 0.82840, saving model to /content/drive/MyDrive/best_cauliflower_model.h5\n"]},{"output_type":"stream","name":"stderr","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"]},{"output_type":"stream","name":"stdout","text":["\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m21s\u001b[0m 873ms/step - accuracy: 0.9976 - loss: 0.8267 - val_accuracy: 0.9961 - val_loss: 0.8284 - learning_rate: 5.0000e-04\n","Epoch 54/150\n","\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 635ms/step - accuracy: 0.9928 - loss: 0.8248\n","Epoch 54: val_loss improved from 0.82840 to 0.80889, saving model to /content/drive/MyDrive/best_cauliflower_model.h5\n"]},{"output_type":"stream","name":"stderr","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"]},{"output_type":"stream","name":"stdout","text":["\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m21s\u001b[0m 871ms/step - accuracy: 0.9929 - loss: 0.8248 - val_accuracy: 0.9974 - val_loss: 0.8089 - learning_rate: 5.0000e-04\n","Epoch 55/150\n","\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 634ms/step - accuracy: 0.9967 - loss: 0.8102\n","Epoch 55: val_loss did not improve from 0.80889\n","\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m18s\u001b[0m 738ms/step - accuracy: 0.9967 - loss: 0.8101 - val_accuracy: 0.9935 - val_loss: 0.8188 - learning_rate: 5.0000e-04\n","Epoch 56/150\n","\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 635ms/step - accuracy: 0.9961 - loss: 0.8080\n","Epoch 56: val_loss improved from 0.80889 to 0.79353, saving model to /content/drive/MyDrive/best_cauliflower_model.h5\n"]},{"output_type":"stream","name":"stderr","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"]},{"output_type":"stream","name":"stdout","text":["\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m21s\u001b[0m 892ms/step - accuracy: 0.9961 - loss: 0.8079 - val_accuracy: 0.9974 - val_loss: 0.7935 - learning_rate: 5.0000e-04\n","Epoch 57/150\n","\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 631ms/step - accuracy: 0.9972 - loss: 0.7980\n","Epoch 57: val_loss did not improve from 0.79353\n","\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m18s\u001b[0m 734ms/step - accuracy: 0.9972 - loss: 0.7980 - val_accuracy: 0.9961 - val_loss: 0.7942 - learning_rate: 5.0000e-04\n","Epoch 58/150\n","\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 630ms/step - accuracy: 0.9943 - loss: 0.7894\n","Epoch 58: val_loss improved from 0.79353 to 0.78713, saving model to /content/drive/MyDrive/best_cauliflower_model.h5\n"]},{"output_type":"stream","name":"stderr","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"]},{"output_type":"stream","name":"stdout","text":["\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m21s\u001b[0m 880ms/step - accuracy: 0.9943 - loss: 0.7892 - val_accuracy: 0.9935 - val_loss: 0.7871 - learning_rate: 5.0000e-04\n","Epoch 59/150\n","\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 633ms/step - accuracy: 0.9942 - loss: 0.7827\n","Epoch 59: val_loss improved from 0.78713 to 0.77502, saving model to /content/drive/MyDrive/best_cauliflower_model.h5\n"]},{"output_type":"stream","name":"stderr","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"]},{"output_type":"stream","name":"stdout","text":["\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m21s\u001b[0m 877ms/step - accuracy: 0.9942 - loss: 0.7826 - val_accuracy: 0.9974 - val_loss: 0.7750 - learning_rate: 5.0000e-04\n","Epoch 60/150\n","\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 631ms/step - accuracy: 0.9994 - loss: 0.7655\n","Epoch 60: val_loss improved from 0.77502 to 0.76808, saving model to /content/drive/MyDrive/best_cauliflower_model.h5\n"]},{"output_type":"stream","name":"stderr","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"]},{"output_type":"stream","name":"stdout","text":["\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m19s\u001b[0m 808ms/step - accuracy: 0.9993 - loss: 0.7656 - val_accuracy: 0.9961 - val_loss: 0.7681 - learning_rate: 5.0000e-04\n","Epoch 61/150\n","\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 632ms/step - accuracy: 0.9966 - loss: 0.7650\n","Epoch 61: val_loss improved from 0.76808 to 0.75806, saving model to /content/drive/MyDrive/best_cauliflower_model.h5\n"]},{"output_type":"stream","name":"stderr","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"]},{"output_type":"stream","name":"stdout","text":["\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m19s\u001b[0m 810ms/step - accuracy: 0.9966 - loss: 0.7648 - val_accuracy: 0.9987 - val_loss: 0.7581 - learning_rate: 5.0000e-04\n","Epoch 62/150\n","\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 634ms/step - accuracy: 0.9956 - loss: 0.7558\n","Epoch 62: val_loss improved from 0.75806 to 0.75400, saving model to /content/drive/MyDrive/best_cauliflower_model.h5\n"]},{"output_type":"stream","name":"stderr","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"]},{"output_type":"stream","name":"stdout","text":["\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m19s\u001b[0m 810ms/step - accuracy: 0.9956 - loss: 0.7558 - val_accuracy: 0.9948 - val_loss: 0.7540 - learning_rate: 5.0000e-04\n","Epoch 63/150\n","\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 632ms/step - accuracy: 0.9963 - loss: 0.7483\n","Epoch 63: val_loss improved from 0.75400 to 0.73819, saving model to /content/drive/MyDrive/best_cauliflower_model.h5\n"]},{"output_type":"stream","name":"stderr","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"]},{"output_type":"stream","name":"stdout","text":["\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m21s\u001b[0m 863ms/step - accuracy: 0.9963 - loss: 0.7484 - val_accuracy: 0.9974 - val_loss: 0.7382 - learning_rate: 5.0000e-04\n","Epoch 64/150\n","\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 633ms/step - accuracy: 0.9942 - loss: 0.7453\n","Epoch 64: val_loss improved from 0.73819 to 0.73649, saving model to /content/drive/MyDrive/best_cauliflower_model.h5\n"]},{"output_type":"stream","name":"stderr","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"]},{"output_type":"stream","name":"stdout","text":["\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m19s\u001b[0m 813ms/step - accuracy: 0.9942 - loss: 0.7451 - val_accuracy: 0.9974 - val_loss: 0.7365 - learning_rate: 5.0000e-04\n","Epoch 65/150\n","\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 629ms/step - accuracy: 0.9985 - loss: 0.7251\n","Epoch 65: val_loss improved from 0.73649 to 0.72561, saving model to /content/drive/MyDrive/best_cauliflower_model.h5\n"]},{"output_type":"stream","name":"stderr","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"]},{"output_type":"stream","name":"stdout","text":["\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m19s\u001b[0m 807ms/step - accuracy: 0.9985 - loss: 0.7251 - val_accuracy: 0.9974 - val_loss: 0.7256 - learning_rate: 5.0000e-04\n","Epoch 66/150\n","\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 634ms/step - accuracy: 0.9967 - loss: 0.7187\n","Epoch 66: val_loss improved from 0.72561 to 0.71785, saving model to /content/drive/MyDrive/best_cauliflower_model.h5\n"]},{"output_type":"stream","name":"stderr","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"]},{"output_type":"stream","name":"stdout","text":["\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m21s\u001b[0m 875ms/step - accuracy: 0.9967 - loss: 0.7187 - val_accuracy: 0.9961 - val_loss: 0.7178 - learning_rate: 5.0000e-04\n","Epoch 67/150\n","\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 632ms/step - accuracy: 0.9955 - loss: 0.7187\n","Epoch 67: val_loss improved from 0.71785 to 0.71320, saving model to /content/drive/MyDrive/best_cauliflower_model.h5\n"]},{"output_type":"stream","name":"stderr","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"]},{"output_type":"stream","name":"stdout","text":["\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m21s\u001b[0m 