Christopher Manning @chrmanning · 23 Aug 2026
@chrmanning @akarshkumar0101 @phillip_isola Another cool connection is the ability to parallelize GRUs over the sequence length for both forward and backward passes, allowing pretraining at scale without teacher forcing. As for example in this paper! https://t.co/9QB0cr7rqf
1 926Views
36Likes
5Reposts
1Replies
0Quotes
24Bookmarks
Is that a lot?
0.07×vs this author's median29 579 views is typical
0Percentile for this authorof 8 recent posts
0.43×vs 100K–1M median4 505 views is typical
1.14%Reachviews ÷ followers
2.18%Engagement rateof viewers reacted