awesome-multi-task-learning

Why we included this project

Multi-task learning has a scattered literature, with methods spread across vision, NLP, and reinforcement learning papers, and this list does the work of pulling it together. It collects surveys, benchmark datasets, and open-source codebases into sections that follow how the field actually organizes itself: architecture, optimization, task relationship learning, and so on. The editors flag toy datasets and mark the surveys they consider most influential, which saves you from wasting time on a benchmark that will not tell you much. Since entries link straight to the papers and repositories instead of summarizing them, the list works as a map for planning a literature review or picking a baseline codebase to start from. Someone new to MTL can use it to get oriented, and someone who has been in the field a while can scan it to make sure they have not missed a relevant method.