Paper discussion on 3 November - High-resolution image synthesis with latent diffusion models This update is for participants only
Stable Diffusion is a deep learning, text-to-image model, trained and released as open source by Stability AI. Stable Diffiusion has gained traction as the preferred text-to-image model to deep dive both into practical and technical aspects of this technology.
By attending this hands-on workshop, you will:
- Create your own generative AI art using popular web based tools.
- Deploy an application to production using python.
- Learn fundamental concepts and techniques used in text-to-image models, including deep learning algorithms, neural networks, and image processing.
If you are interested in beginning your journey towards mastering text-to-image models like Stable Diffusion, and understanding the underlying technology of how it generates digital art, this workshop is the perfect opportunity for you.
This is a two-day workshop with four 90-min session each day. You can buy a ticket to attend either both days of the workshop, or just the first day of the workshop.
Session wise break-up of both days is published on the schedule.
Following topics will be covered:
- Prompt engineering.
- Advanced Prompt Engineering (anatomy, keywords, negative prompts, attention weights).
- Parameter tuning (CFG, Quality, models, seed, schedulers).
- Img2img (in-painting, out-painting, masking, object removal, instruct pix2pix, upscaling, face fixes).
- ControlNet (controlling poses, lighting, hands).
- Model finetuning techniques theory (Textual Inversion, Dreambooth, LORA).
- Hands on model finetuning using LORA.
- Model merging.
- Stable Diffusion deep learning model components.
- Image pipeline using Diffusers.
- Deployment using Replicate.
- Applications (fantasy profile pics, discord/whatsapp stickers, t-shirt designs, website designs, personalized memes, model photoshoot).
If you are a technologist or an entrepreneur, looking to take first steps in Generative AI space or deep dive into the underlying technology, this workshop is a perfect opportunity for you.
The workshop is structured such that on the first day, there will be a deep dive of text2image technology without going in too much detail of its underlying components. Participants will use advance yet intutive tools to explore the potential of this technology.
You do not need a strong technical background to participate actively for the first day of the workshop. Hence, you can attend only the first day of the workshop. The instructor will provide source material and easy follow through instructions for participants to be actively engaged in the workshop.
For the second day of the workshop, participants must have hands-on technical experience in coding. Participants do not need to have a deep understanding of Machine Learning or Python. The instructor will use intutive SaaS providers for deploying AI models in a serverless way using templated code.
This is a hands-on workshop. Participants must have a working laptop, and a reliable and fast internet connection.
Resource material will be shared on Google Drive.
Installations will be on the cloud. No local installation is required, though a local installation of your preferred code editor like VS Code is recommended.
The pricing of the workshop includes access to Colab Pro and Replicate API keys (around ~$20 = 1600 INR).
Amir Nagri is a geek with 17+ years of experience in Technology. Over the past decade, he failed as a startup entrepreneur, busted all his savings, went back industry to lead and build Gojek (now GoTo), Superapp of Indonesia, as Head of Mobile Engineering. In GoTo, he played various roles and gained experience in Mobile, Data Engineering, DevOps, Distributed Systems etc., scaling systems from 20k bookings per day to 5M bookings per day.
For past 2 years, he is on his second stint as an entrepreneur, exploring and building products using technologies like NLP, Large Language Models, Text to Image models. Apart from sharing his insights and learning on Twitter, he is full-time building his Generative AI startup YouDazzl.
This is a paid workshop, and will be held online, via Zoom.
For tickets and inquiries about this workshop, contact Hasgeek at firstname.lastname@example.org or call (91)7676332020.