agentic-grpo-longhorizon
Fixing GRPO training collapse in long-horizon multi-tool agents. A lightweight PRM-Lite + LATA joint approach achieves +37% over vanilla GRPO on τ-bench airline (50-task, multi-turn).
qiqihezh/agentic-grpo-longhorizon
View on GitHub ↗Fixing GRPO training collapse in long-horizon multi-tool agents. A lightweight PRM-Lite + LATA joint approach achieves +37% over vanilla GRPO on τ-bench airline (50-task, multi-turn).
agentic-grpo-longhorizon grew +7.4% in the last 30 days, ranks 49th of 82 in Reinforcement Learning / RLHF, last commit 119 days ago.
Star trajectory
79 days tracked+7.4% in 30d, above the Reinforcement Learning / RLHF median of +1.1%.
Last commit 119 days ago, staler than the median of 46d.
How it compares in Reinforcement Learning / RLHF
Ranked 49th of 82 by stars.
- Language
- Python
- License
- -
- Contributors
- 2
- Forks
- 18
- Watchers
- 0
- Open issues
- 2
- Releases
- 0
- Created
- 2026-04-29
- Stars this week
- +2
You've seen where agentic-grpo-longhorizon stands today.
Get the moves before everyone else: the 5 Reinforcement Learning / RLHF repos climbing fastest, every Tuesday.
Free weekly digest. Double opt-in, unsubscribe anytime.