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
This claim has expired. It is kept for the record and no longer describes the current state.
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% similarUnverifiedBehavior cloning pretraining solves the cold-start problem in reinforcement learning: pure reinforcement learning in sparse-reward environments often requires hundreds of millions of interactions before learning a minimally usable policy.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)