Skip to content

Commit

Permalink
added genre and story style input params
Browse files Browse the repository at this point in the history
  • Loading branch information
Sunil Surendra Singh committed Oct 4, 2023
1 parent 65f6ddd commit 3253a7e
Show file tree
Hide file tree
Showing 6 changed files with 140 additions and 46 deletions.
4 changes: 3 additions & 1 deletion README.md
View file
Original file line number Diff line number Diff line change
Expand Up @@ -30,14 +30,16 @@ BLIP Image2Text model details can be found [here](https://huggingface.co/Sof22/i

* It's important to note that this sample demonstration app is hosted on the free tiers of Huggingface Spaces, which means it is functional but may exhibit slower performance.
* Additionally, when using the app for the first time or after an extended period (more than 1 hour), you might encounter an "Internal Error" message or receive a story unrelated to the provided image. This is a normal occurrence during the model loading process. Please wait a few seconds and try again; it should function as intended.
* Please be aware that due to cost and resource constraints, the app currently has a maximum story length limit of 100 words per request.
* Please be aware that due to cost and resource constraints, the app currently has a maximum story length limit of 200 words per request.

App UI is shown below:

<img src="https://github.com/sssingh/pic-to-story/blob/main/assets/story-teller-app.png?raw=true" width="1000" height="450"/><br><br>

**Dark Mode Toggle**: Activate it to switch between dark and light mode.
**Image Selector**: Click on it to pick an image from your computer, or drag and drop an image onto it directly. Click the 'X' to clear the selection and resets the app.
**Story Genre Dropdown**: Select the desired story genre from the dropdown list.
**Story Writing Style Dropdown**: Select the desired story writing style from the dropdown list.
**Story Length (in words) Slider**: Adjust the slider to specify the desired length of the generated story.
**Creativity Index Slider**: Modify the slider to indicate the desired level of creativity for the generated story. A range between 0.5 and 0.7 is recommended. Setting it to 1.0 results in highly creative, sometimes amusing output.
**Generate Story Button**: Press this button to initiate the story generation process.
Expand Down
115 changes: 82 additions & 33 deletions app.py
View file
Original file line number Diff line number Diff line change
Expand Up @@ -30,21 +30,25 @@ def create_interface():
api_name=False,
)
with gr.Row():
with gr.Column(scale=5):
with gr.Column():
gr.Markdown(
"""
# Storyteller
# The Storyteller
**This app can craft captivating narratives from captivating images,
potentially surpassing even Shakespearean standards. Select an image
that inspires a story, choose a story length (up to 100 words), and
adjust the creativity index to enhance its creative flair.**
potentially surpassing even Shakespearean standards.
<br>
Select an `Image` that inspires a story, choose a `Story Genre`,
`Story Writing Style`, `Story Length (up to 200 words)`, and
adjust the `Creativity Index` to enhance its creative flair. Then
hit `Generate Story` button.
Alternatively, just select one the pre-configured `Examples`**
<br>
***Please exercise patience, as the models employed are extensive and may
require a few seconds to load. If you encounter an unrelated story,
it is likely still loading; wait a moment and try again.***
***Please exercise patience, as the models employed are extensive
and may require a few seconds to load. If you encounter an unrelated
story, it is likely still loading; wait a moment and try again.***
"""
)
with gr.Column(scale=2):
with gr.Column():
max_count = gr.Textbox(
label="Max allowed OpenAI requests:",
value=app_config.openai_max_access_count,
Expand All @@ -63,21 +67,34 @@ def create_interface():
image = gr.Image(
type="filepath",
)
# Word Count Slider
word_count = gr.Slider(
label="Story Length (words):",
minimum=25,
maximum=100,
value=50,
step=5,
)
creativity = gr.Slider(
label="Creativity Index:",
minimum=0.3,
maximum=1.0,
value=0.7,
step=0.1,
)
with gr.Row():
with gr.Column():
genre = gr.Dropdown(
label="Story Genre: ",
value="Poetry",
choices=app_config.genre,
)
style = gr.Dropdown(
label="Story Writing Style:",
value="Cinematic",
choices=app_config.writing_style_list,
)
with gr.Column():
# Word Count Slider
word_count = gr.Slider(
label="Story Length (words):",
minimum=30,
maximum=200,
value=50,
step=10,
)
creativity = gr.Slider(
label="Creativity Index:",
minimum=0.3,
maximum=1.0,
value=0.7,
step=0.1,
)
with gr.Row():
submit_button = gr.Button(
value="Generate Story", elem_classes="orange-button"
Expand All @@ -87,24 +104,56 @@ def create_interface():
story = gr.Textbox(
label="Story:",
placeholder="Generated story will appear here.",
lines=12,
lines=21,
)
with gr.Row():
with gr.Accordion("Expand for examples:", open=False):
gr.Examples(
examples=[
["assets/examples/cheetah-deer.jpg", 50, 0.5],
["assets/examples/man-child-pet-dog.jpg", 100, 0.6],
["assets/examples/man-child.jpeg", 60, 1.0],
["assets/examples/men-fighting.jpg", 50, 0.4],
["assets/examples/teacher-school.jpg", 80, 0.7],
[
"assets/examples/cheetah-deer.jpg",
"Horror",
"Narrative",
80,
0.5,
],
[
"assets/examples/man-child-pet-dog.jpg",
"Fiction",
"Formal",
100,
0.6,
],
[
"assets/examples/man-child.jpeg",
"Children Literature",
"Symbolic",
120,
1.0,
],
[
"assets/examples/men-fighting.jpg",
"Comedy",
"Experimental",
60,
0.4,
],
[
"assets/examples/teacher-school.jpg",
"Surrealism",
"Non-linear",
80,
0.7,
],
],
inputs=[image, word_count, creativity],
outputs=[story],
fn=model.generate_story,
inputs=[image, genre, style, word_count, creativity],
outputs=[story, max_count, curr_count, available_count],
run_on_click=True,
)
submit_button.click(
fn=model.generate_story,
inputs=[image, word_count, creativity],
inputs=[image, genre, style, word_count, creativity],
outputs=[story, max_count, curr_count, available_count],
)
clear_button.click(
Expand Down
4 changes: 2 additions & 2 deletions assets/story-teller-app.png
View file
Loading
Sorry, something went wrong. Reload?
Sorry, we cannot display this file.
Sorry, this file is invalid so it cannot be displayed.
4 changes: 2 additions & 2 deletions assets/story-teller-examples.png
View file
Loading
Sorry, something went wrong. Reload?
Sorry, we cannot display this file.
Sorry, this file is invalid so it cannot be displayed.
35 changes: 35 additions & 0 deletions config.py
View file
Original file line number Diff line number Diff line change
Expand Up @@ -17,6 +17,41 @@ class AppConfig:
OPENAI_KEY = os.getenv("OPENAI_KEY")
I2T_API_URL = os.getenv("I2T_API_URL")
MONGO_CONN_STR = os.getenv("MONGO_CONN_STR")
genre_list = genre = [
"Adventure",
"Children Literature",
"Comedy",
"Drama",
"Fantasy",
"Fiction",
"Horror",
"Mystery",
"Non-fiction",
"Poetry",
"Romance",
"Satire",
"Surrealism",
"Urban Fantasy",
]
writing_style_list = [
"Cinematic",
"Conversational",
"Descriptive",
"Experimental",
"First-Person",
"Formal",
"Informal",
"Metaphorical",
"Minimalist",
"Narrative",
"Non-linear",
"Objective",
"Sensory",
"Stream of Consciousness",
"Symbolic",
"Third-Person Limited",
"Third-Person Omniscient",
]


