Mind Crafting – Unveiling the Power of Open-Source AI Models
05.10.2023 | 4 min read
In the ever-evolving landscape of technology, the realm of artificial intelligence continues to transform industries. Recently, as part of our knowledge-sharing initiative during Mind Crafting week, we had the privilege of delving into the world of open-source AI models, led by Jakub Bilko, a Senior JS Developer at 10Clouds. In this blog post, we share with you what we’ve learnt from his session on open-source AI models, with a particular focus on Hugging Face.
Catching the perfect match
With over 300,000 AI models available, it could be a neck-breaking challenge - but not for Jakub.
Following his vision of “Image to Horror” app Jakub reached a platform called Hugging Face that has come a long way since its inception in 2017 as a … chatbot for teenagers! Over time, it has evolved into a thriving hub for hosting, training, and collaborative development of AI models. provides a gateway to a vast repository of AI models, datasets, and projects, opening doors to endless possibilities. That’s the one Jakub went for.
A case in point – The 'Image to Horror' App
Creating an app from scratch is a fascinating and challenging thing. And using a variety of combinations of programming and AI models is still a terra incognita that motivates and enables us to push the boundaries of what’s possible in programming.
So what exactly is an 'Image to Horror' app about?
This AI image app generates a vivid description of a selected stock image and proceeds to craft a horror story based on that description. To create such an app from scratch, Jakub guided us through the process within Hugging Face:
2. Image description: Write a function that takes the image file and generates a textual description of its content.
3. Voice generation: To add voice narration to your horror story, you can use the 'Get Voiceover' function. Alternatively, you can employ 'Text-to-Speech' to bring your story to life.
4. Story generation: To craft the horror story, utilize ChatGPT. Create a file and define a prompt that instructs ChatGPT to generate a short horror story based on the image description.
5. Seamless communication: Utilize Lanchain to establish communication with the OpenAI object and create a function that takes the image description and generates the story.
Jakub emphasized that Hugging Face streamlines the process of building such apps, eliminating the need for advanced coding skills (which he was looking for!). While Python boasts a vast library landscape, Hugging Face stands out for its user-friendly approach and accessibility.
Accessibility and scalability
What’s in it for you? The great notable advantage of such platforms is its ease of use, making it accessible to a broader audience and enabling them to create something new on their own. Like your own app for instance!
It simplifies the integration of AI models into applications, opening doors for creativity and innovation. However, it's worth noting that while some programs enable you to set up models locally, Hugging Face primarily operates in a cloud-based environment.
Stay tuned for more…
Stay tuned for more insightful sessions and discoveries from 10Clouds’ MindCrafting Week!