868ms/step - accuracy: 0.9955 - loss: 0.7186 - val_accuracy: 0.9935 - val_loss: 0.7132 - learning_rate: 5.0000e-04\n","Epoch 68/150\n","\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 635ms/step - accuracy: 0.9977 - loss: 0.7042\n","Epoch 68: val_loss improved from 0.71320 to 0.69843, saving model to /content/drive/MyDrive/best_cauliflower_model.h5\n"]},{"output_type":"stream","name":"stderr","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"]},{"output_type":"stream","name":"stdout","text":["\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m19s\u001b[0m 809ms/step - accuracy: 0.9977 - loss: 0.7041 - val_accuracy: 0.9961 - val_loss: 0.6984 - learning_rate: 5.0000e-04\n","Epoch 69/150\n","\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 633ms/step - accuracy: 0.9964 - loss: 0.6960\n","Epoch 69: val_loss did not improve from 0.69843\n","\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m18s\u001b[0m 737ms/step - accuracy: 0.9964 - loss: 0.6959 - val_accuracy: 0.9935 - val_loss: 0.6994 - learning_rate: 5.0000e-04\n","Epoch 70/150\n","\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 631ms/step - accuracy: 0.9952 - loss: 0.6909\n","Epoch 70: val_loss improved from 0.69843 to 0.68642, saving model to /content/drive/MyDrive/best_cauliflower_model.h5\n"]},{"output_type":"stream","name":"stderr","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"]},{"output_type":"stream","name":"stdout","text":["\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m21s\u001b[0m 874ms/step - accuracy: 0.9952 - loss: 0.6907 - val_accuracy: 0.9935 - val_loss: 0.6864 - learning_rate: 5.0000e-04\n","Epoch 71/150\n","\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 635ms/step - accuracy: 0.9979 - loss: 0.6760\n","Epoch 71: val_loss improved from 0.68642 to 0.67345, saving model to /content/drive/MyDrive/best_cauliflower_model.h5\n"]},{"output_type":"stream","name":"stderr","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"]},{"output_type":"stream","name":"stdout","text":["\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m19s\u001b[0m 815ms/step - accuracy: 0.9978 - loss: 0.6760 - val_accuracy: 0.9974 - val_loss: 0.6734 - learning_rate: 5.0000e-04\n","Epoch 72/150\n","\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 646ms/step - accuracy: 0.9981 - loss: 0.6671\n","Epoch 72: val_loss improved from 0.67345 to 0.66849, saving model to /content/drive/MyDrive/best_cauliflower_model.h5\n"]},{"output_type":"stream","name":"stderr","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"]},{"output_type":"stream","name":"stdout","text":["\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m20s\u001b[0m 822ms/step - accuracy: 0.9980 - loss: 0.6670 - val_accuracy: 0.9948 - val_loss: 0.6685 - learning_rate: 5.0000e-04\n","Epoch 73/150\n","\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 635ms/step - accuracy: 0.9989 - loss: 0.6544\n","Epoch 73: val_loss did not improve from 0.66849\n","\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m18s\u001b[0m 737ms/step - accuracy: 0.9989 - loss: 0.6543 - val_accuracy: 0.9923 - val_loss: 0.6758 - learning_rate: 5.0000e-04\n","Epoch 74/150\n","\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 627ms/step - accuracy: 0.9976 - loss: 0.6519\n","Epoch 74: val_loss improved from 0.66849 to 0.65017, saving model to /content/drive/MyDrive/best_cauliflower_model.h5\n"]},{"output_type":"stream","name":"stderr","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"]},{"output_type":"stream","name":"stdout","text":["\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m21s\u001b[0m 872ms/step - accuracy: 0.9977 - loss: 0.6518 - val_accuracy: 0.9961 - val_loss: 0.6502 - learning_rate: 5.0000e-04\n","Epoch 75/150\n","\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 644ms/step - accuracy: 0.9959 - loss: 0.6465\n","Epoch 75: val_loss did not improve from 0.65017\n","\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m18s\u001b[0m 747ms/step - accuracy: 0.9960 - loss: 0.6463 - val_accuracy: 0.9910 - val_loss: 0.6533 - learning_rate: 5.0000e-04\n","Epoch 76/150\n","\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 627ms/step - accuracy: 0.9986 - loss: 0.6287\n","Epoch 76: val_loss did not improve from 0.65017\n","\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m17s\u001b[0m 730ms/step - accuracy: 0.9986 - loss: 0.6288 - val_accuracy: 0.9884 - val_loss: 0.6547 - learning_rate: 5.0000e-04\n","Epoch 77/150\n","\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 627ms/step - accuracy: 0.9963 - loss: 0.6290\n","Epoch 77: val_loss improved from 0.65017 to 0.62182, saving model to /content/drive/MyDrive/best_cauliflower_model.h5\n"]},{"output_type":"stream","name":"stderr","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"]},{"output_type":"stream","name":"stdout","text":["\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m21s\u001b[0m 874ms/step - accuracy: 0.9963 - loss: 0.6288 - val_accuracy: 0.9948 - val_loss: 0.6218 - learning_rate: 5.0000e-04\n","Epoch 78/150\n","\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 634ms/step - accuracy: 0.9979 - loss: 0.6143\n","Epoch 78: val_loss improved from 0.62182 to 0.61170, saving model to /content/drive/MyDrive/best_cauliflower_model.h5\n"]},{"output_type":"stream","name":"stderr","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"]},{"output_type":"stream","name":"stdout","text":["\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m19s\u001b[0m 812ms/step - accuracy: 0.9979 - loss: 0.6143 - val_accuracy: 0.9961 - val_loss: 0.6117 - learning_rate: 5.0000e-04\n","Epoch 79/150\n","\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 634ms/step - accuracy: 0.9994 - loss: 0.6036\n","Epoch 79: val_loss improved from 0.61170 to 0.60763, saving model to /content/drive/MyDrive/best_cauliflower_model.h5\n"]},{"output_type":"stream","name":"stderr","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"]},{"output_type":"stream","name":"stdout","text":["\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m21s\u001b[0m 868ms/step - accuracy: 0.9994 - loss: 0.6035 - val_accuracy: 0.9974 - val_loss: 0.6076 - learning_rate: 5.0000e-04\n","Epoch 80/150\n","\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 635ms/step - accuracy: 0.9962 - loss: 0.6007\n","Epoch 80: val_loss improved from 0.60763 to 0.59148, saving model to /content/drive/MyDrive/best_cauliflower_model.h5\n"]},{"output_type":"stream","name":"stderr","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"]},{"output_type":"stream","name":"stdout","text":["\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m21s\u001b[0m 873ms/step - accuracy: 0.9962 - loss: 0.6007 - val_accuracy: 0.9974 - val_loss: 0.5915 - learning_rate: 5.0000e-04\n","Epoch 81/150\n","\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 635ms/step - accuracy: 0.9978 - loss: 0.5901\n","Epoch 81: val_loss did not improve from 0.59148\n","\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m18s\u001b[0m 738ms/step - accuracy: 0.9977 - loss: 0.5901 - val_accuracy: 0.9935 - val_loss: 0.6017 - learning_rate: 5.0000e-04\n","Epoch 82/150\n","\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 630ms/step - accuracy: 0.9986 - loss: 0.5793\n","Epoch 82: val_loss improved from 0.59148 to 0.57572, saving model to /content/drive/MyDrive/best_cauliflower_model.h5\n"]},{"output_type":"stream","name":"stderr","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"]},{"output_type":"stream","name":"stdout","text":["\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m21s\u001b[0m 876ms/step - accuracy: 0.9985 - loss: 0.5794 - val_accuracy: 0.9961 - val_loss: 0.5757 - learning_rate: 5.0000e-04\n","Epoch 