Zeitpunkt Nutzer Delta Tröts TNR Titel Version maxTL Di 06.08.2024 00:00:03 191.116 -167 9.390.529 49,1 Mastodon 4.3.0... 500 Mo 05.08.2024 00:00:05 191.283 -32 9.381.093 49,0 Mastodon 4.3.0... 500 So 04.08.2024 00:00:02 191.315 -3 9.372.708 49,0 Mastodon 4.3.0... 500 Sa 03.08.2024 00:00:00 191.318 -1 9.364.577 48,9 Mastodon 4.3.0... 500 Fr 02.08.2024 00:00:03 191.319 -1 9.355.247 48,9 Mastodon 4.3.0... 500 Do 01.08.2024 00:00:07 191.320 0 9.345.588 48,8 Mastodon 4.3.0... 500 Mi 31.07.2024 00:00:02 191.320 -2 9.336.598 48,8 Mastodon 4.3.0... 500 Di 30.07.2024 00:00:03 191.322 -8 9.327.492 48,8 Mastodon 4.3.0... 500 Mo 29.07.2024 00:00:00 191.330 -1 9.318.622 48,7 Mastodon 4.3.0... 500 So 28.07.2024 00:00:03 191.331 0 9.309.563 48,7 Mastodon 4.3.0... 500
Scott Williams 🐧 (@vwbusguy) · 09/2020 · Tröts: 15.603 · Folger: 2.798
Di 06.08.2024 02:45
I got IBM's 34b granite instruction model up and running in a box with a bunch of nVidia GPUs today, but basing on their demo code it oddly pulled memory from the GPUs and then did all the processing on one cpu thread, which took several minutes for one query. The query results were actually good, but about the most compute expensive way to possibly do it.
[Öffentlich] Antw.: 0 Wtrl.: 0 Fav.: 0 · via Tusky