Unverified50% confidenceBenchmarkExact time
On multiple LongBench subtasks, MERLIN improves by 13.5% on Gated DeltaNet-1.3B.
1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
10/28/2026
First Seen
Valid until: 1/26/2027
Sources
Related Entities
Related Claims
UnverifiedOn multiple LongBench subtasks, MERLIN improves by 12.0% on Mamba-2-780M.74% similarUnverifiedDeepSeek-V4.1 achieves a 22% improvement in time-to-first-token speed on long prompts.64% similarUnverifiedIn DeepSeek-V4 optimizations, skipping unnecessary topk/router during the decoding phase brings a 3.4% end-to-end TTFT improvement, and workspace reuse brings a 3.9% TTFT improvement.61% similarUnverifiedDeepSeek的编程智能体DeepSWE在SWE-bench基准测试中从预览版的12.8分提升至62.7分55% similarUnverifiedResNet（2015，He 等人）通过引入跳跃连接构造恒等映射的学习旁路，将可训练网络深度推进至 152 层乃至千层，并在 ImageNet 上以显著优势夺冠54% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1105932API
curl https://kongchang.com/api/v1/knowledge/claims/1105932MCP
get_claim(id=1105932)