83/150\n","\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 628ms/step - accuracy: 0.9989 - loss: 0.5695\n","Epoch 83: val_loss did not improve from 0.57572\n","\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m18s\u001b[0m 731ms/step - accuracy: 0.9989 - loss: 0.5695 - val_accuracy: 0.9948 - val_loss: 0.5809 - learning_rate: 5.0000e-04\n","Epoch 84/150\n","\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 628ms/step - accuracy: 0.9963 - loss: 0.5645\n","Epoch 84: val_loss improved from 0.57572 to 0.57188, saving model to /content/drive/MyDrive/best_cauliflower_model.h5\n"]},{"output_type":"stream","name":"stderr","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"]},{"output_type":"stream","name":"stdout","text":["\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m21s\u001b[0m 904ms/step - accuracy: 0.9963 - loss: 0.5644 - val_accuracy: 0.9923 - val_loss: 0.5719 - learning_rate: 5.0000e-04\n","Epoch 85/150\n","\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 629ms/step - accuracy: 0.9977 - loss: 0.5552\n","Epoch 85: val_loss improved from 0.57188 to 0.55612, saving model to /content/drive/MyDrive/best_cauliflower_model.h5\n"]},{"output_type":"stream","name":"stderr","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"]},{"output_type":"stream","name":"stdout","text":["\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m21s\u001b[0m 862ms/step - accuracy: 0.9976 - loss: 0.5552 - val_accuracy: 0.9948 - val_loss: 0.5561 - learning_rate: 5.0000e-04\n","Epoch 86/150\n","\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 637ms/step - accuracy: 0.9988 - loss: 0.5453\n","Epoch 86: val_loss improved from 0.55612 to 0.54696, saving model to /content/drive/MyDrive/best_cauliflower_model.h5\n"]},{"output_type":"stream","name":"stderr","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"]},{"output_type":"stream","name":"stdout","text":["\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m21s\u001b[0m 875ms/step - accuracy: 0.9988 - loss: 0.5453 - val_accuracy: 0.9948 - val_loss: 0.5470 - learning_rate: 5.0000e-04\n","Epoch 87/150\n","\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 639ms/step - accuracy: 0.9986 - loss: 0.5359\n","Epoch 87: val_loss improved from 0.54696 to 0.53636, saving model to /content/drive/MyDrive/best_cauliflower_model.h5\n"]},{"output_type":"stream","name":"stderr","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"]},{"output_type":"stream","name":"stdout","text":["\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m21s\u001b[0m 881ms/step - accuracy: 0.9986 - loss: 0.5358 - val_accuracy: 0.9961 - val_loss: 0.5364 - learning_rate: 5.0000e-04\n","Epoch 88/150\n","\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 647ms/step - accuracy: 0.9985 - loss: 0.5276\n","Epoch 88: val_loss improved from 0.53636 to 0.52795, saving model to /content/drive/MyDrive/best_cauliflower_model.h5\n"]},{"output_type":"stream","name":"stderr","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"]},{"output_type":"stream","name":"stdout","text":["\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m21s\u001b[0m 888ms/step - accuracy: 0.9985 - loss: 0.5275 - val_accuracy: 0.9974 - val_loss: 0.5279 - learning_rate: 5.0000e-04\n","Epoch 89/150\n","\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 644ms/step - accuracy: 0.9999 - loss: 0.5163\n","Epoch 89: val_loss improved from 0.52795 to 0.52327, saving model to /content/drive/MyDrive/best_cauliflower_model.h5\n"]},{"output_type":"stream","name":"stderr","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"]},{"output_type":"stream","name":"stdout","text":["\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m21s\u001b[0m 889ms/step - accuracy: 0.9999 - loss: 0.5162 - val_accuracy: 0.9961 - val_loss: 0.5233 - learning_rate: 5.0000e-04\n","Epoch 90/150\n","\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 646ms/step - accuracy: 0.9960 - loss: 0.5148\n","Epoch 90: val_loss improved from 0.52327 to 0.51487, saving model to /content/drive/MyDrive/best_cauliflower_model.h5\n"]},{"output_type":"stream","name":"stderr","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"]},{"output_type":"stream","name":"stdout","text":["\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m20s\u001b[0m 826ms/step - accuracy: 0.9961 - loss: 0.5147 - val_accuracy: 0.9961 - val_loss: 0.5149 - learning_rate: 5.0000e-04\n","Epoch 91/150\n","\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 637ms/step - accuracy: 0.9991 - loss: 0.5001\n","Epoch 91: val_loss improved from 0.51487 to 0.51086, saving model to /content/drive/MyDrive/best_cauliflower_model.h5\n"]},{"output_type":"stream","name":"stderr","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"]},{"output_type":"stream","name":"stdout","text":["\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m20s\u001b[0m 819ms/step - accuracy: 0.9991 - loss: 0.5000 - val_accuracy: 0.9935 - val_loss: 0.5109 - learning_rate: 5.0000e-04\n","Epoch 92/150\n","\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 642ms/step - accuracy: 0.9982 - loss: 0.4945\n","Epoch 92: val_loss improved from 0.51086 to 0.50982, saving model to /content/drive/MyDrive/best_cauliflower_model.h5\n"]},{"output_type":"stream","name":"stderr","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"]},{"output_type":"stream","name":"stdout","text":["\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m21s\u001b[0m 882ms/step - accuracy: 0.9982 - loss: 0.4945 - val_accuracy: 0.9923 - val_loss: 0.5098 - learning_rate: 5.0000e-04\n","Epoch 93/150\n","\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 641ms/step - accuracy: 0.9996 - loss: 0.4865\n","Epoch 93: val_loss improved from 0.50982 to 0.50296, saving model to /content/drive/MyDrive/best_cauliflower_model.h5\n"]},{"output_type":"stream","name":"stderr","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"]},{"output_type":"stream","name":"stdout","text":["\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m20s\u001b[0m 825ms/step - accuracy: 0.9996 - loss: 0.4864 - val_accuracy: 0.9910 - val_loss: 0.5030 - learning_rate: 5.0000e-04\n","Epoch 94/150\n","\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 638ms/step - accuracy: 1.0000 - loss: 0.4760\n","Epoch 94: val_loss improved from 0.50296 to 0.49139, saving model to /content/drive/MyDrive/best_cauliflower_model.h5\n"]},{"output_type":"stream","name":"stderr","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"]},{"output_type":"stream","name":"stdout","text":["\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m21s\u001b[0m 887ms/step - accuracy: 1.0000 - loss: 0.4759 - val_accuracy: 0.9935 - val_loss: 0.4914 - learning_rate: 5.0000e-04\n","Epoch 95/150\n","\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 637ms/step - accuracy: 0.9977 - loss: 0.4693\n","Epoch 95: val_loss improved from 0.49139 to 0.48243, saving model to /content/drive/MyDrive/best_cauliflower_model.h5\n"]},{"output_type":"stream","name":"stderr","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"]},{"output_type":"stream","name":"stdout","text":["\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m21s\u001b[0m 880ms/step - accuracy: 0.9977 - loss: 0.4692 - val_accuracy: 0.9935 - val_loss: 0.4824 - learning_rate: 5.0000e-04\n","Epoch 96/150\n","\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 638ms/step - accuracy: 0.9981 - loss: 0.4634\n","Epoch 96: val_loss improved from 0.48243 to 0.47229, saving model to /content/drive/MyDrive/best_cauliflower_model.h5\n"]},{"output_type":"stream","name":"stderr","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"]},{"output_type":"stream","name":"stdout","text":["\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m21s\u001b[0m 891ms/step - accuracy: 0.9981 - loss: 0.4634 - val_accuracy: 0.9935 - val_loss: 0.4723 - learning_rate: 5.0000e-04\n","Epoch 