Unverified50% confidenceBenchmarkExact time
NVIDIA的双塔扩散语言模型保留了原始自回归模型98.7%的质量，速度提升约2.4倍
1
Sources
50%
Confidence
Long-term
Relevance
7/20/2026
First Seen
Sources
Related Claims
UnverifiedNVIDIA从A100到H100再到B100的迭代周期约两年，每代FP8混合精度算力提升约3-4倍74% similarUnverifiedNVIDIA TensorRT加速方案可将Stable Diffusion的推理速度提升2-4倍73% similarUnverifiedNVIDIA H100 GPU原生支持FP8格式，其吞吐量相较FP16提升约2倍69% similarUnverifiedNVIDIA的cuFFT库在处理大规模多维FFT时性能可达CPU实现的数十倍69% similarUnverifiedNVIDIA每代GPU性价比提升约2-3倍69% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/565258API
curl https://kongchang.com/api/v1/knowledge/claims/565258MCP
get_claim(id=565258)