laoding utils & app

app.py

@@ -0,0 +1,55 @@
+# Streamlit Library
+import streamlit as st 
+
+from utils import load_model, generate
+
+# Main Page
+# Title
+st.title("Butterflies Generator")
+# Sumamry Description Model
+st.write("This is Light GAN trained model")
+
+# Sidebar
+st.sidebar.subheader("This butterfly does not exist!, could you believe it?")
+# Logo folder
+st.sidebar.image("assets/logo.png", width=200)
+#
+st.sidebar.caption("Demo live created")
+
+# Loading & Model's Name
+repo_id = "ceyda/butterfly_cropped_uniq1K_512"
+gan_model = load_model(repo_id)
+
+# Butterflies Generator (4)
+n_butterflies = 4
+
+
+def run():
+    with st.spinner("Loading, Please! be patient..."):
+        # Inner Processing 
+        ims = generate(gan_model, n_butterflies)
+        # To save on
+        st.session_state["ims"]
+        
+     
+if "ims" not in st.session_state:
+    st.session_state["ims"] = None
+    run()
+    
+# Extracting info from ims    
+ims = st.session_state["ims"]
+
+#
+run_button = st.button(
+    "Please! Generate buttlerflies",
+    on_click= run(),
+    help="we are in flight, fasten your seatbelt."
+)
+
+if ims is not None:
+    cols = st.columns(n_butterflies)
+    for j, im in enumerate(ims):
+        i = j % n_butterflies
+        cols[i].image(im, use_column_width=True)
+        
+        

utils.py

@@ -0,0 +1,27 @@
+# Utils Standards
+
+# Numerical Management
+import numpy as np
+# Pythorch
+import torch
+# A hugging face library
+from huggan.pytorch.lightweight_gan.leightweight_gan import LightweightGAN
+
+"""Let's now set up the model functions"""
+
+def load_model(model_name = "ceyda/butterfly_cropped_uniq1K_512", model_version=None):
+    # GAN set-up
+    gan = LightweightGAN.from_pretrained(model_name, versio=model_version)
+    # GAN Inference Evaluation 
+    gan.eval()
+    return gan
+
+# Let's set-up a second function for generation form
+
+def generate(gan, batch_size=1):
+    with torch.no_grad():
+        # Cleaning process to properly fit it to the model
+        ims = gan.G(torch.randn(batch_size, gan.latent_dim)).clamp_(0.0, 1.0) * 255
+        #
+        ims = ims.permute(0, 2, 3, 1).deatch().cpu().numpy().astype(np.uint8)
+    return ims