Research map / Vision-language and multimodal / Robot action and navigation
Behavior Transformers: Cloning $k$ modes with one stone
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| Use BeT rather than LWR | behavior cloning on CARLA · CARLA | driving success: 1 vs. 0.98 | No clear effect |
| Use BeT rather than IBC | behavior cloning on Block push · Block push | push two blocks (P2): 0 vs. 0.71 | Helped |
| Use BeT rather than IBC | behavior cloning on Kitchen · Kitchen | probability of completing 4 tasks: 0.24 -> 0.44 | Helped |
| Add residual action offsets (ablation) | BeT · Kitchen | relative task success: 0.78 relative success without offsets | Helped |
| Add discrete action binning (ablation) | BeT · Block push | relative task success: 0.25 relative success without binning | Helped |
| Add historical context (ablation) | BeT · CARLA | relative task success: 0.65 relative success without history | Helped |
| Use MinGPT instead of MLP (ablation) | BeT · Block push | relative task success: 0 vs. full BeT | Helped |
| Use MinGPT instead of temporal convolution (ablation) | BeT · Block push | relative task success: 0.01 vs. full BeT | Helped |
| Use MinGPT instead of LSTM (ablation) | BeT · CARLA | relative task success: 0.03 vs. full BeT | Helped |
| Use k-means bins instead of MDN (ablation) | BeT · CARLA | relative task success: 0.30 with GPT-MDN | Helped |
| Use k-means bins instead of uniform quantization (ablation) | BeT · CARLA | relative task success: 0.90 with uniform quantization | Helped |
Closest papers by meaning
- Behavior Generation with Latent Actions (ICML 2024 · similarity 0.90)
- Is Behavior Cloning All You Need? Understanding Horizon in Imitation Learning (NeurIPS 2024 · similarity 0.87)
- Rethinking Latent Redundancy in Behavior Cloning: An Information Bottleneck Approach for Robot Manipulation (ICML 2025 · similarity 0.87)
- Self-Improving Embodied Foundation Models (NeurIPS 2025 · similarity 0.87)
- Videos are Sample-Efficient Supervisions: Behavior Cloning from Videos via Latent Representations (NeurIPS 2025 · similarity 0.87)
- How to Leverage Diverse Demonstrations in Offline Imitation Learning (ICML 2024 · similarity 0.86)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.