Reposted by Tim Vieira
Tokenization is an often-overlooked aspect of modern #NLP, but it’s experiencing a resurgence — thanks in large part to @karpathy.bsky.social and his classic tweet:
x.com/karpathy/sta...
Come hang out with us and let's fix these problems!
x.com/karpathy/sta...
Come hang out with us and let's fix these problems!
February 10, 2025 at 4:26 PM
Tokenization is an often-overlooked aspect of modern #NLP, but it’s experiencing a resurgence — thanks in large part to @karpathy.bsky.social and his classic tweet:
x.com/karpathy/sta...
Come hang out with us and let's fix these problems!
x.com/karpathy/sta...
Come hang out with us and let's fix these problems!
No amazing live music at the closing ceremony...
December 17, 2024 at 1:17 AM
No amazing live music at the closing ceremony...
Also, what will the role of adaptive/amortized inference be?
E.g., twisted SMC
arxiv.org/abs/2404.17546
Variational best-of-N
Our version:
arxiv.org/abs/2407.06057
Google's: arxiv.org/abs/2407.14622
E.g., twisted SMC
arxiv.org/abs/2404.17546
Variational best-of-N
Our version:
arxiv.org/abs/2407.06057
Google's: arxiv.org/abs/2407.14622
December 16, 2024 at 8:01 PM
Also, what will the role of adaptive/amortized inference be?
E.g., twisted SMC
arxiv.org/abs/2404.17546
Variational best-of-N
Our version:
arxiv.org/abs/2407.06057
Google's: arxiv.org/abs/2407.14622
E.g., twisted SMC
arxiv.org/abs/2404.17546
Variational best-of-N
Our version:
arxiv.org/abs/2407.06057
Google's: arxiv.org/abs/2407.14622
What will the hardware-friendly search algorithms be? My top picks are best-of-N and sequential Monte Carlo because they do search without backtracking.
December 16, 2024 at 7:06 PM
What will the hardware-friendly search algorithms be? My top picks are best-of-N and sequential Monte Carlo because they do search without backtracking.
