Expired60% confidenceFactTime unknown
Composer 2 adopts a two-phase advancement strategy: mid-training followed by large-scale reinforcement learning.
2
Sources
60%
Confidence
Medium-term (~90 days)
Relevance
5/31/2026
First Seen
Valid until: 8/29/2026(expired)
Sources
Related Entities
Related Claims
Unverified85% of the compute resources in Composer 2.5's training were invested in additional continued pre-training and reinforcement learning post-training.78% similarUnverifiedQwQ-32B采用两阶段强化学习训练策略：第一阶段在数学和编程任务上进行强化学习，第二阶段扩展到通用领域68% similarUnverifiedTactico employs a two-phase training approach: first using tens of thousands of human games for imitation learning, then using millions of self-play games for reinforcement learning to converge toward a Nash equilibrium.64% similarUnverified行为克隆预训练解决了强化学习的冷启动问题：纯强化学习在稀疏奖励环境中往往需要数亿次交互才能学到基本可用策略64% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/6586API
curl https://kongchang.com/api/v1/knowledge/claims/6586MCP
get_claim(id=6586)