Harness-S: Learning from Sibling Harnesses for Generalizable Agent Training
cs.AI
Submitted: 2026-11-03
Updated: 2026-11-03
Code: https://github.com/DeepExperience/Harness-S
Terminology
Sources
- On-Policy Distillation of Language Models: Learning from Self-Generated Mistakes
- Hindsight Experience Replay
- Thinking Fast and Slow with Deep Learning and Tree Search
- RUDDER: Return Decomposition for Delayed Rewards
- Large Language Models as Tool Makers
- Learning by Cheating
- Reinforcement Learning for Long-Horizon Interactive LLM Agents
- Go-Explore: a New Approach for Hard-Exploration Problems
- MiniLLM: On-Policy Distillation of Large Language Models
- Distilling Step-by-Step! Outperforming Larger Language Models with Less Training Data and Smaller Model Sizes
- Search-R1: Training LLMs to Reason and Leverage Search Engines with Reinforcement Learning
- Sequence-Level Knowledge Distillation
- Unifying distillation and privileged information
- WebGPT: Browser-assisted question-answering with human feedback
- A Reduction of Imitation Learning and Structured Prediction to No-Regret Online Learning
- Policy Distillation
- Toolformer: Language Models Can Teach Themselves to Use Tools
- Kickstarting Deep Reinforcement Learning
- Harness-R1: Learning to Edit Executable Runtime Harnesses from Agent Failure Trajectories
Related papers
- Multi-Objective Structured Pruning of LLMs for Latency and Model Size Optimization
- 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
- Embeddings for Preferences, Not Semantics
- Can We Trust Item Response Theory for AI Evaluation?
- Diagnosing Task Insensitivity in Language Agents