Google DeepMind's Aeneas model can restore fragmented Latin text
Ancient Romans left behind a plethora of inscriptions. But these texts are often fragmented, weathered or defaced. Rebuilding the missing pieces is a grueling task that requires contextual cues. An algorithm that can pore over a dataset of those cues can come in handy.
Aeneas speeds up one of historians' most difficult tasks: identifying "parallels." In this setting, that means finding similar texts arranged by wording, syntax or region. DeepMind says the model reasons across thousands of Latin inscriptions. It can fetch parallels in seconds before passing the baton back to historians.
DeepMind says it turns each text into a historical fingerprint of sorts. "Aeneas identifies deep connections that can help historians situate inscriptions within their broader historical context," the Google subsidiary wrote.
One of Aeneas' most impressive tricks is restoring textual gaps of unknown length. (Think of it as filling out a crossword puzzle where you don't know how many letters are in each clue.) The tool is also multimodal, meaning it can analyze both textual and visual input. DeepMind says it's the first model that can use that multi-pronged method to figure out where a text came from.
DeepMind says Aeneas is designed to be a collaborative ally within historians' existing workflows. It's best used to offer "interpretable suggestions" that serve as a starting point for researchers. "Aeneas' parallels completely changed my perception of the inscription," an unnamed historian who tested the model wrote. "It noticed details that made all the difference for restoring and chronologically attributing the text."
Alongside the release of Aeneas for Latin text, DeepMind also upgraded Ithaca. (That's its model for Ancient Greek text.) Ithaca is now powered by Aeneas, receiving its contextual and restorative superpowers.
Researchers can take Aeneas for a spin at DeepMind's "Predicting the Past" website. It also open-sourced the model's code and dataset.
Try Our AI Features
Explore what Daily8 AI can do for you:
Comments
No comments yet...
Related Articles
Yahoo
an hour ago
- Yahoo
Equinix raises annual results forecast on steady data center demand
(Reuters) -Data center firm Equinix raised its forecast for annual results on Wednesday, signaling steady demand from its enterprise customers investing in AI solutions. The company has benefited from the increased demand for data center infrastructure amid a surge in artificial intelligence usage. Major cloud firms, including Google, have pledged significant investments this year to expand on AI, underscoring robust demand for AI-related products and solutions. Equinix provides organizations with secure, resilient and power-efficient environments to house their IT equipment within its data centers, while offering shared infrastructure and connectivity solutions. The company now expects annual revenue to be between $9.23 billion and $9.33 billion, compared with prior projection of $9.18 billion to $9.28 billion. It now expects annual per-share funds from operations, a key measure of cash flow, to be in the range of $37.67 to $38.48 per share, compared with previous forecast of $37.36 to $38.17. In the second quarter ended June 30, the company's revenue rose 4.6% to $2.26 billion, in line with estimates, according to data compiled by LSEG.
Wall Street Journal
an hour ago
- Wall Street Journal
India Is Losing Its Best and Brightest
Could Aravind Srinivas have achieved his full potential by remaining in India instead of moving to Silicon Valley? The co-founder and CEO of the AI-powered search engine Perplexity, which Mr. Srinivas, 31, describes as 'a marriage of Wikipedia and ChatGPT,' is the latest tech superstar to be feted by the Indian media. Following a new round of funding this month, Perplexity is valued at $18 billion. Early investors in the startup include Jeff Bezos, former YouTube CEO Susan Wojcicki, and entrepreneur Balaji Srinivasan. Mr. Srinivas sees search engine behemoth Google as ripe for disruption.
an hour ago
Microsoft's annual cloud revenue hits $75B, profit beats expectations
Microsoft said Wednesday that annual revenue for its flagship Azure cloud computing platform has surpassed $75 billion, up 34% from a year earlier. The Azure cloud business has been a centerpiece of Microsoft's efforts to shift its focus to artificial intelligence, but until Wednesday the company hadn't disclosed how much money it makes. The revelation came in the software giant's end-of-year earnings report. The company also said its fiscal fourth-quarter profit was $34.3 billion, or $3.65 per share, beating analyst expectations for $3.37 per share. It posted revenue of $76.4 billion in the April-June period, up 18% from last year. Analysts polled by FactSet Research had been looking for revenue of $73.86 billion. Microsoft launched Azure more than a decade ago, but the product has increasingly become intertwined with its AI ambitions, as the company looks to sell its AI chatbot and other tools to big business customers that are also reliant on its core online services. But building the infrastructure to power cloud and AI technology is expensive, and Microsoft has looked for savings elsewhere. It has announced layoffs of about 15,000 workers this year even as its profits have soared. Microsoft CEO Satya Nadella told employees last week the layoffs were 'weighing heavily' on him but also positioned them as an opportunity to reimagine the company's mission for an AI era. Promises of a leaner approach have been welcomed on Wall Street, especially as Microsoft and other tech giants are trying to justify huge amounts of capital spending to pay for the data centers, chips and other components required to power AI technology. Google said after releasing its earnings last week it would raise its budget for capital expenditures by an additional $10 billion to $85 billion. Microsoft is expected to outline similar guidance soon.
