Pith. sign in

← back to paper

Review history

arxiv: 2605.16928 · 2 revisions

Full Attention Strikes Back: Transferring Full Attention into Sparse within Hundred Training Steps

  1. 2026-06-30 UNVERDICTED LOW v0.9.1-grok novelty 6.0
    32995 ms 5817 in 1566 out 2026-06-30T19:15:02.085116+00:00
  2. 2026-05-19 UNVERDICTED LOW v0.9.0 novelty 6.0
    38220 ms 5817 in 1251 out 2026-05-19T20:47:59.513839+00:00