Recent searches

No recent searches

Search options

Only available when logged in.
sigmoid.social is one of the many independent Mastodon servers you can use to participate in the fediverse.
A social space for people researching, working with, or just interested in AI!

Administered by:

Server stats:

596
active users

sigmoid.social: AboutProfiles directoryPrivacy policy

Mastodon: AboutGet the appKeyboard shortcutsView source codev4.4.2

#tokenization

1 post1 participant0 posts today
Miguel Afonso Caetano

"Prominent crypto firms such as Robinhood, Republic, Coinbase, and Kraken are rapidly moving towards “tokenizing” traditional stocks, and pressuring regulators to allow it. Instead of buying your shares of publicly traded firms via a brokerage account that places orders on the NYSE or Nasdaq, you would use a crypto trading app to purchase a token representing a share. Companies hoping to develop such platforms usually promote the idea by saying that a blockchainified stock market would expand trading hours, and would be more accessible to international investors who didn’t want to go through the somewhat onerous process of opening an American brokerage account.

These companies don’t usually admit that, by encasing stocks in a blockchainy wrapper, they hope to tap into lucrative equities markets while sidestepping the expensive compliance and oversight requirements of traditional American brokerages and exchanges. This fits the long history of companies trying to use blockchains as a magic get-out-of-regulation-free wand, reminiscent of the 2017 bubble when companies used “initial coin offerings” (ICOs) to try to sidestep IPO regulations.d Indeed, Robinhood has been heavily lobbying for “a new regulatory approach [that’s] needed to allow tokenization to flourish” and not “stifle growth and innovation”.1 Regular readers of this newsletter will recognize this language as the standard rhetoric of a crypto company asking for carveouts and exemptions from regulations we collectively learned are necessary, oh, about a century ago — when a speculative bubble emerged around stocks sold to the public based on false or incomplete information and we wound up in the Great Depression."

citationneeded.news/issue-88/

Citation Needed · Issue 88 – The stockchainCrypto firms hope putting a blockchain veneer on traditional equities will allow them to sidestep lessons learned in the 1929 Wall Street crash, crypto firms look to become banks, and Congress celebrates crypto surveillance while claiming to outlaw it
#Crypto#Cryptocurrencies#Tokenization
Paul Houle

⛓️‍💥 Broken Tokens? Your Language Model can Secretly Handle Non-Canonical Tokenizations

arxiv.org/abs/2506.19004

arXiv.orgBroken Tokens? Your Language Model can Secretly Handle Non-Canonical TokenizationsModern tokenizers employ deterministic algorithms to map text into a single "canonical" token sequence, yet the same string can be encoded as many non-canonical tokenizations using the tokenizer vocabulary. In this work, we investigate the robustness of LMs to text encoded with non-canonical tokenizations entirely unseen during training. Surprisingly, when evaluated across 20 benchmarks, we find that instruction-tuned models retain up to 93.4% of their original performance when given a randomly sampled tokenization, and 90.8% with character-level tokenization. We see that overall stronger models tend to be more robust, and robustness diminishes as the tokenization departs farther from the canonical form. Motivated by these results, we then identify settings where non-canonical tokenization schemes can *improve* performance, finding that character-level segmentation improves string manipulation and code understanding tasks by up to +14%, and right-aligned digit grouping enhances large-number arithmetic by +33%. Finally, we investigate the source of this robustness, finding that it arises in the instruction-tuning phase. We show that while both base and post-trained models grasp the semantics of non-canonical tokenizations (perceiving them as containing misspellings), base models try to mimic the imagined mistakes and degenerate into nonsensical output, while post-trained models are committed to fluent responses. Overall, our findings suggest that models are less tied to their tokenizer than previously believed, and demonstrate the promise of intervening on tokenization at inference time to boost performance.
#ai#llm#cs
*
AxelPolleres (Q54860587)

We're #hiring!

Two(!) full #professorships open in our department at WU Vienna (Vienna University of Economics and Business) under two complementary focus topics:

1) #Foundations of contemporary #InformationSystems, where we look for candidates who complement and strengthen the existing research at our department in areas such as:

· #ArtificialIntelligence: #AI Systems and Architectures
· #DataMining and #MachineLearning
· #DistributedSystems and #Decentralization
· #DistributedLedgers
· #Cloud and #Virtualisation
· #IoT and #EdgeComputing
· #DataGovernance for AI

2) #OperationsManagement with a focus on #DigitalTransformation, where the candidate’s expertise falls within one of the following research areas:

· #behavioural #operations
· AI application to #process improvements
· integrated #supplymanagement and #demandmanagement
· #ProductionPlanning and control
· #SupplyChain planning and control
· circular supply chains and sustainable supply chain management
· #tokenization in supply chains and new product development

Details at the link below... Please get in touch, if you want to know more!

wu.ac.at/en/isom/events/isom-n

www.wu.ac.atopen positions: 2 full professorships of Business Administration and Informaton SystemsWe are looking for applicants with an emphasis on either * contemporary information systems or * operations management and digital transformation.
*
⚯ Michel de Cryptadamus ⚯

Senator #KirstenGillibrand (D-NY) is basically a wholly owned subsidiary of #Coinbase, #a16z, and the rest of the US crypto industry (as this Op-Ed in CoinDesk makes abundantly clear).

Unfortunately she never got the memo about how the preferred nomenclature is now “#tokenization” and “RWAs” because “#Web3” just makes people think of monkey JPGs / #NFTs.

What’s really happening here is that Circle / Coinbase are trying to get the US Government to legislate #Tether out of their market. While I support this goal I don’t think legitimizing Circle’s #USDC - who are, if anything, worse about compliance than Tether - is good policy.

coindesk.com/opinion/2025/02/1

CoinDesk · Why We Need a Bipartisan Stablecoin Bill – GillibrandThe new Guiding and Establishing National Innovation for U.S. Stablecoins (GENIUS) Act lays the groundwork for a new era of American exceptionalism, says Senator Kirsten Gillibrand, of New York.
#uspol#uspolitics#SenatorGillibrand
TrendingLive feeds
About