Research map / Image generation / Text-to-image diffusion
GenCtrl -- A Formal Controllability Toolkit for Generative Models
Emily Cheng, Carmen Amo Alonso, Federico Danieli, Arno Blaas, Luca Zappella, Pau Rodriguez, Xavier Suau
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 5-shot examples instead of 0-shot prompting (ablation) | Qwen3-4B and Gemma3-4B · Formality dialogue, 5 turns | controllability: reach full controllability by t=5 vs none fully controllable at 0-shot | Improved |
| add 5-shot examples (ablation) | SmolLM3-3B · Formality dialogue, 5 turns | controllability: not controllable | No clear effect |
| scale Qwen3 from 0.6B to 14B | Qwen3 · Formality (0-shot, 5 turns) | controllability and calibration: calibration saturates at 4B; MAE approx 0.25 at 14B | Improved |
| scale Qwen3 sizes | Qwen3 · Formality (0-shot, 1 turn) | correlation R: R plateaus around 8B | Mixed |
| add stochastic sampling instead of greedy decoding | LLMs · Even/odd, formality, word length | faithfulness: worse faithfulness metrics | Hurt |
| switch task from formality to even/odd | Gemma3-4B · controllable fraction | controllability: not fully controllable on even/odd | Hurt |
| add more dialogue turns | LLMs · Formality 0-shot | controllable set: controllable set grows within 5 steps | Improved |
Closest papers by meaning
- Controllable User Simulation (NeurIPS 2026 · similarity 0.87)
- Adaptable Logical Control for Large Language Models (NeurIPS 2024 · similarity 0.86)
- Teaching Models to Improve on Tape (AAAI 2025 · similarity 0.86)
- Controlling Language and Diffusion Models by Transporting Activations (ICLR 2025 · similarity 0.86)
- Test-Time Alignment for Large Language Models via Textual Model Predictive Control (ICLR 2026 · similarity 0.85)
- TRACE Back from the Future: A Probabilistic Reasoning Approach to Controllable Language Generation (ICML 2025 · similarity 0.85)
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.
