Update app.py

app.py

@@ -50,7 +50,7 @@ api_url = 'https://a02q342s5b.execute-api.us-east-2.amazonaws.com/reinvent-demo-
 
 
 # Creating Tabs
-tab1, tab2, tab3 = st.tabs(["Image Generation",  "Architecture", "Code"])
+tab1, tab2, tab3, tab4 = st.tabs(["Image Generation",  "Architecture", "Stable Diffusion Architecture", "Code"])
 
 with tab1:
     # Create two columns for layout
@@ -74,7 +74,7 @@ with tab1:
             "A lost city rediscovered in the Amazon jungle, overgrown with plants, in the style of a vintage travel poster",
             "A steampunk train emitting clouds of steam as it races through a mountain pass, digital art",
             "An enchanted forest with bioluminescent trees and fairies dancing, in a Studio Ghibli style",
-            "A polar bear working as an auto mechanic, changing the oil on a car, reminiscent of Art Nouveau",
+            "A portrait of an elegant alien empress with a detailed headdress, reminiscent of Art Nouveau",
             "A post-apocalyptic Tokyo with nature reclaiming skyscrapers, in the style of a concept art",
             "A mythical phoenix rising from ashes, vibrant colors, with a nebula in the background",
             "A cybernetic wolf in a neon-lit city, cyberpunk theme, rain-drenched streets",
@@ -150,78 +150,78 @@ with tab1:
                 else:
                     st.error(f"Error: {response_one.text}")
 
-# with tab2:
+# with pass:
    
-# st.title('Llama 2 7B Text Generation with AWS Inferentia 2')
-
-# params = {
-#     "do_sample" : True,
-#     "top_p": 0.6,
-#     "temperature": 0.9,
-#     "top_k": 50,
-#     "max_new_tokens": 512,
-#     "repetition_penalty": 1.03,
-# }
-    
-# if "messages" not in st.session_state:
-#     st.session_state.messages = [
-#         {"role": "system", "content": "You are a helpful Travel Planning Assistant. You respond with only 1-2 sentences."},
-#         {'role': 'user', 'content': 'Where can I travel in the fall for cloudy, rainy, and beautiful views?'},
-#     ]
-
-# for message in st.session_state.messages:
-#     with st.chat_message(message["role"]):
-#         st.markdown(message["content"])
-
-# with st.chat_message("assistant"):
-#     message_placeholder = st.empty()
-#     full_response = ""
-#     prompt_input_one = {
-#             "prompt": st.session_state.messages,
-#             "parameters": params,
-#             "endpoint": "huggingface-pytorch-inference-neuronx-2023-11-28-16-09-51-708"
-#         }
-    
-#     response_one = requests.post(api_url, json=prompt_input_one)
-
-#     if response_one.status_code == 200:
-#         result_one = response_one.json()
-#         # st.success(f"Prediction result: {result}")
-#         full_response += result_one["generation"]
-#     else:
-#         st.error(f"Error: {response_one.text}")
-
-#     message_placeholder.markdown(full_response)
-# st.session_state.messages.append({"role": "assistant", "content": full_response})
-
-# if prompt := st.chat_input("What is up?"):
-#     st.session_state.messages.append({"role": "user", "content": prompt})
-#     print(st.session_state.messages)
-#     with st.chat_message("user"):
-#         st.markdown(prompt)
-    
-#     with st.chat_message("assistant"):
-#         message_placeholder = st.empty()
-#         new_response = ""
-#         prompt_input_one = {
-#                 "prompt": st.session_state.messages,
-#                 "parameters": params,
-#                 "endpoint": "huggingface-pytorch-inference-neuronx-2023-11-28-16-09-51-708"
-#             }
+    # st.title('Llama 2 7B Text Generation with AWS Inferentia 2')
+
+    # params = {
+    #     "do_sample" : True,
+    #     "top_p": 0.6,
+    #     "temperature": 0.9,
+    #     "top_k": 50,
+    #     "max_new_tokens": 512,
+    #     "repetition_penalty": 1.03,
+    # }
         
