app.py

import streamlit as st
# Set the page layout to 'wide'
st.set_page_config(layout="wide")
import requests
from PIL import Image
from io import BytesIO
# from IPython.display import display
import base64
import time
import random



# helper decoder
def decode_base64_image(image_string):
    base64_image = base64.b64decode(image_string)
    buffer = BytesIO(base64_image)
    return Image.open(buffer)

# display PIL images as grid
def display_image(image=None,width=500,height=500):
    img = image.resize((width, height))
    return img

def pretty_print(messages):
    for message in messages:
        return f"{message['role']}: {message['content']}"

# API Gateway endpoint URL
api_url = 'https://a02q342s5b.execute-api.us-east-2.amazonaws.com/reinvent-demo-inf2-sm-20231114'

# # Define the CSS to change the text input background color
# input_field_style = """
# <style>
#     /* Customize the text input field background and text color */
#     .stTextInput input {
#         background-color: #fbd8bf; /* 'Rind' color */
#         color: #232F3E; /* Dark text color */
#     }
#     /* You might also want to change the color for textarea if you're using it */
#     .stTextArea textarea {
#         background-color: #fbd8bf; /* 'Rind' color */
#         color: #232F3E; /* Dark text color */
#     }
# </style>
# """

# # Inject custom styles into the Streamlit app
# st.markdown(input_field_style, unsafe_allow_html=True)


# Creating Tabs
tab1, tab2, tab3, tab4 = st.tabs(["Image Generation",  "Architecture", "Stable Diffusion Architecture", "Code"])

with tab1:
    # Create two columns for layout
    left_column, right_column = st.columns(2)

    with right_column:
        cont = st.container()

    # ===========
    with left_column:
        # Define Streamlit UI elements
        st.title('Stable Diffusion XL Image Generation with AWS Inferentia2')

        sample_prompts = [
            "A futuristic cityscape at sunset, cyberpunk",
            "A serene landscape with mountains and a river, photorealistic style",
            "An astronaut riding a horse, artistic and surreal",
            "A robot playing chess in a medieval setting, high detail",
            "An underwater scene with colorful coral reefs and fish, vibrant colors",
            "Raccoon astronaut in space, sci-fi, future, cold color palette, muted colors, detailed, 8k",
            "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 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",
            "A high fantasy battle scene with dragons in the sky and knights on the ground, epic scale",
            "An ice castle on a lonely mountain peak, under the northern lights, fantasy illustration",
            "A surreal landscape where giant flowers bloom in the desert, with a distant thunderstorm, hyperrealism"
        ]

        def set_random_prompt():
            # This function will be called when the button is clicked
            random_prompt = random.choice(sample_prompts)
            # Update the session state for the input field
            st.session_state.prompt_one = random_prompt

        prompt_one = st.text_area("Enter your prompt:", 
                                  
                                  key="prompt_one")
        
        st.button('Random Prompt', on_click=set_random_prompt)

        # Number of inference steps
        num_inference_steps_one = st.slider("Number of Inference Steps", 
                                    min_value=1, 
                                    max_value=100, 
                                    value=15, 
                                    help="More steps might improve quality, with diminishing marginal returns. 30-50 seems best, but your mileage may vary.")

        # Create an expandable section for optional parameters
        with st.expander("Optional Parameters"):
            # Random seed input
            seed_one = st.number_input("Random seed", 
                                value=555, 
                                help="Set to the same value to generate the same image if other inputs are the same, change to generate a different image for same inputs.")

            # Negative prompt input
            negative_prompt_one = st.text_area("Enter your negative prompt:", 
                            "cartoon, graphic, text, painting, crayon, graphite, abstract glitch, blurry")

        if st.button('Generate Image'):
            with st.spinner(f'Generating Image with {num_inference_steps_one} iterations'):
                start_time = time.time()
                # ===============
                # Example input data
                prompt_input_one = {
                    "prompt": prompt_one,
                    "parameters": {
                        "num_inference_steps": num_inference_steps_one,
                        "seed": seed_one,
                        "negative_prompt": negative_prompt_one
                    },
                    "endpoint": "huggingface-pytorch-inference-neuronx-2023-11-14-21-22-10-388"
                }

