Pluralis Research @Pluralis · 20 May 2026
Today we're releasing Agora: the first ever pretraining stack that allows non-collocated consumer GPUs to be competitive with centralized clusters Agora is 15x faster than Megatron-LM in this setting and is only 1.5x less efficient in terms of tokens per unit compute than https://t.co/Rs3G9JS4Os
91 406Views
311Likes
45Reposts
29Replies
23Quotes
147Bookmarks
Is that a lot?
16.8×vs this author's median5 437 views is typical
100Percentile for this authorof 6 recent posts
42.5×vs 10K–100K median2 151 views is typical
6.9× audienceReachviews ÷ followers
0.45%Engagement rateof viewers reacted