Unverified50% confidenceFactExact time
Existing interface randomization strategies can only extend the Agent's dependence on the interface from a 'single interface' to the 'training interface distribution', and out-of-distribution perturbations still lead to degradation in Agent performance.
1
Sources
50%
Confidence
Long-term
Relevance
10/15/2026
First Seen
Sources
Related Entities
Related Claims
VerifiedBehavioral cloning suffers from the distribution shift problem: during training the policy only encounters the state distribution covered by expert trajectories, and during deployment the accumulation of small deviations causes the actual state distribution visited to gradually diverge from the training distribution.66% similarUnverifiedThe capabilities of LLM Agents depend heavily on the interaction interface used during training, and semantically equivalent but differently expressed interfaces (such as changing an action name from search to lookup, or changing parameters from JSON to YAML) can cause a marked decline in their performance.66% similarUnverifiedPrevious multimodal diffusion work usually trained unimodal base models first and then extended to bimodal generation through fine-tuning; this route lowers training costs but may limit the depth of cross-modal interaction.64% similarVerifiedWhen training a dexterous hand to solve the Rubik's Cube, OpenAI used Automatic Domain Randomization, which maintains training efficiency by dynamically adjusting environment difficulty.64% similarVerifiedDomain randomization technology randomly perturbs physical parameters (such as mass, friction coefficient, motor delay, etc.) during training, forcing the policy to learn to operate robustly under various conditions.63% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1037938API
curl https://kongchang.com/api/v1/knowledge/claims/1037938MCP
get_claim(id=1037938)