                # Make API request
                response_one = requests.post(api_url, json=prompt_input_one)

                # Process and display the response
                if response_one.status_code == 200:
                    result_one = response_one.json()
                    # st.success(f"Prediction result: {result}")
                    image_one = display_image(decode_base64_image(result_one["generated_images"][0]))
                    cont.image(image_one, 
                        caption=f"{prompt_one}") 
                    end_time = time.time()
                    total_time = round(end_time - start_time, 2)
                    cont.text(f"Prompt: {prompt_one}")
                    cont.text(f"Number of Iterations: {num_inference_steps_one}")
                    cont.text(f"Random Seed: {seed_one}")
                    cont.text(f'Total time taken: {total_time} seconds')
                    # Calculate and display the time per iteration in milliseconds
                    time_per_iteration_ms = (total_time / num_inference_steps_one)
                    cont.text(f'Time per iteration: {time_per_iteration_ms:.2f} seconds')
                else:
                    st.error(f"Error: {response_one.text}")

# 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"
    #             }
            
    #         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:        
    # ===========
    left_column, _, right_column = st.columns([2,.2,3])

    with right_column:
        # Define Streamlit UI elements
        st.markdown("""<br>""", unsafe_allow_html=True)
        st.markdown("""<br>""", unsafe_allow_html=True)
        st.markdown("""<br>""", unsafe_allow_html=True)
        st.markdown("""<br>""", unsafe_allow_html=True)
        st.markdown("""<br>""", unsafe_allow_html=True)
        st.image('./architecture.png', caption=f"Application Architecture") 

    with left_column:  
        st.write("## Architecture Overview")
        st.write("This diagram illustrates the architecture of our Generative AI service, which is composed of several interconnected AWS services, notable Amazon Elastic Compute Cloud (Amazon EC2). Here's a detailed look at each component:")

        with st.expander("(1) Inference Models"):
            st.markdown("""
            - The architecture starts with our trained machine learning models hosted on Amazon SageMaker, running on AWS Inferentia 2 instance (`inf2.xlarge`).
            - There are two models shown here, Stable Diffusion XL for image generation, and Llama 2 7B for text generation.
            """)

        with st.expander("(2) Amazon SageMaker Endpoints"):
            st.markdown("""
            - The models are exposed via SageMaker Endpoints, which provide scalable and secure real-time inference services.
            - These endpoints are the interfaces through which the models receive input data and return predictions.
            """)

        with st.expander("(3) AWS Lambda"):
            st.markdown("""
            - AWS Lambda functions serve as the middle layer, handling the logic of communicating with the SageMaker Endpoints.
            - Lambda can process the incoming requests, perform any necessary transformations, call the endpoints, and then process the results before sending them back.
            """)

        with st.expander("(4) Amazon API Gateway"):
            st.markdown("""
            - The processed results from Lambda are then routed through Amazon API Gateway.
            - API Gateway acts as a front door to manage all incoming API requests, including authorization, throttling, and CORS handling.
            """)

        with st.expander("(5) Streamlit Frontend"):
            st.markdown("""
            - Finally, our Streamlit application provides a user-friendly interface for end-users to interact with the service.
            - It sends requests to the API Gateway and displays the returned predictions from the machine learning models.
            """)

        st.write("""
        In summary, this architecture enables a scalable, serverless, and responsive Generative AI service that can serve real-time predictions to users directly from a web interface.
        """)

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(
            """
            [Source] This code is modified from this fantastic blog by Phil Schmid at HuggingFace: https://www.philschmid.de/inferentia2-stable-diffusion-xl

            # Deploy Stable Diffusion on AWS inferentia2 with Amazon SageMaker

            In this end-to-end tutorial, you will learn how to deploy and speed up Stable Diffusion XL inference using AWS Inferentia2 and [optimum-neuron](https://huggingface.co./docs/optimum-neuron/index) on Amazon SageMaker. [Optimum Neuron](https://huggingface.co./docs/optimum-neuron/index) is the interface between the Hugging Face Transformers & Diffusers library and AWS Accelerators including AWS Trainium and AWS Inferentia2. 

