Research map / Vision-language and multimodal / Robot action and navigation
SILG: The Multi-domain Symbolic Interactive Language Grounding Benchmark
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 |
|---|---|---|---|
| add recurrent state-tracking LSTM (ablation) | SIR · RTFM | success rate: 88.8 (22.4) -> 99.2 (0.7) | Helped |
| add recurrent state-tracking LSTM (ablation) | SIR · Messenger | success rate | No clear effect |
| add recurrent state-tracking LSTM (ablation) | SIR · ALFWorld new inst | success rate: 21.0 (1.5) -> 23.6 (2.8) | Helped |
| add recurrent state-tracking LSTM (ablation) | SIR · SymTD | success rate: 9.7 (1.3) -> 14.9 (1.8) | Helped |
| add entity-centric local convolution (ablation) | SIR · SILGNetHack | success rate: 23.8 (0.8) -> 25.4 (3.3) | Helped |
| add entity-centric local convolution (ablation) | SIR · RTFM, Messenger, ALFWorld, SymTD | success rate | No clear effect |
| add entity-centric attention (ablation) | SIR · RTFM, SILGNetHack, SymTD | success rate | Hurt |
| add entity-centric attention (ablation) | SIR · Messenger | success rate | Helped |
| add frozen BERT text encoding (ablation) | SIR · Messenger | success rate: 0 (0) -> 31 (2.6) | Helped |
| add frozen BERT text encoding (ablation) | SIR · RTFM, SILGNetHack | success rate | No clear effect |
| concatenate text fields into one string instead of separate fields (ablation) | SIR · RTFM | success rate | Hurt |
| use Masked-RCNN detections instead of oracle detector for ALFWorld transfer (ablation) | SIR base · ALFRED new inst | success rate: 11.2 (3.3) -> 3.0 (1.7) | Hurt |
Closest papers by meaning
- GroundingME: Exposing the Visual Grounding Gap in MLLMs through Multi-Dimensional Evaluation (CVPR 2026 · similarity 0.90)
- GeM-VG: Towards Generalized Multi-image Visual Grounding with Multimodal Large Language Models (AAAI 2026 · similarity 0.88)
- Grounding Multimodal Large Language Models to the World (ICLR 2024 · similarity 0.87)
- iVGR: Internalizing Visually Grounded Reasoning for MLLMs with Reinforcement Learning (ICML 2026 · similarity 0.87)
- Towards GUI Agents: Vision-Language Diffusion Models for GUI Grounding (CVPR 2026 · similarity 0.87)
- Grounding Multimodal Large Language Models in Actions (NeurIPS 2024 · similarity 0.87)
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.