Create app.py

app.py

@@ -1,38 +1,65 @@
-import streamlit as st
-import process
-import pandas as pd
+import gradio as gr
+import torch
+from semdiffusers import SemanticEditPipeline
 
-st.set_page_config(page_title="Data Anonymizer App")
+device='cuda'
 
-st.title("Data Anonymizer App")
+pipe = SemanticEditPipeline.from_pretrained(
+    "runwayml/stable-diffusion-v1-5",
+).to(device)
 
-st.sidebar.title("Data Upload")
-uploaded_file = st.sidebar.file_uploader("Choose a CSV file", type="csv")
+def infer(prompt, seed):
+    gen = torch.Generator(device=device)
+    gen.manual_seed(seed)
+    
+    out = pipe(
+        prompt=prompt, 
+        generator=gen, 
+        num_images_per_prompt=1, 
+        guidance_scale=7
+    )
+    
+    images = out.images[0]
+    
+    out_edit = pipe(
+        prompt=prompt, 
+        generator=gen, 
+        num_images_per_prompt=1, 
+        guidance_scale=7,
+        editing_prompt=['male person', 'female person'],     # Concepts to apply 
+        reverse_editing_direction=[True, False],             # Direction of guidance i.e. decrease the first and increase the second concept
+        edit_warmup_steps=[10, 10],                         # Warmup period for each concept
+        edit_guidance_scale=[4, 4],                         # Guidance scale for each concept
+        edit_threshold=[0.95, 0.95],                       # Threshold for each concept. Threshold equals the percentile of the latent space that will be discarded. I.e. threshold=0.99 uses 1% of the latent dimensions
+        edit_momentum_scale=0.3,                             # Momentum scale that will be added to the latent guidance
+        edit_mom_beta=0.6,                                   # Momentum beta
+        edit_weights=[1, 1]                                 # Weights of the individual concepts against each other
+    )
+    
+    images_edited = out_edit.images[0]
+    
+    return [
+        (images, 'Stable Diffusion'), 
+        (images_edited, 'Fair Diffusion')
+    ]
 
-if uploaded_file:
-    df = pd.read_csv(uploaded_file)
-    st.write("Original Data:")
-    st.write(df)
+inputs = [
+    gr.inputs.Textbox(label='Prompt'),
+    gr.inputs.Number(label='Seed', default=0, step=1)
+]
 
-    # process the data
-    processed_df, sensitive_cols = process.process_data(df)
+outputs = gr.outputs.Image(label='Images', type='numpy', number=2)
 
-    # display processed data
-    st.write("Processed Data:")
-    st.write(processed_df)
+title = 'Semantic Edit Pipeline'
+description = 'Semantic Edit Pipeline implementation using SemDiffusers.'
+article = "<h3 style='text-align: center'><a href='https://github.com/crowsonkb/semdiffusers'>SemDiffusers</a></h3>"
 
-    # ask for sensitive columns removal
-    if sensitive_cols:
-        st.write(f"The following columns contain sensitive data: {', '.join(sensitive_cols)}")
-        if st.checkbox("Remove sensitive columns"):
-            processed_df.drop(columns=sensitive_cols, inplace=True)
-        else:
-            st.write("Sensitive columns will not be removed.")
-
-    # ask for k-anonymity
-    if st.checkbox("Apply k-anonymity"):
-        k = st.number_input("Enter the value of k", min_value=1)
-        processed_df = process.apply_k_anonymity(processed_df, k)
-
-    st.write("Final Processed Data:")
-    st.write(processed_df)
+gr.Interface(
+    infer,
+    inputs,
+    outputs,
+    title=title,
+    description=description,
+    article=article,
+    theme='compact'
+).launch();