            You will learn how to: 

            1. Convert Stable Diffusion XL to AWS Neuron (Inferentia2) with `optimum-neuron`
            2. Create a custom `inference.py` script for Stable Diffusion
            3. Upload the neuron model and inference script to Amazon S3
            4. Deploy a Real-time Inference Endpoint on Amazon SageMaker
            5. Generate images using the deployed model

            ## Quick intro: AWS Inferentia 2

            [AWS inferentia (Inf2)](https://aws.amazon.com/de/ec2/instance-types/inf2/) are purpose-built EC2 for deep learning (DL) inference workloads. Inferentia 2 is the successor of [AWS Inferentia](https://aws.amazon.com/ec2/instance-types/inf1/?nc1=h_ls), which promises to deliver up to 4x higher throughput and up to 10x lower latency.

            | instance size | accelerators | Neuron Cores | accelerator memory | vCPU | CPU Memory | on-demand price ($/h) |
            | ------------- | ------------ | ------------ | ------------------ | ---- | ---------- | --------------------- |
            | inf2.xlarge   | 1            | 2            | 32                 | 4    | 16         | 0.76                  |
            | inf2.8xlarge  | 1            | 2            | 32                 | 32   | 128        | 1.97                  |
            | inf2.24xlarge | 6            | 12           | 192                | 96   | 384        | 6.49                  |
            | inf2.48xlarge | 12           | 24           | 384                | 192  | 768        | 12.98                 |

            Additionally, inferentia 2 will support the writing of custom operators in c++ and new datatypes, including `FP8` (cFP8).

            Let's get started! 🚀

            *If you are going to use Sagemaker in a local environment (not SageMaker Studio or Notebook Instances). You need access to an IAM Role with the required permissions for Sagemaker. You can find [here](https://docs.aws.amazon.com/sagemaker/latest/dg/sagemaker-roles.html) more about it.*

            ## 1. Convert Stable Diffusion to AWS Neuron (Inferentia2) with `optimum-neuron`

            We are going to use the [optimum-neuron](https://huggingface.co./docs/optimum-neuron/index) to compile/convert our model to neuronx. Optimum Neuron provides a set of tools enabling easy model loading, training and inference on single- and multi-Accelerator settings for different downstream tasks. 

            As a first step, we need to install the `optimum-neuron` and other required packages.

            *Tip: If you are using Amazon SageMaker Notebook Instances or Studio you can go with the `conda_python3` conda kernel.*



            ```python
            # Install the required packages
            %pip install "optimum-neuron==0.0.13" "diffusers==0.21.4" --upgrade
            %pip install "sagemaker>=2.197.0"  --upgrade
            ```

            After we have installed the `optimum-neuron` we can convert load and convert our model.

            We are going to use the [stabilityai/stable-diffusion-xl-base-1.0](hstabilityai/stable-diffusion-xl-base-1.0) model. Stable Diffusion XL (SDXL) from [Stability AI](https://stability.ai/) is the newset text-to-image generation model, which can create photorealistic images with detailed imagery and composition compared to previous SD models, including SD 2.1.

            At the time of writing, the [AWS Inferentia2 does not support dynamic shapes for inference](https://awsdocs-neuron.readthedocs-hosted.com/en/latest/general/arch/neuron-features/dynamic-shapes.html?highlight=dynamic%20shapes#), which means that the we need to specify our image size in advanced for compiling and inference. 

            In simpler terms, this means we need to define the input shapes for our prompt (sequence length), batch size, height and width of the image.

            We precompiled the model with the following parameters and pushed it to the Hugging Face Hub: 
            * `height`: 1024
            * `width`: 1024
            * `sequence_length`: 128
            * `num_images_per_prompt`: 1
            * `batch_size`: 1
            * `neuron`: 2.15.0


            _Note: If you want to compile your own model or a different Stable Diffusion XL checkpoint you need to use ~120GB of memory and the compilation can take ~45 minutes. We used an `inf2.8xlarge` ec2 instance with the [Hugging Face Neuron Deep Learning AMI](https://aws.amazon.com/marketplace/pp/prodview-gr3e6yiscria2) to compile the model._


            ```python
            from huggingface_hub import snapshot_download

            # compiled model id
            compiled_model_id = "aws-neuron/stable-diffusion-xl-base-1-0-1024x1024"

            # save compiled model to local directory
            save_directory = "sdxl_neuron"
            # Downloads our compiled model from the HuggingFace Hub 
            # using the revision as neuron version reference
            # and makes sure we exlcude the symlink files and "hidden" files, like .DS_Store, .gitignore, etc.
            snapshot_download(compiled_model_id, revision="2.15.0", local_dir=save_directory, local_dir_use_symlinks=False, allow_patterns=["[!.]*.*"])


