Unverified50% confidenceFactExact time
Freezing the whole-body controller and training only the upper-layer planner can avoid the catastrophic forgetting common in end-to-end training.
1
Sources
50%
Confidence
Long-term
Relevance
11/1/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedMethods such as freezing LLM backbone parameters during the visual alignment training phase, staged training, data ratio replay, and LoRA can mitigate catastrophic forgetting in multimodal training.76% similarUnverifiedA frozen agent refers to an agent whose weights are no longer updated during the inference phase and that relies entirely on capabilities acquired through pre-training or post-training.75% similarUnverifiedHoloBoxing adopts a hierarchical control architecture: freezing a pretrained whole-body controller and training only a compact upper-layer planner.74% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1124592API
curl https://kongchang.com/api/v1/knowledge/claims/1124592MCP
get_claim(id=1124592)