Papers › BAD: Bidirectional Auto-regressive Diffusion for Text-to-Motion Generation

BAD: Bidirectional Auto-regressive Diffusion for Text-to-Motion Generation

17 Sep 2024arXiv:2409.10847archive 2025-07-28

S. Rohollah Hosseyni, Ali Ahmad Rahmani, S. Jamal Seyedmohammadi, Sanaz Seyedin, Arash Mohammadi

Autoregressive models excel in modeling sequential dependencies by enforcing causal constraints, yet they struggle to capture complex bidirectional patterns due to their unidirectional nature. In contrast, mask-based models leverage bidirectional context, enabling richer dependency modeling. However, they often assume token independence during prediction, which undermines the modeling of sequential dependencies. Additionally, the corruption of sequences through masking or absorption can introduce unnatural distortions, complicating the learning process. To address these issues, we propose Bidirectional Autoregressive Diffusion (BAD), a novel approach that unifies the strengths of autoregressive and mask-based generative models. BAD utilizes a permutation-based corruption technique that preserves the natural sequence structure while enforcing causal dependencies through randomized ordering, enabling the effective capture of both sequential and bidirectional relationships. Comprehensive experiments show that BAD outperforms autoregressive and mask-based models in text-to-motion generation, suggesting a novel pre-training strategy for sequence modeling. The codebase for BAD is available on https://github.com/RohollahHS/BAD.

PaperPDFCode

In Syntology Open this paper in Syntology's Atlas, the map of the papers in Syntology's graph and their citations.

Code

rohollahhs/bad officialmentioned in papermentioned on GitHubpytorch report

Repository list and official/mentioned flags are the archive's, frozen 2025-07-28. Reachability, where shown, is from one Syntology probe window (2026-09-16 to 2026-09-18); repositories not probed show nothing. GitHub stars are not tracked.

Code Syntology ran Syntology

Not run by Syntology. Nothing on this page verifies that the listed code works.

Tasks

Human motion predictionMotion ForecastingMotion GenerationMotion Synthesis

Results from the paper archive 2025-07-28

TaskDatasetModelMetricValueRank at snapshotLeaderboardReport
Motion Synthesis HumanML3D BAD (CBS) Diversity 9.688 #6 of 37 Archive leaderboard report
Motion Synthesis HumanML3D BAD (CBS) FID 0.049 #6 of 37 Archive leaderboard report
Motion Synthesis HumanML3D BAD (CBS) Multimodality 1.119 #6 of 37 Archive leaderboard report
Motion Synthesis HumanML3D BAD (CBS) R Precision Top3 0.800 #6 of 37 Archive leaderboard report
Motion Synthesis HumanML3D BAD (OAAS) Diversity 9.694 #10 of 37 Archive leaderboard report
Motion Synthesis HumanML3D BAD (OAAS) FID 0.065 #10 of 37 Archive leaderboard report
Motion Synthesis HumanML3D BAD (OAAS) Multimodality 1.194 #10 of 37 Archive leaderboard report
Motion Synthesis HumanML3D BAD (OAAS) R Precision Top3 0.808 #10 of 37 Archive leaderboard report
Motion Synthesis KIT Motion-Language BAD (OAAS) Diversity 11.000 #9 of 31 Archive leaderboard report
Motion Synthesis KIT Motion-Language BAD (OAAS) FID 0.221 #9 of 31 Archive leaderboard report
Motion Synthesis KIT Motion-Language BAD (OAAS) Multimodality 1.170 #9 of 31 Archive leaderboard report
Motion Synthesis KIT Motion-Language BAD (OAAS) R Precision Top3 0.750 #9 of 31 Archive leaderboard report

Ranks are positions in the archive's leaderboards as they stood at the 2025-07-28 snapshot. Results published since then are not among these rows, so a rank here is not a current standing.

Methods

Diffusion

Report a problem or propose a change · a person checks every report against the paper or source before anything changes; decisions are listed on /corrections