[KongchangAI]
Unverified50% confidenceFactExact time

The scope of this study's conclusions is limited to: length-normalized reward plus an SFT retention term, sparse LoRA at 8/32 layers, a single 4B model, easily separable synthetic preference data, and three structured tasks.

1
Sources
50%
Confidence
Long-term
Relevance
11/5/2026
First Seen

Sources

Related Entities

Related Claims

Cite This Claim

Stable URI
https://kongchang.com/claim/1138295
API
curl https://kongchang.com/api/v1/knowledge/claims/1138295
MCP
get_claim(id=1138295)