Blackwell B200

Search everything

Mentions · 7d0
Mentions · 30d0
All mentions1
First seenSep 26, 2026
Last mentionedSep 26, 2026

Timeline

1 articles · newest first

Approximating Softmax in Pretrained LLMs: Model Sensitivity and Kernel Acceleration

Approximating softmax in ten frozen LLMs yields a FlashAttention-4 kernel up to 25.8% faster on NVIDIA B200.

Hugging Face daily papersSep 26, 20:00 UTC · Sep 26, 2026AI research1 · 1 read

Appears with

NVIDIAvendor · 1
FlashAttention-4product · 1

Entities are extracted by the model from each article. First seen Sep 26, 2026, last mentioned Sep 26, 2026. Watching keeps it in this browser only (no account); the watchlist page and dashboard alerts use it.