MoMHa: Multi-Objective Optimization of LLM Harnesses over Accuracy, Safety, and Tokens
cs.AI
Submitted: 2026-09-25
Updated: 2026-09-25
Terminology
Sources
- GEPA: Reflective Prompt Evolution Can Outperform Reinforcement Learning
- Trace is the Next AutoDiff: Generative Optimization with Rich Feedback, Execution Traces, and LLMs
- Is Safety Standard Same for Everyone? User-Specific Safety Evaluation of Large Language Models
- Llama Guard: LLM-based Input-Output Safeguard for Human-AI Conversations
- SWE-bench: Can Language Models Resolve Real-World GitHub Issues?
- Beyond Accuracy: Decomposing the Reasoning Efficiency of LLMs
- DSPy: Compiling Declarative Language Model Calls into Self-Improving Pipelines
- Meta-Harness: End-to-End Optimization of Model Harnesses
- The Power of Scale for Parameter-Efficient Prompt Tuning
- Prefix-Tuning: Optimizing Continuous Prompts for Generation
- Holistic Evaluation of Language Models
- AutoHarness: improving LLM agents by automatically synthesizing a code harness
- Terminal-Bench: Benchmarking Agents on Hard, Realistic Tasks in Command Line Interfaces
- Multi-Objective Alignment of Large Language Models Through Hypervolume Maximization
- Optimizing Instructions and Demonstrations for Multi-Stage Language Model Programs
- Automatic Prompt Optimization with "Gradient Descent" and Beam Search
- NeMo Guardrails: A Toolkit for Controllable and Safe LLM Applications with Programmable Rails
- Identifying the Risks of LM Agents with an LM-Emulated Sandbox
- AutoPrompt: Eliciting Knowledge from Language Models with Automatically Generated Prompts
- Steering Language Models With Activation Engineering
Related papers
- MAVEN-T: Reinforced Heterogeneous Distillation for Real-Time Multi-Agent Trajectory Prediction
- Model Discovery Agent: LLM-assisted Bayesian experiment design for data-efficient discovery of mechanistic world models
- The Clinician's Veto: Navigating Trust, Liability, and Uncertainty in Autonomous AI Prescribing
- MindHelper: Closed-Loop Embodied Mental-State Reasoning for Precision Intervention
- Incumbent Advantage: Brand Bias and Cognitive Manipulation Dynamics in LLM Recommendation Systems
- VSAL: A Vision Solver with Adaptive Layouts for Graph Property Detection