Research map / Image generation / Text-to-image diffusion
SymbOmni: Evolving Agentic Omni Models via Symbolic Concept Learning
Jinxiu Liu, Jianru Li, Tanqing Kuang, Xuanming Liu, Kangfu Mei, Yandong Wen, Weiyang Liu
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (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 symbolic concept memory retrieval (vs w/o EXP) (ablation) | SymbOmni · ReasonEdit | total tokens per case: 21,727.1 -> 14,361.4 | Improved |
| add symbolic concept memory retrieval (ablation) | SymbOmni · ReasonEdit | API requests per case: 9.85 -> 7.25 | Improved |
| add experience-driven evolution (ablation) | SymbOmni · GenEval2 Soft-TIFA | Overall: 71.06 -> 77.34 | Improved |
| replace Gemini 2.5 Flash with open-source Qwen3-VL + DeepSeek (ablation) | SymbOmni · ComfyBench Creative | Resolve rate: 67.5 -> 55.0 | Hurt |
| replace Gemini 2.5 Flash with open-source Qwen3-VL + DeepSeek (ablation) | SymbOmni · ComfyBench total | Resolve rate: 86.0 -> 82.0 | Hurt |
| learn concepts from out-of-domain workflows | SymbOmni · ComfyBench total | Resolve rate: 67.0 -> 75.0 | Improved |
| use offline symbolic memory instead of workflow descriptions | SymbOmni · ComfyBench Complex offline | Resolved rate: 66.7 -> 70.0 | Improved |
| use SymbOmni vs ComfyMind | SymbOmni · ComfyBench Creative | Resolve rate: 42.5 -> 67.5 | Improved |
| swap in Nano Banana as generator (ablation) | SymbOmni · ComfyBench total | Resolve rate: 86.0 -> 91.0 | Improved |
| SymbOmni vs Nano Banana on Left-Right and Color subcategories | SymbOmni · ReasonEdit | score | Mixed |
Closest papers by meaning
- Gen-Searcher: Reinforcing Agentic Search for Image Generation (NeurIPS 2026 · similarity 0.87)
- SEAgent: Self-Evolving Computer Use Agent with Autonomous Learning from Experience (ICML 2026 · similarity 0.86)
- DeepEyesV2: Toward Agentic Multimodal Model (ICLR 2026 · similarity 0.86)
- VLM Agents Generate Their Own Memories: Distilling Experience into Embodied Programs of Thought (NeurIPS 2024 · similarity 0.86)
- ATLAS: Agentic or Latent Visual Reasoning? One Word is Enough for Both (NeurIPS 2026 · similarity 0.86)
- ViLoMem: Agentic Learner with Grow-and-Refine Multimodal Semantic Memory (CVPR 2026 · 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.
