Unverified50% confidenceBenchmarkExact time
At moderate acceleration factors, FM-Recon achieved better reconstruction quality than baseline methods without foundation model features; at extremely high acceleration factors, the improvement was limited.
1
Sources
50%
Confidence
Long-term
Relevance
11/17/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedIn FM-Recon the foundation model parameters remain frozen, and only the lightweight feature refinement and fusion modules are trained, to avoid catastrophic forgetting and reduce computational overhead.70% similarUnverifiedThe study proposes correcting CFG in the early flow steps: reducing or reweighting the CFG influence during the noise stage, then gradually restoring standard CFG; this method requires no retraining of the model and adds no inference cost, being completed entirely within the sampling stage.64% similarUnverifiedExperiments found that runtime speedup alone cannot fully characterize optimization capability: some models may achieve decent speedup but with very low fidelity, whereas conversely models with high fidelity often also bring actual speedup.64% similarUnverifiedAlthough traditional restoration-based methods can improve input quality, they incur additional latency, and the downstream accuracy gains are unstable.61% similarUnverifiedIf the input signal does not have sparsity or low-rank structure, the speedup of OpenAI's method shrinks substantially, and may even degrade to the complexity of the classic FFT.60% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1179262API
curl https://kongchang.com/api/v1/knowledge/claims/1179262MCP
get_claim(id=1179262)