Unverified50% confidenceFactExact time
FSPO (Function-Space Policy Optimization) proposes sparsely storing the next-token distributions of historical policies on specific inputs during training as fixed temporal anchor targets, rather than maintaining a continuously updated teacher model.
1
Sources
50%
Confidence
Long-term
Relevance
10/15/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedFSPO directly operates in function space on the policy's output distribution (the next-token probability distribution given an input), avoiding parameter symmetry and optimization path complexity, and is more direct and stable than alignment in parameter space.73% similarUnverifiedFSDP shards model parameters, gradients, and optimizer states across multiple GPUs, and is the mainstream solution for training models with hundreds of billions of parameters64% similarUnverifiedFSRS (Free Spaced Repetition Scheduler) was developed by researcher Jarrett Ye using machine learning methods, modeling memory states with two variables: retrievability and stability.63% similarUnverified潜在缓存技术将LLM推理中的KV Cache思想迁移至训练阶段，对轨迹历史帧预先提取潜在表示并缓存63% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1037323API
curl https://kongchang.com/api/v1/knowledge/claims/1037323MCP
get_claim(id=1037323)