Four pages on Bittensor subnet 9, built from what Macrocosmos and others have published, and a note on the SDK announced on 28 September.
Every Orion training run, with its numbers, dates and sources. Cells the published record does not fill say unknown.
Everyone training, or selling training, on compute outside a single reserved cluster, placed by their relation to IOTA.
Nine Macrocosmos claims, each followed from the post where it started to every page that repeated it.
From the 2024 pretraining contest to the compression the Orion runs use, in the order they were written.
Announced at Exploit Summit on 28 September. Liquid Compute turns “the long tail of global compute into capacity you can actually train on,” and the SDK runs any training workload across it “as if it were one cluster.”