Semantic Blending

Follow this topic Ask about this topic

Momentum

3 papers in the last four weeks, against 2 the four weeks before. 0.0% of all new papers.

Jul 6Week of Sep 21

Latest papers 18

All topics

Ready2Blend: From Natural-Language Instructions to Composable Alignment Prompts
Sep 30, 2026Jeesu Jung, Hwan Chang, Juseon Do +5Value Alignment Large Language Model Adaptation
Korea Advanced Institute of Science and Technology · Samsung SDS
MeRoTune: RoPE-Safe Merging with a Tunable Dial
Sep 7, 2026Salman FarozRotary Position Embedding Soft-Token Representations
From Monolithic Blending to Agentic Orchestration: Dynamic Response for Conversational Assistants at Scale
Sep 4, 2026Cen Mia Zhao, Peng Wang, Chuan Shi +6Orchestration Responses
Airbnb, Inc.
Sequential Trajectories and Simultaneous Blending: Multi-Emotion Modeling for Instruction-Following TTS
Aug 31, 2026Yan Zhou, Yun Hong, Yang FengEmotion Recognition Emotion
Key Laboratory of Intelligent Information Processing, Institute of Computing Technology, Chinese Academy of Sciences (ICT/CAS) · State Key Laboratory of AI Safety, Institute of Computing Technology, Chinese Academy of Sciences · University of Chinese Academy of Sciences, Beijing, China
Wiring Beats Blending: Structure-Aware Compensation for Transformer Downscaling
Aug 3, 2026Ravi Satya Durga Prasad YenugulaModel Size Transformer Architectures
Independent Researcher
DeGS: A Scalable 3DGS Architecture via Decoupled Workload Parsing and Reorganization
Aug 3, 2026Minnan Pei, Gang Li, Zeyu Zhu +7Dynamic 3D Gaussian Splatting Scalability
Institute of Automation, Chinese Academy of Sciences, C2DL, Beijing, China · University of Chinese Academy of Sciences, Beijing, China · Shanghai Jiao Tong University, Shanghai, China
Lantern: Conflict-Aware Gradient Blending for Physics-Guided Diffusion Models in Calorimeter Simulation
Jul 27, 2026Farzana Yasmin Ahmad, Vanamala Venkataswamy, Geoffrey FoxPhysics-Guided Diffusion Diffusion Models
Computer Science Department, University of Virginia USA · Visiting Scholar, University of Virginia USA · Computer Science Department, Biocomplexity Institute, University of Virginia USA
DriveWeaver: Point-Conditioned Video Inpainting for Controllable Vehicle Insertion in Autonomous Driving Simulation
Jun 30, 2026Junzhe Jiang, Zipei Ma, Zijie Pan +1Autonomous Driving Simulation Drives
School of Data Science, Fudan University · Shanghai Innovation Institute
Large Databases Need Small, Open-Weight Language Models
Jun 30, 2026Parker Glenn, Alfy SamuelSmall Large Language Models Large Databases
Capital One
BV-Blend: Uncertainty-Weighted Historical Baselines for Stable Critic-Free RL with Verifiable Rewards
Jun 27, 2026Yupeng Chang, Yuan Wu, Yi ChangReinforcement Learning With Verifiable Reward Critic-Free Reinforcement Learning
School of Artificial Intelligence, Jilin University · 2Engineering Research Center of Knowledge-Driven Human-Machine Intelligence, MOE, China · 3International Center of Future Science, Jilin University
Token-to-Token Alignment of Text Embeddings for Semantic Blending
Jun 22, 2026Saar Huberman, Ron Mokady, Or Patashnik +1Image-Text Alignment Text-To-Image
1Tel Aviv University · 2BRIA AI
BindEdit: Taming Attention Leakage for Precise Multi-Object Image Editing
Jun 17, 2026Chaewon Park, Soyoon Lee, Naeun Lee +3Diffusion-Based Image Editing Image Editing
1Sookmyung Women’s University · 2Yonsei University · 3Samsung Research
Deep Temporal Modeling and Ensemble Fusion for Multimodal Emotion Recognition from Physiological Signals
Jun 12, 2026Desta Haileselassie Hagos, Saurav Keshari Aryal, Patrick Ymele-Leki +2Physiological Signals Emotion Recognition
Howard University Washington, DC, USA
Training-Free Occluded Text Rendering via Glyph Priors and Attention-Guided Semantic Blending
May 16, 2026Jingqi Hou, Hongtian WangFew-Shot Font Generation Occlusion
1,2College of Computer Science, Beijing University of Technology, Beijing, China
AttnRouter: Per-Category Attention Routing for Training-Free Image Editing on MMDiT
May 2, 2026Guandong Li, Mengxia YeImage Editing Diffusion Transformers
iFLYTEK · Aegon THTF
Beyond Heuristics: Learnable Density Control for 3D Gaussian Splatting
May 1, 2026Zhenhua Ning, Xin Li, Jun Yu +3Semantic Blending Heuristics
Pengcheng Laboratory, Shenzhen · Harbin Institute of Technology, Shenzhen
Softmax-GS: Generalized Gaussians Learning When to Blend or Bound
Apr 30, 2026Chen Ziwen, Peng Wang, Hao Tan +2Novel View Synthesis Gaussian Primitives
Adobe Research · Tripo AI · Hillbot +1
SG-Blend: Learning an Interpolation Between Improved Swish and GELU for Robust Neural Representations
May 29, 2025Gaurav Sarkar, Syed Affan Daimi, Jay Gala +1Activation Functions Semantic Blending
IIIT · Microsoft · Intel

Sep 30, 2026, cs.CLJ/K move · Enter open · S save

Ready2Blend: From Natural-Language Instructions to Composable Alignment Prompts

Jeesu Jung, Hwan Chang, Juseon Do, Jeonghwan Choi+4

Korea Advanced Institute of Science and Technology · Samsung SDS

Value Alignment Large Language Model Adaptation Semantic Blending Post-Training Instruction Prompt Engineering+1

Open paper AskPDF arXiv

Continual alignment requires LLMs to adapt to new requirements without forgetting previously acquired behaviors. Natural-language instructions are flexible and composable but offer only indirect control, whereas post-training provides stronger adaptation at the cost of repeated parameter updates. We introduce Ready2Blend, which combines the flexibility of natural language with learned alignment. AlignFormer maps each requirement to a fixed-length alignment prompt stored in a modular prompt bank, while the backbone and prior prompts remain frozen. Composability regularization transfers the semantic geometry of textual requirements into prompt space, enabling inference-time blending and reweighting. Across two practical continual alignment settings, Ready2Blend is the only frozen-backbone method that matches post-training-based alignment methods, reaching

93.1

98.5\%

of a joint-training reference with competitive retention, while requiring only a few prompt tokens and up to

4.3\times

less training time. Its modular design further enables weighted personalization and order-free composition without retraining. Code will be released upon acceptance.

Momentum

Related topics

Latest papers 18