Quantization of Diffusion model. -- 2
Budget: ₹37,500 – ₹40,000 INR
I am looking for a skilled deep learning researcher/engineer to collaborate on a project. I have a foundational paper on the quantization of diffusion models, along with its corresponding codebase, and I've developed a new idea for a hierarchical, bi-policy quantization system that needs to be implemented. The ultimate goal is to generate results suitable for a top-tier academic publication.
I will provide a clear and comprehensive document outlining the new idea, including all the theoretical explanations and logic required for implementation. The project requires a strong background in deep learning, particularly with diffusion models and model compression techniques.
Key Requirements and Deliverables:
•Implementation: You will be responsible for implementing the new framework, which involves a two-stage optimization pipeline, a lightweight semantic analysis network, and a dynamic bit modulation mechanism. The core technical stack is PyTorch and CUDA.
•Experimental Validation: The work must include a comprehensive evaluation pipeline. You will need to calculate and report various quantitative metrics, including FID, IS, LPIPS, and KID scores across datasets like CIFAR-10, ImageNet, CelebA-HQ, and FFHQ.
•Performance Analysis: A crucial part of the project is to measure and document efficiency metrics such as model size reduction, inference speedup, and memory usage analysis.
•Comparative Analysis: The final implementation must be able to perform a head-to-head comparison against 5+ state-of-the-art baselines.
•Reproducibility: All experiments must be documented for reproducibility. You will need to provide reusable scripts or notebooks that allow for easy rerunning of the entire pipeline, including exact hyperparameters and environment specifications.
I will provide all necessary paper PDFs, code repositories, and datasets (if available with me).
I will provide a clear and comprehensive document outlining the new idea, including all the theoretical explanations and logic required for implementation. The project requires a strong background in deep learning, particularly with diffusion models and model compression techniques.
Key Requirements and Deliverables:
•Implementation: You will be responsible for implementing the new framework, which involves a two-stage optimization pipeline, a lightweight semantic analysis network, and a dynamic bit modulation mechanism. The core technical stack is PyTorch and CUDA.
•Experimental Validation: The work must include a comprehensive evaluation pipeline. You will need to calculate and report various quantitative metrics, including FID, IS, LPIPS, and KID scores across datasets like CIFAR-10, ImageNet, CelebA-HQ, and FFHQ.
•Performance Analysis: A crucial part of the project is to measure and document efficiency metrics such as model size reduction, inference speedup, and memory usage analysis.
•Comparative Analysis: The final implementation must be able to perform a head-to-head comparison against 5+ state-of-the-art baselines.
•Reproducibility: All experiments must be documented for reproducibility. You will need to provide reusable scripts or notebooks that allow for easy rerunning of the entire pipeline, including exact hyperparameters and environment specifications.
I will provide all necessary paper PDFs, code repositories, and datasets (if available with me).
Related categories:
Python
CUDA
Machine Learning (ML)
Pytorch
Deep Learning
Machine Learning Algorithms