47 repos across 2 sub-areas
Techniques and systems for capturing, representing, and synthesizing 3D human body pose, motion, and shape from images and video. Repositories in this cluster cover foundational topics like pose estimation, SMPL body model fitting, diffusion-based motion generation, and neural rendering approaches (including gaussian splatting variants). The cluster reflects both research implementations and educational resources aimed at understanding embodied AI and computer vision for human-centric 3D reconstruction.
Cluster 639426
31 repos
Large Language Model Fine-Tuning & Optimization
16 repos
Deep learning techniques for adapting and improving large language models through methods like QLoRA, DPO alignment, RLHF, and RAG pipelines. These repositories provide implementations, tutorials, and frameworks for fine-tuning LLMs efficiently on consumer hardware, optimizing model behavior through reinforcement learning and preference-based training, and augmenting models with retrieval systems. Covers both theoretical foundations in attention mechanisms and practical applied examples using PyTorch.