            ###############################################
            # COMMENT IN BELOW TO COMPILE DIFFERENT MODEL #
            ###############################################
            #
            # from optimum.neuron import NeuronStableDiffusionXLPipeline
            #
            # # model id you want to compile
            # vanilla_model_id = "stabilityai/stable-diffusion-xl-base-1.0"
            #
            # # configs for compiling model
            # compiler_args = {"auto_cast": "all", "auto_cast_type": "bf16"}
            # input_shapes = {
            #   "height": 1024, # width of the image
            #   "width": 1024, # height of the image
            #   "num_images_per_prompt": 1, # number of images to generate per prompt
            #   "batch_size": 1 # batch size for the model
            #   }
            #
            # sd = NeuronStableDiffusionXLPipeline.from_pretrained(vanilla_model_id, export=True, **input_shapes, **compiler_args)
            #
            # # Save locally or upload to the HuggingFace Hub
            # save_directory = "sdxl_neuron"
            # sd.save_pretrained(save_directory)
            ```

            ## 2. Create a custom `inference.py` script for Stable Diffusion

            The [Hugging Face Inference Toolkit](https://github.com/aws/sagemaker-huggingface-inference-toolkit) supports zero-code deployments on top of the [pipeline feature](https://huggingface.co./transformers/main_classes/pipelines.html) from 🤗 Transformers. This allows users to deploy Hugging Face transformers without an inference script [[Example](https://github.com/huggingface/notebooks/blob/master/sagemaker/11_deploy_model_from_hf_hub/deploy_transformer_model_from_hf_hub.ipynb)]. 

            Currently is this feature not supported with AWS Inferentia2, which means we need to provide an `inference.py` for running inference. But `optimum-neuron` has integrated support for the 🤗 Diffusers pipeline feature. That way we can use the `optimum-neuron` to create a pipeline for our model.

            If you want to know more about the `inference.py` script check out this [example](https://github.com/huggingface/notebooks/blob/master/sagemaker/17_custom_inference_script/sagemaker-notebook.ipynb). It explains amongst other things what the `model_fn` and `predict_fn` are. 


            ```python
            # create code directory in our model directory
            !mkdir {save_directory}/code
            ```

            We are using the `NEURON_RT_NUM_CORES=2` to make sure that each HTTP worker uses 2 Neuron core to maximize throughput.


            ```python
            %%writefile {save_directory}/code/inference.py
            import os
            # To use two neuron core per worker
            os.environ["NEURON_RT_NUM_CORES"] = "2"
            import torch
            import torch_neuronx
            import base64
            from io import BytesIO
            from optimum.neuron import NeuronStableDiffusionXLPipeline


            def model_fn(model_dir):
                # load local converted model into pipeline
                pipeline = NeuronStableDiffusionXLPipeline.from_pretrained(model_dir, device_ids=[0, 1])
                return pipeline


            def predict_fn(data, pipeline):
                # extract prompt from data
                prompt = data.pop("inputs", data)
                
                parameters = data.pop("parameters", None)
                
                if parameters is not None:
                    generated_images = pipeline(prompt, **parameters)["images"]
                else:
                    generated_images = pipeline(prompt)["images"]
                    
                # postprocess convert image into base64 string
                encoded_images = []
                for image in generated_images:
                    buffered = BytesIO()
                    image.save(buffered, format="JPEG")
                    encoded_images.append(base64.b64encode(buffered.getvalue()).decode())

                # always return the first 
                return {"generated_images": encoded_images}
            ```

            ## 3. Upload the neuron model and inference script to Amazon S3

            Before we can deploy our neuron model to Amazon SageMaker we need to upload it all our model artifacts to Amazon S3.

            _Note: Currently `inf2` instances are only available in the `us-east-2` & `us-east-1` region [[REF](https://aws.amazon.com/de/about-aws/whats-new/2023/05/sagemaker-ml-inf2-ml-trn1-instances-model-deployment/)]. Therefore we need to force the region to us-east-2._

            Lets create our SageMaker session and upload our model to Amazon S3.


            ```python
            import sagemaker
            import boto3
            sess = sagemaker.Session()
            # sagemaker session bucket -> used for uploading data, models and logs
            # sagemaker will automatically create this bucket if it not exists
            sagemaker_session_bucket=None
            if sagemaker_session_bucket is None and sess is not None:
                # set to default bucket if a bucket name is not given
                sagemaker_session_bucket = sess.default_bucket()

            try:
                role = sagemaker.get_execution_role()
            except ValueError:
                iam = boto3.client('iam')
                role = iam.get_role(RoleName='sagemaker_execution_role')['Role']['Arn']

            sess = sagemaker.Session(default_bucket=sagemaker_session_bucket)

            print(f"sagemaker role arn: {role}")
            print(f"sagemaker bucket: {sess.default_bucket()}")
            print(f"sagemaker session region: {sess.boto_region_name}")
            assert sess.boto_region_name in ["us-east-2", "us-east-1"] , "region must be us-east-2 or us-west-2, due to instance availability"
            ```

