Unverified50% confidenceFactExact time
GNoME基于图神经网络架构，将晶体结构表示为原子节点和化学键边的图，在Materials Project等数据库中约4.8万种已知稳定材料上训练
1
Sources
50%
Confidence
Long-term
Relevance
8/14/2026
First Seen
Sources
Related Claims
UnverifiedDeepMind's GNoME project predicted approximately 2.2 million potentially stable crystal structures, about an order of magnitude more than the stable materials previously known to humans.65% similarUnverifiedGrok 4.5's technical foundation is the V9 architecture with 1.5 trillion parameters, and it incorporates Cursor's programming data for reinforcement training.58% similarUnverifiedAnalysis speculates that the model is distributed across approximately 70 to 100 Cerebras wafer-scale chips, with each chip hosting one neural network layer, totaling 70 to 90 layers.58% similarUnverifiedGhidra is a reverse engineering framework open-sourced by the National Security Agency in 2019, supporting dozens of processor architectures and providing collaborative analysis capabilities.57% similarUnverified混合架构的核心理念是将不同类型的神经网络模块（如RNN/GRU、Transformer注意力、GNN、统计因子分支）组合以捕捉金融数据中的多维度模式57% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/746291API
curl https://kongchang.com/api/v1/knowledge/claims/746291MCP
get_claim(id=746291)