97/150\n","\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 641ms/step - accuracy: 0.9990 - loss: 0.4529\n","Epoch 97: val_loss improved from 0.47229 to 0.45122, saving model to /content/drive/MyDrive/best_cauliflower_model.h5\n"]},{"output_type":"stream","name":"stderr","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"]},{"output_type":"stream","name":"stdout","text":["\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m21s\u001b[0m 880ms/step - accuracy: 0.9990 - loss: 0.4528 - val_accuracy: 0.9948 - val_loss: 0.4512 - learning_rate: 5.0000e-04\n","Epoch 98/150\n","\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 642ms/step - accuracy: 0.9975 - loss: 0.4485\n","Epoch 98: val_loss did not improve from 0.45122\n","\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m18s\u001b[0m 745ms/step - accuracy: 0.9975 - loss: 0.4484 - val_accuracy: 0.9910 - val_loss: 0.4778 - learning_rate: 5.0000e-04\n","Epoch 99/150\n","\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 636ms/step - accuracy: 0.9984 - loss: 0.4402\n","Epoch 99: val_loss did not improve from 0.45122\n","\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m18s\u001b[0m 742ms/step - accuracy: 0.9984 - loss: 0.4401 - val_accuracy: 0.9910 - val_loss: 0.4646 - learning_rate: 5.0000e-04\n","Epoch 100/150\n","\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 636ms/step - accuracy: 0.9962 - loss: 0.4374\n","Epoch 100: val_loss did not improve from 0.45122\n","\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m18s\u001b[0m 741ms/step - accuracy: 0.9962 - loss: 0.4371 - val_accuracy: 0.9910 - val_loss: 0.4657 - learning_rate: 5.0000e-04\n","Epoch 101/150\n","\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 633ms/step - accuracy: 0.9992 - loss: 0.4205\n","Epoch 101: val_loss improved from 0.45122 to 0.43177, saving model to /content/drive/MyDrive/best_cauliflower_model.h5\n"]},{"output_type":"stream","name":"stderr","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"]},{"output_type":"stream","name":"stdout","text":["\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m21s\u001b[0m 877ms/step - accuracy: 0.9992 - loss: 0.4205 - val_accuracy: 0.9923 - val_loss: 0.4318 - learning_rate: 5.0000e-04\n","Epoch 102/150\n","\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 638ms/step - accuracy: 0.9969 - loss: 0.4167\n","Epoch 102: val_loss did not improve from 0.43177\n","\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m18s\u001b[0m 741ms/step - accuracy: 0.9970 - loss: 0.4165 - val_accuracy: 0.9910 - val_loss: 0.4391 - learning_rate: 5.0000e-04\n","Epoch 103/150\n","\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 637ms/step - accuracy: 0.9992 - loss: 0.4052\n","Epoch 103: val_loss improved from 0.43177 to 0.40880, saving model to /content/drive/MyDrive/best_cauliflower_model.h5\n"]},{"output_type":"stream","name":"stderr","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"]},{"output_type":"stream","name":"stdout","text":["\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m21s\u001b[0m 889ms/step - accuracy: 0.9992 - loss: 0.4052 - val_accuracy: 0.9948 - val_loss: 0.4088 - learning_rate: 5.0000e-04\n","Epoch 104/150\n","\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 635ms/step - accuracy: 0.9970 - loss: 0.4042\n","Epoch 104: val_loss did not improve from 0.40880\n","\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m18s\u001b[0m 739ms/step - accuracy: 0.9970 - loss: 0.4042 - val_accuracy: 0.9923 - val_loss: 0.4174 - learning_rate: 5.0000e-04\n","Epoch 105/150\n","\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 637ms/step - accuracy: 0.9992 - loss: 0.3907\n","Epoch 105: val_loss improved from 0.40880 to 0.39847, saving model to /content/drive/MyDrive/best_cauliflower_model.h5\n"]},{"output_type":"stream","name":"stderr","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"]},{"output_type":"stream","name":"stdout","text":["\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m21s\u001b[0m 885ms/step - accuracy: 0.9992 - loss: 0.3906 - val_accuracy: 0.9935 - val_loss: 0.3985 - learning_rate: 5.0000e-04\n","Epoch 106/150\n","\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 639ms/step - accuracy: 0.9982 - loss: 0.3843\n","Epoch 106: val_loss improved from 0.39847 to 0.39325, saving model to /content/drive/MyDrive/best_cauliflower_model.h5\n"]},{"output_type":"stream","name":"stderr","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"]},{"output_type":"stream","name":"stdout","text":["\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m20s\u001b[0m 823ms/step - accuracy: 0.9982 - loss: 0.3843 - val_accuracy: 0.9948 - val_loss: 0.3933 - learning_rate: 5.0000e-04\n","Epoch 107/150\n","\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 641ms/step - accuracy: 0.9968 - loss: 0.3804\n","Epoch 107: val_loss improved from 0.39325 to 0.38860, saving model to /content/drive/MyDrive/best_cauliflower_model.h5\n"]},{"output_type":"stream","name":"stderr","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"]},{"output_type":"stream","name":"stdout","text":["\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m20s\u001b[0m 818ms/step - accuracy: 0.9968 - loss: 0.3803 - val_accuracy: 0.9948 - val_loss: 0.3886 - learning_rate: 5.0000e-04\n","Epoch 108/150\n","\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 636ms/step - accuracy: 0.9992 - loss: 0.3710\n","Epoch 108: val_loss improved from 0.38860 to 0.37875, saving model to /content/drive/MyDrive/best_cauliflower_model.h5\n"]},{"output_type":"stream","name":"stderr","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"]},{"output_type":"stream","name":"stdout","text":["\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m21s\u001b[0m 884ms/step - accuracy: 0.9991 - loss: 0.3709 - val_accuracy: 0.9948 - val_loss: 0.3788 - learning_rate: 5.0000e-04\n","Epoch 109/150\n","\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 647ms/step - accuracy: 0.9990 - loss: 0.3621\n","Epoch 109: val_loss improved from 0.37875 to 0.37353, saving model to /content/drive/MyDrive/best_cauliflower_model.h5\n"]},{"output_type":"stream","name":"stderr","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"]},{"output_type":"stream","name":"stdout","text":["\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m21s\u001b[0m 891ms/step - accuracy: 0.9990 - loss: 0.3621 - val_accuracy: 0.9961 - val_loss: 0.3735 - learning_rate: 5.0000e-04\n","Epoch 110/150\n","\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 636ms/step - accuracy: 0.9996 - loss: 0.3536\n","Epoch 110: val_loss did not improve from 0.37353\n","\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m18s\u001b[0m 743ms/step - accuracy: 0.9995 - loss: 0.3536 - val_accuracy: 0.9948 - val_loss: 0.3752 - learning_rate: 5.0000e-04\n","Epoch 111/150\n","\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 643ms/step - accuracy: 0.9975 - loss: 0.3509\n","Epoch 111: val_loss improved from 0.37353 to 0.36717, saving model to /content/drive/MyDrive/best_cauliflower_model.h5\n"]},{"output_type":"stream","name":"stderr","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"]},{"output_type":"stream","name":"stdout","text":["\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m21s\u001b[0m 886ms/step - accuracy: 0.9976 - loss: 0.3508 - val_accuracy: 0.9935 - val_loss: 0.3672 - learning_rate: 5.0000e-04\n","Epoch 112/150\n","\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 639ms/step - accuracy: 0.9983 - loss: 0.3423\n","Epoch 112: val_loss improved from 0.36717 to 0.35028, saving model to /content/drive/MyDrive/best_cauliflower_model.h5\n"]},{"output_type":"stream","name":"stderr","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"]},{"output_type":"stream","name":"stdout","text":["\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m21s\u001b[0m 