Alexander Korolev 2bde637398 r36 @512K + Engram NVMe: 154/176, 3,774 pp / 822 ms TTFT (2.3x prod prefill, decode -3%) — candidate lane; + forum 381350#521 b12x loader finding (verified: 320B/18B from the checkpoint config) 1 tháng trước cách đây
..
reports 2bde637398 r36 @512K + Engram NVMe: 154/176, 3,774 pp / 822 ms TTFT (2.3x prod prefill, decode -3%) — candidate lane; + forum 381350#521 b12x loader finding (verified: 320B/18B from the checkpoint config) 1 tháng trước cách đây
recipe.toml 6798a62b8a r36 recipe: 512K context (our floor; their 131K profile is not usable) + note the b12x vs instanttensor load-format finding (forum 381350 #521: -35% prefill on the b12x loader at TP4, and it cannot load a DFlash2 drafter) 1 tháng trước cách đây