Expired50% confidenceFactExact time
The experiments used IPPO (Independent PPO) and GPPO (Graph Independent PPO) policy types, including heterogeneous variants.
1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
7/12/2026
First Seen
This claim has expired. It is kept for the record and no longer describes the current state.
Valid until: 10/10/2026(expired)
Sources
Adversarial RL: Why Are Critic Attacks Stronger in Multi-Agent Environments?
redditr/reinforcementlearning7/10/2026
Related Claims
Cite This Claim
Stable URI
https://kongchang.com/claim/493663API
curl https://kongchang.com/api/v1/knowledge/claims/493663MCP
get_claim(id=493663)