Unverified50% confidenceOpinionExact time
The trend revealed by Latent Compaction is that static components in the inference process can dynamically adapt to the input through test-time training.
1
Sources
50%
Confidence
Long-term
Relevance
10/15/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedThe representative work of soft prompt techniques is Prompt Tuning proposed by Lester et al. 2021, which adds trainable continuous vectors only at the front of the input sequence while keeping model parameters frozen.73% similarVerified推理时扩展（inference-time scaling）通过在推理阶段增加计算量提升模型性能，存在类似预训练的幂律关系71% similarUnverifiedThis approach embodies a compute-shifting philosophy that moves the computational burden from repeated test attempts to a one-time learning investment.70% similarUnverifiedAn efficient inference engine can accelerate rollout generation, thereby shortening the iteration cycle of RL training.70% similarVerified域随机化是指在仿真训练中刻意随机扰动物理参数，迫使策略学习到对参数变化具有泛化能力的行为69% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1037901API
curl https://kongchang.com/api/v1/knowledge/claims/1037901MCP
get_claim(id=1037901)