876ms/step - accuracy: 0.9983 - loss: 0.3422 - val_accuracy: 0.9935 - val_loss: 0.3503 - learning_rate: 5.0000e-04\n","Epoch 113/150\n","\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 641ms/step - accuracy: 0.9995 - loss: 0.3337\n","Epoch 113: val_loss did not improve from 0.35028\n","\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m18s\u001b[0m 744ms/step - accuracy: 0.9994 - loss: 0.3338 - val_accuracy: 0.9948 - val_loss: 0.3555 - learning_rate: 5.0000e-04\n","Epoch 114/150\n","\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 639ms/step - accuracy: 0.9986 - loss: 0.3371\n","Epoch 114: val_loss improved from 0.35028 to 0.33313, saving model to /content/drive/MyDrive/best_cauliflower_model.h5\n"]},{"output_type":"stream","name":"stderr","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"]},{"output_type":"stream","name":"stdout","text":["\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m21s\u001b[0m 902ms/step - accuracy: 0.9986 - loss: 0.3368 - val_accuracy: 0.9948 - val_loss: 0.3331 - learning_rate: 5.0000e-04\n","Epoch 115/150\n","\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 643ms/step - accuracy: 0.9991 - loss: 0.3209\n","Epoch 115: val_loss improved from 0.33313 to 0.32908, saving model to /content/drive/MyDrive/best_cauliflower_model.h5\n"]},{"output_type":"stream","name":"stderr","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"]},{"output_type":"stream","name":"stdout","text":["\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m20s\u001b[0m 824ms/step - accuracy: 0.9991 - loss: 0.3209 - val_accuracy: 0.9961 - val_loss: 0.3291 - learning_rate: 5.0000e-04\n","Epoch 116/150\n","\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 638ms/step - accuracy: 0.9978 - loss: 0.3147\n","Epoch 116: val_loss improved from 0.32908 to 0.32053, saving model to /content/drive/MyDrive/best_cauliflower_model.h5\n"]},{"output_type":"stream","name":"stderr","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"]},{"output_type":"stream","name":"stdout","text":["\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m21s\u001b[0m 892ms/step - accuracy: 0.9978 - loss: 0.3146 - val_accuracy: 0.9961 - val_loss: 0.3205 - learning_rate: 5.0000e-04\n","Epoch 117/150\n","\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 636ms/step - accuracy: 0.9975 - loss: 0.3117\n","Epoch 117: val_loss improved from 0.32053 to 0.31414, saving model to /content/drive/MyDrive/best_cauliflower_model.h5\n"]},{"output_type":"stream","name":"stderr","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"]},{"output_type":"stream","name":"stdout","text":["\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m20s\u001b[0m 820ms/step - accuracy: 0.9975 - loss: 0.3116 - val_accuracy: 0.9961 - val_loss: 0.3141 - learning_rate: 5.0000e-04\n","Epoch 118/150\n","\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 646ms/step - accuracy: 0.9993 - loss: 0.3011\n","Epoch 118: val_loss improved from 0.31414 to 0.30237, saving model to /content/drive/MyDrive/best_cauliflower_model.h5\n"]},{"output_type":"stream","name":"stderr","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"]},{"output_type":"stream","name":"stdout","text":["\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m21s\u001b[0m 895ms/step - accuracy: 0.9992 - loss: 0.3011 - val_accuracy: 0.9974 - val_loss: 0.3024 - learning_rate: 5.0000e-04\n","Epoch 119/150\n","\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 645ms/step - accuracy: 0.9993 - loss: 0.2954\n","Epoch 119: val_loss improved from 0.30237 to 0.29839, saving model to /content/drive/MyDrive/best_cauliflower_model.h5\n"]},{"output_type":"stream","name":"stderr","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"]},{"output_type":"stream","name":"stdout","text":["\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m21s\u001b[0m 882ms/step - accuracy: 0.9993 - loss: 0.2954 - val_accuracy: 0.9961 - val_loss: 0.2984 - learning_rate: 5.0000e-04\n","Epoch 120/150\n","\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 644ms/step - accuracy: 0.9990 - loss: 0.2896\n","Epoch 120: val_loss improved from 0.29839 to 0.28987, saving model to /content/drive/MyDrive/best_cauliflower_model.h5\n"]},{"output_type":"stream","name":"stderr","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"]},{"output_type":"stream","name":"stdout","text":["\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m21s\u001b[0m 894ms/step - accuracy: 0.9990 - loss: 0.2896 - val_accuracy: 0.9961 - val_loss: 0.2899 - learning_rate: 5.0000e-04\n","Epoch 121/150\n","\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 639ms/step - accuracy: 0.9990 - loss: 0.2827\n","Epoch 121: val_loss improved from 0.28987 to 0.28499, saving model to /content/drive/MyDrive/best_cauliflower_model.h5\n"]},{"output_type":"stream","name":"stderr","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"]},{"output_type":"stream","name":"stdout","text":["\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m21s\u001b[0m 886ms/step - accuracy: 0.9990 - loss: 0.2827 - val_accuracy: 0.9961 - val_loss: 0.2850 - learning_rate: 5.0000e-04\n","Epoch 122/150\n","\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 641ms/step - accuracy: 0.9992 - loss: 0.2761\n","Epoch 122: val_loss improved from 0.28499 to 0.28144, saving model to /content/drive/MyDrive/best_cauliflower_model.h5\n"]},{"output_type":"stream","name":"stderr","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"]},{"output_type":"stream","name":"stdout","text":["\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m21s\u001b[0m 880ms/step - accuracy: 0.9992 - loss: 0.2761 - val_accuracy: 0.9961 - val_loss: 0.2814 - learning_rate: 5.0000e-04\n","Epoch 123/150\n","\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 643ms/step - accuracy: 0.9982 - loss: 0.2712\n","Epoch 123: val_loss improved from 0.28144 to 0.27900, saving model to /content/drive/MyDrive/best_cauliflower_model.h5\n"]},{"output_type":"stream","name":"stderr","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"]},{"output_type":"stream","name":"stdout","text":["\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m20s\u001b[0m 826ms/step - accuracy: 0.9982 - loss: 0.2711 - val_accuracy: 0.9948 - val_loss: 0.2790 - learning_rate: 5.0000e-04\n","Epoch 124/150\n","\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 639ms/step - accuracy: 0.9984 - loss: 0.2656\n","Epoch 124: val_loss improved from 0.27900 to 0.26647, saving model to /content/drive/MyDrive/best_cauliflower_model.h5\n"]},{"output_type":"stream","name":"stderr","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"]},{"output_type":"stream","name":"stdout","text":["\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m21s\u001b[0m 884ms/step - accuracy: 0.9984 - loss: 0.2655 - val_accuracy: 0.9961 - val_loss: 0.2665 - learning_rate: 5.0000e-04\n","Epoch 125/150\n","\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 635ms/step - accuracy: 0.9998 - loss: 0.2577\n","Epoch 125: val_loss improved from 0.26647 to 0.25864, saving model to /content/drive/MyDrive/best_cauliflower_model.h5\n"]},{"output_type":"stream","name":"stderr","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"]},{"output_type":"stream","name":"stdout","text":["\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m20s\u001b[0m 817ms/step - accuracy: 0.9998 - loss: 0.2577 - val_accuracy: 0.9961 - val_loss: 0.2586 - learning_rate: 5.0000e-04\n","Epoch 126/150\n","\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 641ms/step - accuracy: 0.9991 - loss: 0.2538\n","Epoch 126: val_loss improved from 0.25864 to 0.25504, saving model to /content/drive/MyDrive/best_cauliflower_model.h5\n"]},{"output_type":"stream","name":"stderr","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"]},{"output_type":"stream","name":"stdout","text":["\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m22s\u001b[0m 