Update app.py

app.py

@@ -3,26 +3,24 @@ import tensorflow as tf
 from tensorflow import keras
 from tensorflow.keras import layers
 import numpy as np
-import matplotlib.pyplot as plt
 import os
 from PIL import Image
-import math
 
 # =========================================
 # 1. SETTINGS
 # =========================================
-IMG_SIZE = 48   # Slightly larger than CIFAR for better detail
+IMG_SIZE = 48   # Small size for CPU speed
 CHANNELS = 3
 TIMESTEPS = 200
 BATCH_SIZE = 32
 
 # =========================================
-# 2. GENERATE CUSTOM "VEDA" DATASET
+# 2. GENERATE CUSTOM "VEDA" DATASET (MATH ART)
 # =========================================
-def generate_veda_patterns(num_images=1000):
+def generate_veda_patterns(num_images=500):
     """
     Creates a dataset of mathematical/spiritual patterns (Mandalas/Energy Orbs)
-    instead of using boring CIFAR-10 images.
+    instead of using boring photos.
     """
     print(f"Generating {num_images} unique Veda patterns...")
     data = []
@@ -46,7 +44,7 @@ def generate_veda_patterns(num_images=1000):
         # Colorize (RGB)
         img = np.zeros((IMG_SIZE, IMG_SIZE, 3))
         
-        # Random colors
+        # Random color boosters
         r_boost = np.random.rand()
         g_boost = np.random.rand()
         b_boost = np.random.rand()
@@ -62,6 +60,7 @@ def generate_veda_patterns(num_images=1000):
 # =========================================
 # 3. DIFFUSION MATH
 # =========================================
+# Cast to float32 immediately to avoid type errors
 beta_np = np.linspace(0.0001, 0.02, TIMESTEPS)
 beta = tf.cast(beta_np, dtype=tf.float32)
 alpha = 1.0 - beta
@@ -89,7 +88,6 @@ def build_unet():
     x = layers.Conv2D(32, 3, padding="same", activation="relu")(image_input)
     
     # Simple addition of time info
-    # Note: We keep architecture simple for CPU speed
     x = layers.Conv2D(64, 3, strides=2, padding="same", activation="relu")(x) # 24x24
     x = layers.Conv2D(128, 3, strides=2, padding="same", activation="relu")(x) # 12x12
     x = layers.Conv2D(256, 3, padding="same", activation="relu")(x)
@@ -102,29 +100,32 @@ def build_unet():
     outputs = layers.Conv2D(CHANNELS, 1, padding="same")(x)
     return keras.Model([image_input, time_input], outputs)
 
+# Initialize Model
 model = build_unet()
 optimizer = keras.optimizers.Adam(learning_rate=1e-4)
 loss_fn = keras.losses.MeanSquaredError()
 
-# Load weights if trained
+# Load weights if they exist (Persistence)
 if os.path.exists("veda_art_model.keras"):
     try:
         model.load_weights("veda_art_model.keras")
-        print("Brain Loaded!")
+        print("Previous brain loaded successfully!")
     except:
-        pass
+        print("Starting fresh brain.")
 
 # =========================================
 # 5. TRAINING FUNCTION
 # =========================================
-def train_model(epochs):
-    # 1. Generate fresh data
-    x_train = generate_veda_patterns(num_images=500) # 500 images is enough for patterns
+def train_model():
+    # 1. Generate fresh data on the fly
+    yield "Generating Sacred Geometry Data..."
+    x_train = generate_veda_patterns(num_images=500) 
     dataset = tf.data.Dataset.from_tensor_slices(x_train).batch(BATCH_SIZE).shuffle(500)
     
-    yield "Dataset Generated (Mathematical Patterns). Starting Training..."
+    yield "Dataset Ready. Starting Training Loop (5 Epochs)..."
     
-    for epoch in range(int(epochs)):
+    epochs = 5
+    for epoch in range(epochs):
         total_loss = 0
         steps = 0
         for batch in dataset:
@@ -141,10 +142,11 @@ def train_model(epochs):
             total_loss += loss
             steps += 1
             
-        yield f"Epoch {epoch+1}/{epochs} - Loss: {total_loss/steps:.4f}"
+        # Save progress
         model.save_weights("veda_art_model.keras")
+        yield f"Epoch {epoch+1}/{epochs} Complete - Loss: {total_loss/steps:.4f}"
         
-    yield "Training Done! Go to 'Dream' tab."
+    yield "Training Done! Go to 'Dream' tab to generate images."
 
 # =========================================
 # 6. GENERATION FUNCTION
@@ -184,9 +186,9 @@ def generate_art():
 # =========================================
 # 7. UI
 # =========================================
-with gr.Blocks(title="Veda Art Engine") as demo:
-    gr.Markdown("# Veda Art Engine")
-    gr.Markdown("This AI generates its own training data using math, then learns to dream up new patterns.")
+with gr.Blocks(title="Akasha Art Engine") as demo:
+    gr.Markdown("# Akasha Art Engine")
+    gr.Markdown("A custom Diffusion model that learns sacred geometry from pure math.")
     
     with gr.Tab("Dream"):
         gen_btn = gr.Button("Generate Sacred Pattern", variant="primary")
@@ -194,9 +196,11 @@ with gr.Blocks(title="Veda Art Engine") as demo:
         gen_btn.click(generate_art, outputs=out_img)
         
     with gr.Tab("Train"):
-        gr.Markdown("Click to generate a dataset of mathematical mandalas and train the AI on them.")
+        gr.Markdown("Click below to generate data and train the model.")
         train_btn = gr.Button("Train AI (5 Epochs)")
         log = gr.Textbox(label="Status")
-        train_btn.click(lambda: train_model(5), outputs=log)
+        
+        # This connects the stream properly
+        train_btn.click(train_model, outputs=log)
 
 demo.launch()