logoalt Hacker News

verdvermyesterday at 5:54 PM1 replyview on HN

Similar, but I'm using 35B A3B variation with experimental MTP support

OpenCode is pretty good too


Replies

danielblnyesterday at 6:18 PM

A3B is especially nice, MoE really shines on memory bandwidth contained platforms like the DGX Spark.

show 1 reply