Unverified60% confidenceBenchmarkExact time
IMGNet achieved 76.85% vs. 74.62% on CPLFW and 78.80% vs. 77.22% on AgeDB-30, with a more pronounced advantage on higher-difficulty benchmarks
2
Sources
60%
Confidence
Long-term
Relevance
7/10/2026
First Seen
Sources
IMGNet: A New Face Verification Approach Where Sign Pattern Matching Challenges Cosine Similarity
redditr/deeplearning7/9/2026
Related Claims
UnverifiedOn the comprehensive benchmark, IMG Sign achieves 81.02% while cosine similarity achieves 79.49%; on LFW the figures are 96.27% and 95.53% respectively77% similarUnverified业界顶级集群的MFU（模型算力利用率）通常在35%-60%之间66% similarUnverifiedWith context management enabled, Iris-mini achieves 82.2%, 84.8%, 86.9%, and 52.3% on BrowseComp, BrowseComp-ZH, DeepSearchQA, and HLE respectively.65% similarUnverifiedAnthropic states that Opus 5.5 is more than 30% faster than the previous generation and can save approximately 40% of compute under default settings.63% similarUnverifiedAt small scale, the FLOPs share of the MLP layer is approximately 44%, rising to approximately 80% when scaled to 175B parameters62% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/461645API
curl https://kongchang.com/api/v1/knowledge/claims/461645MCP
get_claim(id=461645)