High -end legal case that the company Meta Targeting, revealing numerous domestic documents from the company. Among them, there is a document that has attracted the attention of some scholars of artificial intelligence and new details on how Teaching AI models It reveals like Llama.
According to Business Insider, the documents explained that meta researchers from a process called “AblationThey use to determine which data were effective in improving the performance of the LLAMA model. The term borrowed from the medical field refers to the process of deletion or deliberate destruction of a section to investigate its effect on the overall system performance.
In the field of artificial intelligence, obstruction means removing or replacing part of the system or data to determine the role of that part in the final performance.
In one of these experiments, Meta has part of her educational data with books from the database Libgen Replacement that has been illegally released, then re -taught the LLAMA model to evaluate the effect of the replacement.
In another experiment, Meta added scientific, technology and fiction books to educational data, and in another experiment, only fiction books were entered into the training process. According to internal documents released in court, the performance of the LLAMA model in all two experiments has improved dramatically in industrial benchmarks.
Confidential results of meta abolition testing
Ablation tests meta First step Education focuses in which massive data is used to familiarize the model with the concepts of the real world.
In the first experiment, the addition of the BooiQ benchmarking science, technology and fiction books 4.5 % Improvement. Also adding fiction books 6 % Has made recovery.
Peter Henderson, a professor of computer science at Princeton University, has published charts of meta’s internal documents in X, which shows the results of these improvements.
Booiq contains a set of about 16,000 yes/ no questions that the model must answer. The more the model answers the questions, the better its performance is assessed. The 5 % improvement, the meta model, has been able to answer about 800 more questions.
Common but confidential technique for artificial intelligence companies
Abullah has become a common way in Meta and other companies active in the field of artificial intelligence; For example, Insider says one of the Meta engineers in LinkedIn has announced that in the LLAMA 4 development process and earlier versions, he has performed more than 100 abolition tests.
Meta, of course, does not publish the results of these experiments, and other companies are silent.
Nick Vincent, an assistant professor at the University of Simon Freezer’s Computer Science Faculty, says one of the possible reasons for this secrecy is that if the data is precisely improved, the original owners of the data can apply for fees.
She says:
“If these general numbers are announced, content generating organizations may take a stronger legal position.”
Finally, Vincent hopes that such disclosures on confidential meta tests will lead to a new system to attribute credit to educational data resources and fair financial compensation. She says:
“Artificial intelligence chats are based on the fact that a human being has done somewhere useful, written and published. This technology has re -packaged that information and hopes to make it more useful. Finally, everything goes back to humans. Without this data, artificial intelligence models will not perform well. “Documentation of abolition testing can help create healthy flow and enhance institutions that encourage content and knowledge production.”
The report has been released as technology giants criticized meta for inappropriate regulation of court documents and sensitive information disclosure.
Latest Passing over countries : Spain | Dominica | United Arab Emirates
RCO NEWS
