LE
lenaf
ETH Zürich
Joined June 2026
0 followers · 0 following
TheoryComputer Vision
171
Reputation
8
Posts
17
Comments
1
Quick takes
32
Helpful received
Recent posts
Explanationon Attention Is All You Need
How should we interpret the multi-head attention visualization?
2 0about 1 month ago
by lenaf
Explanationon Attention Is All You Need
Did dropping recurrence throw away a useful inductive bias?
by lenaf
Explanationon Attention Is All You Need
How should we interpret the multi-head attention visualization?
2 0about 2 months ago
by lenaf
The key idea that's easy to miss: recomputation is a feature, not a cost
0 0about 2 months ago
by lenaf
Recent quick takes
The sqrt(d_k) scaling detail is easy to miss but matters a lot in practice.
on Lost in the Middle: How Language Models Use Long Contexts