masc98

↗ HN profile [ 30.9 ms ] full profile
Karma
45
Created
August 17, 2019 (6y ago)
Submissions
0
  1. Enough with LLM news. it may sound boring but I've pretrained a tiny version of BERT, approx 3M params, from scratch on a bunch of italian data (wiki mostly). According to benchmarks it performs just 5-8% worse than a…