917ms/step - accuracy: 0.9990 - loss: 0.2538 - val_accuracy: 0.9974 - val_loss: 0.2550 - learning_rate: 5.0000e-04\n","Epoch 127/150\n","\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 641ms/step - accuracy: 0.9997 - loss: 0.2472\n","Epoch 127: val_loss improved from 0.25504 to 0.25115, saving model to /content/drive/MyDrive/best_cauliflower_model.h5\n"]},{"output_type":"stream","name":"stderr","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"]},{"output_type":"stream","name":"stdout","text":["\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m20s\u001b[0m 821ms/step - accuracy: 0.9997 - loss: 0.2472 - val_accuracy: 0.9961 - val_loss: 0.2512 - learning_rate: 5.0000e-04\n","Epoch 128/150\n","\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 644ms/step - accuracy: 0.9999 - loss: 0.2408\n","Epoch 128: val_loss improved from 0.25115 to 0.24463, saving model to /content/drive/MyDrive/best_cauliflower_model.h5\n"]},{"output_type":"stream","name":"stderr","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"]},{"output_type":"stream","name":"stdout","text":["\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m21s\u001b[0m 883ms/step - accuracy: 0.9999 - loss: 0.2408 - val_accuracy: 0.9987 - val_loss: 0.2446 - learning_rate: 5.0000e-04\n","Epoch 129/150\n","\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 638ms/step - accuracy: 0.9989 - loss: 0.2403\n","Epoch 129: val_loss improved from 0.24463 to 0.24304, saving model to /content/drive/MyDrive/best_cauliflower_model.h5\n"]},{"output_type":"stream","name":"stderr","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"]},{"output_type":"stream","name":"stdout","text":["\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m21s\u001b[0m 883ms/step - accuracy: 0.9989 - loss: 0.2403 - val_accuracy: 0.9974 - val_loss: 0.2430 - learning_rate: 5.0000e-04\n","Epoch 130/150\n","\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 643ms/step - accuracy: 0.9983 - loss: 0.2326\n","Epoch 130: val_loss did not improve from 0.24304\n","\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m18s\u001b[0m 746ms/step - accuracy: 0.9983 - loss: 0.2326 - val_accuracy: 0.9948 - val_loss: 0.2490 - learning_rate: 5.0000e-04\n","Epoch 131/150\n","\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 639ms/step - accuracy: 0.9983 - loss: 0.2274\n","Epoch 131: val_loss improved from 0.24304 to 0.23956, saving model to /content/drive/MyDrive/best_cauliflower_model.h5\n"]},{"output_type":"stream","name":"stderr","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"]},{"output_type":"stream","name":"stdout","text":["\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m21s\u001b[0m 889ms/step - accuracy: 0.9983 - loss: 0.2273 - val_accuracy: 0.9948 - val_loss: 0.2396 - learning_rate: 5.0000e-04\n","Epoch 132/150\n","\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 645ms/step - accuracy: 0.9999 - loss: 0.2200\n","Epoch 132: val_loss improved from 0.23956 to 0.23162, saving model to /content/drive/MyDrive/best_cauliflower_model.h5\n"]},{"output_type":"stream","name":"stderr","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"]},{"output_type":"stream","name":"stdout","text":["\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m21s\u001b[0m 887ms/step - accuracy: 0.9998 - loss: 0.2201 - val_accuracy: 0.9961 - val_loss: 0.2316 - learning_rate: 5.0000e-04\n","Epoch 133/150\n","\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 640ms/step - accuracy: 0.9996 - loss: 0.2153\n","Epoch 133: val_loss improved from 0.23162 to 0.22561, saving model to /content/drive/MyDrive/best_cauliflower_model.h5\n"]},{"output_type":"stream","name":"stderr","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"]},{"output_type":"stream","name":"stdout","text":["\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m20s\u001b[0m 825ms/step - accuracy: 0.9996 - loss: 0.2153 - val_accuracy: 0.9961 - val_loss: 0.2256 - learning_rate: 5.0000e-04\n","Epoch 134/150\n","\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 651ms/step - accuracy: 0.9981 - loss: 0.2159\n","Epoch 134: val_loss improved from 0.22561 to 0.21652, saving model to /content/drive/MyDrive/best_cauliflower_model.h5\n"]},{"output_type":"stream","name":"stderr","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"]},{"output_type":"stream","name":"stdout","text":["\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m21s\u001b[0m 899ms/step - accuracy: 0.9981 - loss: 0.2158 - val_accuracy: 0.9974 - val_loss: 0.2165 - learning_rate: 5.0000e-04\n","Epoch 135/150\n","\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 647ms/step - accuracy: 0.9989 - loss: 0.2073\n","Epoch 135: val_loss did not improve from 0.21652\n","\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m18s\u001b[0m 752ms/step - accuracy: 0.9989 - loss: 0.2073 - val_accuracy: 0.9961 - val_loss: 0.2170 - learning_rate: 5.0000e-04\n","Epoch 136/150\n","\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 631ms/step - accuracy: 0.9991 - loss: 0.2031\n","Epoch 136: val_loss improved from 0.21652 to 0.20744, saving model to /content/drive/MyDrive/best_cauliflower_model.h5\n"]},{"output_type":"stream","name":"stderr","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"]},{"output_type":"stream","name":"stdout","text":["\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m21s\u001b[0m 889ms/step - accuracy: 0.9990 - loss: 0.2031 - val_accuracy: 0.9974 - val_loss: 0.2074 - learning_rate: 5.0000e-04\n","Epoch 137/150\n","\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 635ms/step - accuracy: 0.9980 - loss: 0.1978\n","Epoch 137: val_loss improved from 0.20744 to 0.20264, saving model to /content/drive/MyDrive/best_cauliflower_model.h5\n"]},{"output_type":"stream","name":"stderr","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"]},{"output_type":"stream","name":"stdout","text":["\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m21s\u001b[0m 881ms/step - accuracy: 0.9980 - loss: 0.1978 - val_accuracy: 0.9961 - val_loss: 0.2026 - learning_rate: 5.0000e-04\n","Epoch 138/150\n","\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 639ms/step - accuracy: 0.9992 - loss: 0.1924\n","Epoch 138: val_loss improved from 0.20264 to 0.19639, saving model to /content/drive/MyDrive/best_cauliflower_model.h5\n"]},{"output_type":"stream","name":"stderr","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"]},{"output_type":"stream","name":"stdout","text":["\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m21s\u001b[0m 871ms/step - accuracy: 0.9992 - loss: 0.1924 - val_accuracy: 0.9974 - val_loss: 0.1964 - learning_rate: 5.0000e-04\n","Epoch 139/150\n","\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 647ms/step - accuracy: 0.9993 - loss: 0.1882\n","Epoch 139: val_loss improved from 0.19639 to 0.19403, saving model to /content/drive/MyDrive/best_cauliflower_model.h5\n"]},{"output_type":"stream","name":"stderr","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"]},{"output_type":"stream","name":"stdout","text":["\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m21s\u001b[0m 893ms/step - accuracy: 0.9993 - loss: 0.1882 - val_accuracy: 0.9974 - val_loss: 0.1940 - learning_rate: 5.0000e-04\n","Epoch 140/150\n","\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 641ms/step - accuracy: 0.9987 - loss: 0.1843\n","Epoch 140: val_loss improved from 0.19403 to 0.19324, saving model to /content/drive/MyDrive/best_cauliflower_model.h5\n"]},{"output_type":"stream","name":"stderr","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"]},{"output_type":"stream","name":"stdout","text":["\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m21s\u001b[0m 885ms/step - accuracy: 0.9987 - loss: 0.1843 - val_accuracy: 0.9974 - val_loss: 0.1932 - learning_rate: 