Unverified50% confidenceFactExact time
The model architecture is a cross-attention DiT (896 width × 16 layers) using 2D RoPE, QK-norm, SwiGLU, and adaLN-single.
1
Sources
50%
Confidence
Long-term
Relevance
9/11/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedLLaMA-2 70B采用80层、64个注意力头的架构，并使用GQA（8个KV头共享64个Q头）64% similarUnverified三元/低比特量化加MoE稀疏架构正在成为端侧大模型的主流技术组合62% similarUnverifiedQwen 3.8 27B adopts a hybrid architecture in which, across 64 layers, the majority use Gated DeltaNet layers with Gated Full Attention layers interspersed at regular intervals.60% similarUnverifiedTRT-LLM supports FP8/INT8 quantization, Flash Attention fused operators, and multi-GPU tensor parallelism.60% similarUnverified该攀岩分析工具核心由三个AI模型组合使用：VLM Run的Orion、ViTPose+ Huge、RT-DETR60% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/902328API
curl https://kongchang.com/api/v1/knowledge/claims/902328MCP
get_claim(id=902328)