            We create our `model.tar.gz` with our `inference.py`` script



            ```python
            # create a model.tar.gz archive with all the model artifacts and the inference.py script.
            %cd {save_directory}
            !tar zcvf model.tar.gz *
            %cd ..
            ```

            Next, we upload our `model.tar.gz` to Amazon S3 using our session bucket and `sagemaker` sdk.


            ```python
            from sagemaker.s3 import S3Uploader

            # create s3 uri
            s3_model_path = f"s3://{sess.default_bucket()}/neuronx/sdxl"

            # upload model.tar.gz
            s3_model_uri = S3Uploader.upload(local_path=f"{save_directory}/model.tar.gz", desired_s3_uri=s3_model_path)
            print(f"model artifcats uploaded to {s3_model_uri}")
            ```

            ## 4. Deploy a Real-time Inference Endpoint on Amazon SageMaker

            After we have uploaded our model artifacts to Amazon S3 can we create a custom `HuggingfaceModel`. This class will be used to create and deploy our real-time inference endpoint on Amazon SageMaker.

            The `inf2.xlarge` instance type is the smallest instance type with AWS Inferentia2 support. It comes with 1 Inferentia2 chip with 2 Neuron Cores. This means we can use 2 Neuron Cores to minimize latency for our image generation. 


            ```python
            from sagemaker.huggingface.model import HuggingFaceModel

            # create Hugging Face Model Class
            huggingface_model = HuggingFaceModel(
            model_data=s3_model_uri,        # path to your model.tar.gz on s3
            role=role,                      # iam role with permissions to create an Endpoint
            transformers_version="4.34.1",  # transformers version used
            pytorch_version="1.13.1",       # pytorch version used
            py_version='py310',             # python version used
            model_server_workers=1,         # number of workers for the model server
            )

            # deploy the endpoint endpoint
            predictor = huggingface_model.deploy(
                initial_instance_count=1,      # number of instances
                instance_type="ml.inf2.xlarge", # AWS Inferentia Instance
                volume_size = 100
            )
            # ignore the "Your model is not compiled. Please compile your model before using Inferentia." warning, we already compiled our model.
            ```

            # 5.Generate images using the deployed model

            The `.deploy()` returns an `HuggingFacePredictor` object which can be used to request inference. Our endpoint expects a `json` with at least `inputs` key. The `inputs` key is the input prompt for the model, which will be used to generate the image. Additionally, we can provide inference parameters, e.g. `num_inference_steps`.

            The `predictor.predict()` function returns a `json` with the `generated_images` key. The `generated_images` key contains the `1` generated image as a `base64` encoded string. To decode our response we added a small helper function `decode_base64_to_image` which takes the `base64` encoded string and returns a `PIL.Image` object and `display_image` displays them.


            ```python
            from PIL import Image
            from io import BytesIO
            from IPython.display import display
            import base64

            # helper decoder
            def decode_base64_image(image_string):
            base64_image = base64.b64decode(image_string)
            buffer = BytesIO(base64_image)
            return Image.open(buffer)

            # display PIL images as grid
            def display_image(image=None,width=500,height=500):
                img = image.resize((width, height))
                display(img)
            ```

            Now, lets generate some images. As example `A dog trying catch a flying pizza in style of comic book, at a street corner.`. Generating an image with 25 steps takes around ~6 seconds, except for the first request which can take 45-60s. 
            _note: If the request times out, just rerun again. Only the first request takes a long time._


            ```python
            prompt = "A dog trying catch a flying pizza at a street corner, comic book, well lit, night time"

            # run prediction
            response = predictor.predict(data={
            "inputs": prompt,
            "parameters": {
                "num_inference_steps" : 25,
                "negative_prompt" : "disfigured, ugly, deformed"
                } 
            }
            )

            # decode and display image
            display_image(decode_base64_image(response["generated_images"][0]))
            ```




            ### Delete model and endpoint

            To clean up, we can delete the model and endpoint.


