Parker Seegmiller
@pkseeg.bsky.social
NLP @ Dartmouth College
Formerly @ Amazon AGI (science), CVS Health (science), Jimmy John's (deliveries)
pkseeg.com
Formerly @ Amazon AGI (science), CVS Health (science), Jimmy John's (deliveries)
pkseeg.com
in which I mess around and decide that it would take 104 days and several terabytes to calculate a single actual conditional token probability distribution in the English language
pkseeg.com/post/cost-ac...
pkseeg.com/post/cost-ac...
Costs of Calculating the Actual Causal Language Conditional Probabilities of English | pkseeg
Estimating the cost of a single actual conditional token probability distribution in causal language modeling in a given language
pkseeg.com
January 3, 2024 at 3:10 PM
in which I mess around and decide that it would take 104 days and several terabytes to calculate a single actual conditional token probability distribution in the English language
pkseeg.com/post/cost-ac...
pkseeg.com/post/cost-ac...
First post!
My recent work (w/Sarah Preum) in developing a statistical depth for texts has been accepted at #EMNLP2023. Our depth orders texts along a singular axis for things like outlier detection, representative example selection, and significance testing.
arxiv.org/abs/2310.15010
My recent work (w/Sarah Preum) in developing a statistical depth for texts has been accepted at #EMNLP2023. Our depth orders texts along a singular axis for things like outlier detection, representative example selection, and significance testing.
arxiv.org/abs/2310.15010
November 13, 2023 at 1:32 PM
First post!
My recent work (w/Sarah Preum) in developing a statistical depth for texts has been accepted at #EMNLP2023. Our depth orders texts along a singular axis for things like outlier detection, representative example selection, and significance testing.
arxiv.org/abs/2310.15010
My recent work (w/Sarah Preum) in developing a statistical depth for texts has been accepted at #EMNLP2023. Our depth orders texts along a singular axis for things like outlier detection, representative example selection, and significance testing.
arxiv.org/abs/2310.15010