5.0000e-04\n","Epoch 141/150\n","\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 643ms/step - accuracy: 0.9998 - loss: 0.1785\n","Epoch 141: val_loss improved from 0.19324 to 0.19079, saving model to /content/drive/MyDrive/best_cauliflower_model.h5\n"]},{"output_type":"stream","name":"stderr","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"]},{"output_type":"stream","name":"stdout","text":["\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m21s\u001b[0m 886ms/step - accuracy: 0.9998 - loss: 0.1785 - val_accuracy: 0.9961 - val_loss: 0.1908 - learning_rate: 5.0000e-04\n","Epoch 142/150\n","\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 640ms/step - accuracy: 0.9987 - loss: 0.1760\n","Epoch 142: val_loss improved from 0.19079 to 0.18464, saving model to /content/drive/MyDrive/best_cauliflower_model.h5\n"]},{"output_type":"stream","name":"stderr","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"]},{"output_type":"stream","name":"stdout","text":["\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m21s\u001b[0m 880ms/step - accuracy: 0.9987 - loss: 0.1759 - val_accuracy: 0.9974 - val_loss: 0.1846 - learning_rate: 5.0000e-04\n","Epoch 143/150\n","\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 645ms/step - accuracy: 0.9993 - loss: 0.1711\n","Epoch 143: val_loss improved from 0.18464 to 0.18112, saving model to /content/drive/MyDrive/best_cauliflower_model.h5\n"]},{"output_type":"stream","name":"stderr","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"]},{"output_type":"stream","name":"stdout","text":["\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m22s\u001b[0m 915ms/step - accuracy: 0.9993 - loss: 0.1711 - val_accuracy: 0.9961 - val_loss: 0.1811 - learning_rate: 5.0000e-04\n","Epoch 144/150\n","\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 643ms/step - accuracy: 0.9989 - loss: 0.1678\n","Epoch 144: val_loss improved from 0.18112 to 0.17375, saving model to /content/drive/MyDrive/best_cauliflower_model.h5\n"]},{"output_type":"stream","name":"stderr","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"]},{"output_type":"stream","name":"stdout","text":["\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m21s\u001b[0m 878ms/step - accuracy: 0.9989 - loss: 0.1678 - val_accuracy: 0.9987 - val_loss: 0.1738 - learning_rate: 5.0000e-04\n","Epoch 145/150\n","\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 645ms/step - accuracy: 0.9996 - loss: 0.1624\n","Epoch 145: val_loss improved from 0.17375 to 0.16940, saving model to /content/drive/MyDrive/best_cauliflower_model.h5\n"]},{"output_type":"stream","name":"stderr","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"]},{"output_type":"stream","name":"stdout","text":["\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m21s\u001b[0m 885ms/step - accuracy: 0.9996 - loss: 0.1624 - val_accuracy: 0.9974 - val_loss: 0.1694 - learning_rate: 5.0000e-04\n","Epoch 146/150\n","\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 643ms/step - accuracy: 0.9996 - loss: 0.1587\n","Epoch 146: val_loss improved from 0.16940 to 0.16890, saving model to /content/drive/MyDrive/best_cauliflower_model.h5\n"]},{"output_type":"stream","name":"stderr","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"]},{"output_type":"stream","name":"stdout","text":["\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m23s\u001b[0m 967ms/step - accuracy: 0.9996 - loss: 0.1587 - val_accuracy: 0.9961 - val_loss: 0.1689 - learning_rate: 5.0000e-04\n","Epoch 147/150\n","\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 640ms/step - accuracy: 0.9996 - loss: 0.1542\n","Epoch 147: val_loss improved from 0.16890 to 0.16480, saving model to /content/drive/MyDrive/best_cauliflower_model.h5\n"]},{"output_type":"stream","name":"stderr","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"]},{"output_type":"stream","name":"stdout","text":["\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m21s\u001b[0m 878ms/step - accuracy: 0.9996 - loss: 0.1542 - val_accuracy: 0.9961 - val_loss: 0.1648 - learning_rate: 5.0000e-04\n","Epoch 148/150\n","\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 640ms/step - accuracy: 0.9998 - loss: 0.1502\n","Epoch 148: val_loss improved from 0.16480 to 0.15907, saving model to /content/drive/MyDrive/best_cauliflower_model.h5\n"]},{"output_type":"stream","name":"stderr","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"]},{"output_type":"stream","name":"stdout","text":["\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m21s\u001b[0m 882ms/step - accuracy: 0.9998 - loss: 0.1502 - val_accuracy: 0.9961 - val_loss: 0.1591 - learning_rate: 5.0000e-04\n","Epoch 149/150\n","\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 649ms/step - accuracy: 0.9995 - loss: 0.1479\n","Epoch 149: val_loss did not improve from 0.15907\n","\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m18s\u001b[0m 755ms/step - accuracy: 0.9995 - loss: 0.1479 - val_accuracy: 0.9961 - val_loss: 0.1611 - learning_rate: 5.0000e-04\n","Epoch 150/150\n","\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 631ms/step - accuracy: 0.9989 - loss: 0.1454\n","Epoch 150: val_loss improved from 0.15907 to 0.15092, saving model to /content/drive/MyDrive/best_cauliflower_model.h5\n"]},{"output_type":"stream","name":"stderr","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"]},{"output_type":"stream","name":"stdout","text":["\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r\u001b[1m24/24\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m21s\u001b[0m 882ms/step - accuracy: 0.9989 - loss: 0.1455 - val_accuracy: 0.9974 - val_loss: 0.1509 - learning_rate: 5.0000e-04\n","Restoring model weights from the end of the best epoch: 150.\n","\n","✓ Training completed!\n"]}],"source":["print('STARTING TRAINING')\n","print('=' * 80)\n","print(f'Epochs: {epochs}')\n","print(f'Batch size: {batch_size}')\n","print(f'Steps per epoch: {train_steps}')\n","print(f'Validation steps: {valid_steps}')\n","print('=' * 80)\n","\n","history = model.fit(\n"," x=train_ds,\n"," epochs=epochs,\n"," verbose=1,\n"," callbacks=callbacks,\n"," validation_data=valid_ds,\n"," steps_per_epoch=train_steps,\n"," validation_steps=valid_steps,\n"," shuffle=False,\n"," initial_epoch=0\n",")\n","\n","print('\\n✓ Training completed!')"]},{"cell_type":"markdown","metadata":{"id":"QMzEKeHlR-jd"},"source":["\n","#
Training&Validation Test"]},{"cell_type":"code","execution_count":18,"metadata":{"colab":{"base_uri":"https://localhost:8080/"},"id":"byrTWBwQ3JP3","executionInfo":{"status":"ok","timestamp":1764242599688,"user_tz":-420,"elapsed":36458,"user":{"displayName":"Ping Ping","userId":"16193029904393259603"}},"outputId":"df40c01b-8b28-459c-c29e-4cc827775b08"},"outputs":[{"output_type":"stream","name":"stdout","text":["\n","================================================================================\n","POST-TRAINING ANALYSIS\n","================================================================================\n","\n","Final Training - Loss: 0.1462, Accuracy: 0.9983\n","Final Validation - Loss: 0.1509, Accuracy: 0.9974\n","\n","Best Epoch: 150\n"," Validation Loss: 0.1509\n"," Validation Accuracy: 0.9974\n","\n","🔍 Quick Prediction Check (first 10 validation samples):\n","\u001b[1m1/1\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m36s\u001b[0m 36s/step\n","Predicted classes: ['Healthy', 'Black Rot', 'Alternaria Leaf Spot', 'Alternaria Leaf Spot', 'Healthy', 'Downy Mildew', 'Bacterial Soft Rot', 'Black Rot', 'Alternaria Leaf Spot', 'Downy Mildew']\n","✅ GOOD: Model is predicting MULTIPLE classes (not just Healthy)!