Inference speed is crucial for our applications, and we aim to enhance the performance of our T2I and I2I pipelines. Previous optimizations like DeepCache have improved inference times but at the expense of image quality. We are interested in exploring whether another optimization, OneDiff, can provide better results. As we are currently focused on the core AI subnet software, we are inviting community builders to conduct this research in exchange for a software bounty. If OneDiff proves effective, we will create another bounty to implement this optimization in the T2I and I2I pipelines. The successful researcher will have the first opportunity to take on this new bounty, allowing them to implement this potentially valuable feature for the community and accelerate our pipelines ⚡.
Bounty Tips
Avoid testing DeepCache directly on the livepeer/ai-worker repository, as its complexity may hinder your process. Instead, create a simple inference Python script for your tests.
Requirements
The submission must include a dedicated report outlining the inference speedup and output quality with and without the OneDiff optimization.
The report should cover each of the specified models.
Inference times should be averaged over 10 runs, and comparison images for each run should be provided.
How to Apply
Express Interest: Comment on the bounty GitHub issue to express your interest.
Review Process: We will review submissions within 5 days and select the best candidate.
Assignment: If selected, you will be assigned to the GitHub issue.
Implementation: Implement the work and seek guidance in the #🛋│developer-lounge channel on our Discord server.
Pull Request: Create a pull request for your work and request a review from a team member.
Bounty Payment: Once your pull request is reviewed and approved, we will contact you to arrange the bounty payment.
Thank you for your interest in contributing to our project 💛!
Overview
Inference speed is crucial for our applications, and we aim to enhance the performance of our T2I and I2I pipelines. Previous optimizations like DeepCache have improved inference times but at the expense of image quality. We are interested in exploring whether another optimization, OneDiff, can provide better results. As we are currently focused on the core AI subnet software, we are inviting community builders to conduct this research in exchange for a software bounty. If OneDiff proves effective, we will create another bounty to implement this optimization in the T2I and I2I pipelines. The successful researcher will have the first opportunity to take on this new bounty, allowing them to implement this potentially valuable feature for the community and accelerate our pipelines ⚡.
Bounty Tips
Avoid testing DeepCache directly on the livepeer/ai-worker repository, as its complexity may hinder your process. Instead, create a simple inference Python script for your tests.
Requirements
How to Apply
Thank you for your interest in contributing to our project 💛!