qiqihezh/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).

259

Stars

18

Forks

0.0

Radar Score

Declining

Momentum

Scoring Breakdown

Impact0.0

Velocity0.0

Health0.0

Growth Metrics

7-Day Stars

+0

30-Day Stars

+0

Acceleration

x

Language

Python

Why it ranks

Momentum is cooling relative to the rest of the index.

AI Agents · task agent