\n","================================================================================\n"]}],"source":["print('\\n' + '=' * 80)\n","print('POST-TRAINING ANALYSIS')\n","print('=' * 80)\n","\n","# Check final metrics\n","final_train_loss = history.history['loss'][-1]\n","final_train_acc = history.history['accuracy'][-1]\n","final_val_loss = history.history['val_loss'][-1]\n","final_val_acc = history.history['val_accuracy'][-1]\n","\n","print(f'\\nFinal Training - Loss: {final_train_loss:.4f}, Accuracy: {final_train_acc:.4f}')\n","print(f'Final Validation - Loss: {final_val_loss:.4f}, Accuracy: {final_val_acc:.4f}')\n","\n","# Find best epoch\n","best_epoch = np.argmin(history.history['val_loss']) + 1\n","best_val_loss = np.min(history.history['val_loss'])\n","best_val_acc = history.history['val_accuracy'][best_epoch - 1]\n","\n","print(f'\\nBest Epoch: {best_epoch}')\n","print(f' Validation Loss: {best_val_loss:.4f}')\n","print(f' Validation Accuracy: {best_val_acc:.4f}')\n","\n","# Quick sanity check on a few predictions\n","print('\\n🔍 Quick Prediction Check (first 10 validation samples):')\n","val_batch, val_labels_batch = next(iter(valid_ds))\n","val_predictions = model.predict(val_batch[:10])\n","val_pred_classes = np.argmax(val_predictions, axis=1)\n","val_class_names = [classes[i] for i in val_pred_classes]\n","print('Predicted classes:', val_class_names)\n","\n","unique_preds = np.unique(val_pred_classes)\n","if len(unique_preds) > 1:\n"," print('✅ GOOD: Model is predicting MULTIPLE classes (not just Healthy)!')\n","else:\n"," print('⚠️ WARNING: Model still predicting only one class. Check training logs.')\n","\n","print('=' * 80)"]},{"cell_type":"markdown","metadata":{"id":"8Apjh9d9FPTf"},"source":["\n","#
Define a function to plot the training data"]},{"cell_type":"code","execution_count":19,"metadata":{"colab":{"base_uri":"https://localhost:8080/","height":394},"id":"DogpFfa7FPTf","executionInfo":{"status":"ok","timestamp":1764242600210,"user_tz":-420,"elapsed":496,"user":{"displayName":"Ping Ping","userId":"16193029904393259603"}},"outputId":"30cd7c00-3d03-4135-8294-7032b8ef3470"},"outputs":[{"output_type":"display_data","data":{"text/plain":[""],"image/png":"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\n"},"metadata":{}}],"source":["def tr_plot(tr_data, start_epoch):\n"," #Plot the training and validation data\n"," tacc=tr_data.history['accuracy']\n"," tloss=tr_data.history['loss']\n"," vacc=tr_data.history['val_accuracy']\n"," vloss=tr_data.history['val_loss']\n"," Epoch_count=len(tacc)+ start_epoch\n"," Epochs=[]\n"," for i in range (start_epoch ,Epoch_count):\n"," Epochs.append(i+1)\n"," index_loss=np.argmin(vloss)# this is the epoch with the lowest validation loss\n"," val_lowest=vloss[index_loss]\n"," index_acc=np.argmax(vacc)\n"," acc_highest=vacc[index_acc]\n"," plt.style.use('fivethirtyeight')\n"," sc_label='best epoch= '+ str(index_loss+1 +start_epoch)\n"," vc_label='best epoch= '+ str(index_acc + 1+ start_epoch)\n"," fig,axes=plt.subplots(nrows=1, ncols=2, figsize=(20,8))\n"," axes[0].plot(Epochs,tloss, 'r', label='Training loss')\n"," axes[0].plot(Epochs,vloss,'g',label='Validation loss' )\n"," axes[0].scatter(index_loss+1 +start_epoch,val_lowest, s=150, c= 'blue', label=sc_label)\n"," axes[0].set_title('Training and Validation Loss')\n"," axes[0].set_xlabel('Epochs')\n"," axes[0].set_ylabel('Loss')\n"," axes[0].legend()\n"," axes[1].plot (Epochs,tacc,'r',label= 'Training Accuracy')\n"," axes[1].plot (Epochs,vacc,'g',label= 'Validation Accuracy')\n"," axes[1].scatter(index_acc+1 +start_epoch,acc_highest, s=150, c= 'blue', label=vc_label)\n"," axes[1].set_title('Training and Validation Accuracy')\n"," axes[1].set_xlabel('Epochs')\n"," axes[1].set_ylabel('Accuracy')\n"," axes[1].legend()\n"," plt.tight_layout\n"," plt.show()\n","\n","tr_plot(history,0)"]},{"cell_type":"markdown","metadata":{"id":"5FxDma3xFPTf"},"source":["\n","#
Make Predictions on the test set\n","### Define a function which takes in a test generator and an integer test_steps\n","### and generates predictions on the test set including a confusion matric\n","### and a classification report"]},{"cell_type":"code","execution_count":20,"metadata":{"colab":{"base_uri":"https://localhost:8080/","height":996},"id":"KrxL9EZjFPTf","executionInfo":{"status":"ok","timestamp":1764242650692,"user_tz":-420,"elapsed":50479,"user":{"displayName":"Ping Ping","userId":"16193029904393259603"}},"outputId":"641372fc-a013-428f-d229-c0ceace07572"},"outputs":[{"output_type":"stream","name":"stdout","text":["\u001b[1m7/7\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m50s\u001b[0m 6s/step\n","file /content/cauliflower_training/cali/Alternaria Leaf Spot/alternaria_leaf_spot_0064.jpg was an error\n","file /content/cauliflower_training/cali/Downy Mildew/downy_mildew_0985.jpg was an error\n","file /content/cauliflower_training/cali/Alternaria Leaf Spot/alternaria_leaf_spot_0002.jpg was an error\n","there were 3 errors in 776 tests for an accuracy of 99.61\n"]},{"output_type":"display_data","data":{"text/plain":[""],"image/png":"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\n"},"metadata":{}},{"output_type":"stream","name":"stdout","text":["Classification Report:\n","----------------------\n"," precision recall f1-score support\n","\n","Alternaria Leaf Spot 1.0000 0.9868 0.9933 151\n"," Bacterial Soft Rot 1.0000 1.0000 1.0000 140\n"," Black Rot 0.9753 1.0000 0.9875 79\n"," Downy Mildew 0.9889 0.9889 0.9889 90\n"," Healthy 1.0000 1.0000 1.0000 316\n","\n"," accuracy 0.9961 776\n"," macro avg 0.9928 0.9951 0.9939 776\n"," weighted avg 0.9962 0.9961 0.9961 776\n","\n"]}],"source":["def predictor_from_df(test_df, test_ds):\n"," y_true = test_df['labels'].map(class_to_idx).values\n"," filepaths = test_df['filepaths'].values\n"," class_names_local = class_names\n"," errors = 0\n"," preds = model.predict(test_ds, verbose=1)\n"," pred_indices = np.argmax(preds, axis=1)\n"," for i, (p, t) in enumerate(zip(pred_indices, y_true)):\n"," if p != t:\n"," errors += 1\n"," print('file ', filepaths[i], ' was an error')\n","\n"," acc = (1 - errors/len(y_true)) * 100\n"," print(f'there were {errors} errors in {len(y_true)} tests for an accuracy of {acc:6.2f}')\n","\n"," cm = confusion_matrix(y_true, pred_indices)\n"," plt.figure(figsize=(12, 8))\n"," sns.heatmap(cm, annot=True, vmin=0, fmt='g', cmap='Blues', cbar=False)\n"," plt.xticks(np.arange(class_count)+.5, class_names_local, rotation=90)\n"," plt.yticks(np.arange(class_count)+.5, class_names_local, rotation=0)\n"," plt.xlabel(\"Predicted\")\n"," plt.ylabel(\"Actual\")\n"," plt.title(\"Confusion Matrix\")\n"," plt.show()\n","\n"," clr = classification_report(y_true, pred_indices, target_names=class_names_local, digits=4)\n"," print(\"Classification Report:\\n----------------------\\n\", clr)\n"," return errors, len(y_true)\n","\n","errors, tests = predictor_from_df(test_df, test_ds)"]},{"cell_type":"markdown","metadata":{"id":"FfexEKwBFPTf"},"source":["\n","#
Save the model"]},{"cell_type":"code","execution_count":21,"metadata":{"id":"R55mLBzWFPTf","colab":{"base_uri":"https://localhost:8080/"},"executionInfo":{"status":"ok","timestamp":1764242651841,"user_tz":-420,"elapsed":1147,"user":{"displayName":"Ping Ping","userId":"16193029904393259603"}},"outputId":"b46b3629-3591-4a4c-8b29-7cbff4f0d7dc"},"outputs":[{"output_type":"stream","name":"stderr","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"]},{"output_type":"stream","name":"stdout","text":["model was saved as ./nuts_99.61.h5\n"]}],"source":["subject='nuts'\n","acc=str(( 1-errors/tests) * 100)\n","index=acc.rfind('.')\n","acc=acc[:index + 3]\n","save_id= subject + '_' + str(acc) + '.h5'\n","model_save_loc=os.path.join(working_dir, save_id)\n","model.save(model_save_loc)\n","print ('model was saved as ' , model_save_loc )\n"]}],"metadata":{"accelerator":"GPU","colab":{"gpuType":"A100","provenance":[]},"kernelspec":{"display_name":"env","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":"3.11.9"}},"nbformat":4,"nbformat_minor":0}