149
edits
Arcweld: Difference between revisions
no edit summary
(Created page with "'''Arcweld''' is a proposed server intended for the primary goal of experimenting with novel LLM inference technologies, the secondary goals of a) reducing the cost of using hosted APIs when it would not interfere with the primary goal and b) providing space to train models.") |
No edit summary |
||
| (3 intermediate revisions by the same user not shown) | |||
| Line 1: | Line 1: | ||
'''Arcweld''' is a | '''Arcweld''' is a server for the primary goal of rapidly and efficiently experimenting with open weights models, with secondary goals of 1) researching novel LLM inference technologies 2) reducing the cost of using hosted APIs when it would not interfere with the primary goal. It is not intended for training since it isn't the ideal way to purchase memory bandwidth per dollar (critical for training), although it isn't forbidden. | ||
== Specifications == | |||
* Mac Studio M3 Ultra | |||
** 819 GB/s memory bandwidth | |||
** 60-core GPU | |||
** 32-core Neural Engine | |||
* 512GB Unified Memory | |||
== Setup instructions == | |||
# Download and install [https://tailscale.com/ Tailscale] | |||
#* Some VPNs are incompatible and you may need to disable them. | |||
# Join using a Tailscale invite link | |||
# If you've been granted login access, `ssh sparks@arcweld` to access the shared inference account. '''You may have to reboot for it to work.''' You can also try manually editing the machine IPv4 inside Tailscale portal to something random that doesn't conflict with your LAN. | |||
== See also == | |||
* [[Elysium]] | |||
[[Category:Infrastructure]] | |||