iGEN
Visit IGEN World Explore IGEN Expo
EXPLORE UPGRADE PLANS
BREAKING
WhatsApp tests 'Offers & Updates' folder to declutter business chats Aurora Reports Q2 Loss, Details Per-Mile Pricing for Driverless Truck Services Apple iPad Air OLED display, M5 chip and biggest redesign expected in 2027 India's soyabean acreage recovers as July rains boost Kharif sowing China’s EV Market Surges Past 16 Million as Battery Waste Wave Arrives WIRED Tests Plastic-Free Stainless Steel Water Filters From $199 to $549 FBI Warns Iran-Linked Hackers Hit Water Systems in Seven US States US Crude Bound for Israel for First Time Since 2023, Times of India Reports RBI's Special Swap Facility Draws $40.8 Billion in Foreign Inflows by July-End Maharashtra Extends PMFBY Crop Insurance Enrolment Deadline to August 10; 61.82 Lakh Farmers Registered WhatsApp tests 'Offers & Updates' folder to declutter business chats Aurora Reports Q2 Loss, Details Per-Mile Pricing for Driverless Truck Services Apple iPad Air OLED display, M5 chip and biggest redesign expected in 2027 India's soyabean acreage recovers as July rains boost Kharif sowing China’s EV Market Surges Past 16 Million as Battery Waste Wave Arrives WIRED Tests Plastic-Free Stainless Steel Water Filters From $199 to $549 FBI Warns Iran-Linked Hackers Hit Water Systems in Seven US States US Crude Bound for Israel for First Time Since 2023, Times of India Reports RBI's Special Swap Facility Draws $40.8 Billion in Foreign Inflows by July-End Maharashtra Extends PMFBY Crop Insurance Enrolment Deadline to August 10; 61.82 Lakh Farmers Registered
Home ›› Topics ›› self-play

Topic

self-play

3 stories
Self-Play RL with 30 Minutes of Human Data Trains Coordinated Driving Policies Technology
Artificial Intelligence #artificial intelligence#autonomy

Self-Play RL with 30 Minutes of Human Data Trains Coordinated Driving Policies

A new approach from researchers trains autonomous driving policies using self-play reinforcement learning regularized by only 30 minutes of human demonstrations. The method requires 2500x less human data than imitation learning and completes training in 15 hours on a single consumer-grade GPU. The resulting policies successfully coordinate with held-out human trajectories, avoiding the alien driving conventions common in pure self-play systems.

Jun 20, 2026 1 source
S-SPPO: Semantic Calibration Boosts LLM Preference Alignment Without Human Data Technology
Artificial Intelligence #artificial intelligence#machine learning

S-SPPO: Semantic Calibration Boosts LLM Preference Alignment Without Human Data

S-SPPO, a dual-space semantic calibration framework, fixes instability in Self-Play Preference Optimization (SPPO) for large language models. By annealing win targets and enforcing geometric diversity, it achieves superior alignment results on AlpacaEval 2.0 without extra human preferences.

Jun 17, 2026 1 source
Tree-like Self-Play Framework Teaches LLMs to Fix Security Flaws in Code Generation Technology
Artificial Intelligence #ai#llm

Tree-like Self-Play Framework Teaches LLMs to Fix Security Flaws in Code Generation

Researchers introduce Tree-like Self-Play (TSP), a framework that treats secure code generation as a fine-grained sequential decision process. TSP significantly outperforms standard supervised fine-tuning (SFT) and reinforcement learning (RL) on Python security benchmarks, achieving a 75.8% pass rate and reducing unseen vulnerabilities by 24.5% while generalising across programming languages.

Jun 16, 2026 1 source