Unverified50% confidenceBenchmarkExact time
When using INT4 HQQ quantization, catastrophic transcription loops occurring when the model processes West African accents were amplified by 5 to 7 times.
1
Sources
50%
Confidence
Long-term
Relevance
9/25/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedTransformer注意力机制的计算量随序列长度平方增长，早期语言模型上下文窗口上限约4000个token61% 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.60% similarUnverifiedIn real-world production environment tests on DeepSeek V4, inference response speed improved by 60% to 85%, engine throughput increased by 51%, and throughput in high-concurrency scenarios improved by 661%60% similarUnverified多项研究表明，多智能体系统中的错误传播率可达单Agent的3-5倍58% similarUnverifiedvLLM's measured throughput can be improved by 2–4× compared to a naive implementation.57% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/949159API
curl https://kongchang.com/api/v1/knowledge/claims/949159MCP
get_claim(id=949159)