From The #Resistance #Garden on #BlueSky:
"#Liberatory #economics will necessarily include creating ways to #subsist that don't rely on #extracting #resources and #labor from elsewhere, not simply #distributing #imperial spoils more equitably."
Verwaltet von:
#extracting
0 Beiträge0 Beteiligte0 Beiträge heute
Extracting memorized pieces of books from open-weight language models
https://arxiv.org/abs/2505.12546
#HackerNews #Extracting #memorized #pieces #of #books #from #open-weight #language #models #languagemodels #AIresearch #bookextraction #openweightmodels #arxiv
arXiv.orgExtracting memorized pieces of (copyrighted) books from open-weight language modelsPlaintiffs and defendants in copyright lawsuits over generative AI often make sweeping, opposing claims about the extent to which large language models (LLMs) have memorized plaintiffs' protected expression. Drawing on adversarial ML and copyright law, we show that these polarized positions dramatically oversimplify the relationship between memorization and copyright. To do so, we leverage a recent probabilistic extraction technique to extract pieces of the Books3 dataset from 13 open-weight LLMs. Through numerous experiments, we show that it's possible to extract substantial parts of at least some books from different LLMs. This is evidence that the LLMs have memorized the extracted text; this memorized content is copied inside the model parameters. But the results are complicated: the extent of memorization varies both by model and by book. With our specific experiments, we find that the largest LLMs don't memorize most books -- either in whole or in part. However, we also find that Llama 3.1 70B memorizes some books, like Harry Potter and 1984, almost entirely. We discuss why our results have significant implications for copyright cases, though not ones that unambiguously favor either side.
Extracting content from an LCP "protected" ePub
https://shkspr.mobi/blog/2025/03/towards-extracting-content-from-an-lcp-protected-epub/
Terence Eden’s Blog · Extracting content from an LCP "protected" ePub
Mehr von 
Terence Eden
Design patterns for #extracting from #REST APIs
https://blog.sequin.io/design-patterns-for-extracting-from-rest-apis/
The Sequin Blog · Design patterns for extracting from REST APIsMost integrations rely on an extraction process. But REST APIs don't make extraction easy. Follow these design patterns to avoid common pitfalls.
Antwortete im Thread
@thoughtpunks IMHO the #unequal #taxation of #income types is #ClassWarfare.
There's no non-classist asnwer to the question of "Why do #Billionaires not have to contribute the same if not more in both % and total amount of #taxes and fees than any #WageWorker?"
It's not as if prople like #Bezos can't get sick, become unable to work or manage their funds nor that they ain't also #extracting #wealth from #society...
#BuyBorrowDie should be illegal and #CaoitalGains should be taxed higher than #WageWork of equal payout!
https://www.youtube.com/watch?v=t6V9i8fFADI
Never mind #SeaLevelRise: human activity can make the ground go down faster than the seas rise.
"Some land #subsidence, Bekaert said, is related to deep natural processes over long periods of time, such as responding to plate #tectonic activity or to the retreating of the #glaciers from the last Ice Age. Other sinking is linked to human activity, including #extracting oil, #water or minerals from underground. In cities, buildings can also add weight and push land down."
https://www.washingtonpost.com/climate-environment/2023/05/30/land-sinking-us-subsidence-sea-level/
The Washington PostLand around the U.S. is sinking. Here are some of the fastest areas.