            ```python
            predictor.delete_model()
            predictor.delete_endpoint()
            ```


            ```python

            ```

            """

                )
        
    with st.expander("(2) AWS Lambda Function to handle inference requests"):
        st.markdown(
            """
```python
import boto3
import json

def lambda_handler(event, context):
    # SageMaker endpoint details
    endpoint_name = 'INSERT_YOUR_SAGEMAKER_ENDPOINT_NAME_HERE'
    runtime = boto3.client('sagemaker-runtime')

    # Sample input data (modify as per your model's input requirements)
    # Get the prompt from the Lambda function input
    print("======== event payload: ==========")
    print(event['body'])
    
    print("======== prompt payload: ==========")
    event_parsed = json.loads(event['body'])
    prompt = event_parsed.get('prompt', '')
    print(prompt)
    print("======== params payload: ==========")
    params = event_parsed.get('parameters','')
    print(params)
    
    # Prepare input data
    model_input = {
        'inputs': prompt,
        'parameters': params
    }
    
    input_data = json.dumps(model_input)

    # Make a prediction request to the SageMaker endpoint
    response = runtime.invoke_endpoint(EndpointName=endpoint_name,
                                       ContentType='application/json',
                                       Body=input_data)

    # Parse the response
    result = response['Body'].read()
    return {
        'statusCode': 200,
        'body': result
    }

```

            """
        )

    with st.expander("(3) Streamlit app.py, running on Amazon EC2 t2.micro instance"):
        st.markdown(
            """
```python
import streamlit as st
# Set the page layout to 'wide'
st.set_page_config(layout="wide")
import requests
from PIL import Image
from io import BytesIO
import base64
import time



# helper decoder
def decode_base64_image(image_string):
    base64_image = base64.b64decode(image_string)
    buffer = BytesIO(base64_image)
    return Image.open(buffer)

# display PIL images as grid
def display_image(image=None,width=500,height=500):
    img = image.resize((width, height))
    return img

# API Gateway endpoint URL
api_url = 'INSERT_YOUR_API_GATEWAY_ENDPOINT_URL_HERE'
# Create two columns for layout
left_column, right_column = st.columns(2)
# ===========
with left_column:
    # Define Streamlit UI elements
    st.title('Stable Diffusion XL Image Generation with AWS Inferentia')

    prompt_one = st.text_area("Enter your prompt:", 
                        f"Raccoon astronaut in space, sci-fi, future, cold color palette, muted colors, detailed, 8k")

    # Number of inference steps
    num_inference_steps_one = st.slider("Number of Inference Steps", 
                                min_value=1, 
                                max_value=100, 
                                value=30, 
                                help="More steps might improve quality, with diminishing marginal returns. 30-50 seems best, but your mileage may vary.")

    # Create an expandable section for optional parameters
    with st.expander("Optional Parameters"):
        # Random seed input
        seed_one = st.number_input("Random seed", 
                            value=555, 
                            help="Set to the same value to generate the same image if other inputs are the same, change to generate a different image for same inputs.")

        # Negative prompt input
        negative_prompt_one = st.text_area("Enter your negative prompt:", 
                        "cartoon, graphic, text, painting, crayon, graphite, abstract glitch, blurry")

    





if st.button('Generate Image'):
    with st.spinner(f'Generating Image with {num_inference_steps_one} iterations'):
        with right_column:
            start_time = time.time()
            # ===============
            # Example input data
            prompt_input_one = {
                "prompt": prompt_one,
                "parameters": {
                    "num_inference_steps": num_inference_steps_one,
                    "seed": seed_one,
                    "negative_prompt": negative_prompt_one
                }
            }

            # Make API request
            response_one = requests.post(api_url, json=prompt_input_one)

            # Process and display the response
            if response_one.status_code == 200:
                result_one = response_one.json()
                # st.success(f"Prediction result: {result}")
                image_one = display_image(decode_base64_image(result_one["generated_images"][0]))
                st.image(image_one, 
                    caption=f"{prompt_one}") 
                end_time = time.time()
                total_time = round(end_time - start_time, 2)
                st.text(f"Prompt: {prompt_one}")
                st.text(f"Number of Iterations: {num_inference_steps_one}")
                st.text(f"Random Seed: {seed_one}")
                st.text(f'Total time taken: {total_time} seconds')
                # Calculate and display the time per iteration in milliseconds
                time_per_iteration_ms = (total_time / num_inference_steps_one)
                st.text(f'Time per iteration: {time_per_iteration_ms:.2f} seconds')
            else:
                st.error(f"Error: {response_one.text}")
```

            """
        )