app_config = AppConfig()
24 changes: 16 additions & 8 deletions model.py
View file
Original file line number Diff line number Diff line change
Expand Up @@ -22,7 +22,7 @@ def __image2text(image):
return response


def __text2story(image_desc, word_count, creativity):
def __text2story(image_desc, genre, style, word_count, creativity):
""" "Generates a short story based on image description text prompt"""
## chat LLM model
story_model = ChatOpenAI(
Expand All @@ -32,25 +32,29 @@ def __text2story(image_desc, word_count, creativity):
)
## chat message prompts
sys_prompt = PromptTemplate(
template="You are an expert storyteller that can generate"
+ " imaginative {word_count} words long story from the input text",
input_variables=["word_count"],
template="""You are an expert story writer, write a maximum of {word_count}
words long story in {genre} genre in {style} writing style, based on the user
provided story-context.
""",
input_variables=["word_count", "genre", "style"],
)
system_msg_prompt = SystemMessagePromptTemplate(prompt=sys_prompt)
human_prompt = PromptTemplate(
template="{image_desc}", input_variables=["image_desc"]
template="story-context: {context}", input_variables=["context"]
)
human_msg_prompt = HumanMessagePromptTemplate(prompt=human_prompt)
chat_prompt = ChatPromptTemplate.from_messages(
[system_msg_prompt, human_msg_prompt]
)
## LLM chain
story_chain = LLMChain(llm=story_model, prompt=chat_prompt)
response = story_chain.run(image_desc=image_desc, word_count=word_count)
response = story_chain.run(
genre=genre, style=style, word_count=word_count, context=image_desc
)
return response


def generate_story(image_file, word_count, creativity):
def generate_story(image_file, genre, style, word_count, creativity):
"""Generates a story given an image"""
# read image as bytes arrayS
with open(image_file, "rb") as f:
Expand All @@ -59,7 +63,11 @@ def generate_story(image_file, word_count, creativity):
image_desc = __image2text(image=input_image)
# generate story from caption
story = __text2story(
image_desc=image_desc, word_count=word_count, creativity=creativity
image_desc=image_desc,
genre=genre,
style=style,
word_count=word_count,
creativity=creativity,
)
# increment the openai access counter and compute count stats
mongo.increment_curr_access_count()
Expand Down

0 comments on commit 3253a7e

Please sign in to comment.