Unverified50% confidenceFactExact time
On eight related development warnings, increasing the context count from 1 to 50 led to a 49-fold increase in token usage but did not improve the rate of clearing insecure code.
1
Sources
50%
Confidence
Long-term
Relevance
10/28/2026
First Seen
Sources
Related Entities
Related Claims
Unverified某特征族将困难设备风险排名从第37位提升到第7位，Precision@10从0.50提高到0.70，但早期有效预警数量从4个下降到2个，作者据此拒绝了该变更67% similarUnverifiedRaising the share of risky-conversation samples from 10% to 25% improves standalone safety-check recall (53% to 68%) but causes the model to raise more false alarms on ordinary chat; using risky samples only for training the safety question can mitigate the side effects.66% similarUnverifiedAblation analysis shows that fine-tuning on the segment paradigm yields a +5.29-point improvement on the test set, while replacing the first 4,500 transcript words with 2,000 retrieved words yields only a +1.55-point improvement.66% similarUnverifiedUn-tuned PPI with fixed λ=1 inflates human-label consumption by a factor of 2.27, which amounts to injecting noise in medium-quality settings where judge agreement is only 45%–77%.66% similarUnverifiedIn code review tests, Opus 5.5 improved recall on more difficult pull requests from a baseline of 54% to nearly 77%, but the number of comments increased by approximately 20%.65% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1106288API
curl https://kongchang.com/api/v1/knowledge/claims/1106288MCP
get_claim(id=1106288)