-#         response_one = requests.post(api_url, json=prompt_input_one)
-
-#         if response_one.status_code == 200:
-#             result_one = response_one.json()
-#             # st.success(f"Prediction result: {result}")
-#             new_response += result_one["generation"]
-#         else:
-#             st.error(f"Error: {response_one.text}")
-
-#         message_placeholder.markdown(new_response)
-#     st.session_state.messages.append({"role": "assistant", "content": new_response})
-
-
+    # if "messages" not in st.session_state:
+    #     st.session_state.messages = [
+    #         {"role": "system", "content": "You are a helpful Travel Planning Assistant. You respond with only 1-2 sentences."},
+    #         {'role': 'user', 'content': 'Where can I travel in the fall for cloudy, rainy, and beautiful views?'},
+    #     ]
+
+    # for message in st.session_state.messages:
+    #     with st.chat_message(message["role"]):
+    #         st.markdown(message["content"])
+
+    # with st.chat_message("assistant"):
+    #     message_placeholder = st.empty()
+    #     full_response = ""
+    #     prompt_input_one = {
+    #             "prompt": st.session_state.messages,
+    #             "parameters": params,
+    #             "endpoint": "huggingface-pytorch-inference-neuronx-2023-11-28-16-09-51-708"
+    #         }
+        
+    #     response_one = requests.post(api_url, json=prompt_input_one)
+
+    #     if response_one.status_code == 200:
+    #         result_one = response_one.json()
+    #         # st.success(f"Prediction result: {result}")
+    #         full_response += result_one["generation"]
+    #     else:
+    #         st.error(f"Error: {response_one.text}")
+
+    #     message_placeholder.markdown(full_response)
+    # st.session_state.messages.append({"role": "assistant", "content": full_response})
+
+    # if prompt := st.chat_input("What is up?"):
+    #     st.session_state.messages.append({"role": "user", "content": prompt})
+    #     print(st.session_state.messages)
+    #     with st.chat_message("user"):
+    #         st.markdown(prompt)
+        
+    #     with st.chat_message("assistant"):
+    #         message_placeholder = st.empty()
+    #         new_response = ""
+    #         prompt_input_one = {
+    #                 "prompt": st.session_state.messages,
+    #                 "parameters": params,
+    #                 "endpoint": "huggingface-pytorch-inference-neuronx-2023-11-28-16-09-51-708"
+    #             }
+            
+    #         response_one = requests.post(api_url, json=prompt_input_one)
+
+    #         if response_one.status_code == 200:
+    #             result_one = response_one.json()
+    #             # st.success(f"Prediction result: {result}")
+    #             new_response += result_one["generation"]
+    #         else:
+    #             st.error(f"Error: {response_one.text}")
+
+    #         message_placeholder.markdown(new_response)
+    #     st.session_state.messages.append({"role": "assistant", "content": new_response})
+
+    pass
 
 
 with tab2:        
@@ -276,6 +276,28 @@ with tab2:
         """)
 
 with tab3:
+    left_column, _, right_column = st.columns([2,.2,3])
+
+    with right_column:
+        # Define Streamlit UI elements
+        st.markdown("""<br>""", unsafe_allow_html=True)
+        st.image('./sdxl_arch.png', caption=f"SDXL Architecture") 
+
+    with left_column:  
+        st.write("## SDXL Architecture Overview")
+        
+        st.write("""
+The stable diffusion model takes both a latent seed and a text prompt as an input. The latent seed is then used to generate random latent image representations of size 64×64 where as the text prompt is transformed to text embeddings of size 77×768 via CLIP's text encoder.
+
+Next the U-Net iteratively denoises the random latent image representations while being conditioned on the text embeddings. The output of the U-Net, being the noise residual, is used to compute a denoised latent image representation via a scheduler algorithm. Many different scheduler algorithms can be used for this computation, each having its pro- and cons. 
+
+
+Theory on how the scheduler algorithm function is out-of-scope for this demo, but in short one should remember that they compute the predicted denoised image representation from the previous noise representation and the predicted noise residual. 
+
+The denoising process is repeated ca. 50 times to step-by-step retrieve better latent image representations. Once complete, the latent image representation is decoded by the decoder part of the variational auto encoder.
+        """)    
+
+with tab4:
     with st.expander("(1) Deploy GenAI Model to AWS Inferentia 2 Instance and Amazon SageMaker Endpoint"):
         st.markdown(
             """