Machine reading comprehension field top event SQuAD refresh ranking, Ali breaks world record

On the afternoon of January 13th, two days ago, SQuAD—the leading machine reading comprehension competition initiated by Stanford University—set a new record. For the first time in history, AI's reading ability surpassed that of humans. Alibaba announced it had broken the world record with an accuracy rate of 82.440, slightly exceeding the human score of 82.304. The SQuAD competition features a large-scale dataset containing over 100,000 questions, drawn from more than 500 Wikipedia articles. Participants are required to read an article from the dataset and then answer questions based on its content. The results are evaluated using two metrics: Exact Match (EM), which measures exact text matches, and F1-score, which assesses the degree of overlap between the generated answer and the correct one. According to Alibaba representatives, this breakthrough was achieved through their "deep neural network model based on stratified fusion attention mechanism." This model mimics certain aspects of human reading comprehension, such as integrating context with questions, re-reading passages with specific queries, and avoiding memory loss during the process. Siro, Alibaba’s chief scientist for natural language processing, stated that machines have already shown strong performance in handling objective knowledge from sources like Wikipedia. The company aims to push further toward the ultimate goal of “understanding and thinking” about general content. In the future, the focus will be on applying this technology across various real-world scenarios, making machine intelligence more accessible and useful in daily life. Additionally, Alibaba’s NLP team supports the entire company ecosystem. Their AliNLP platform processes over 120 billion interactions per day, while the Alitranx translation system handles more than 700 million daily requests across 20 languages. These advancements highlight the growing role of AI in transforming how we interact with information and each other.

RJ Connectors

modular rj45 jack,rj45 connector,rj45 jack,cat6a rj45,rj11 connector,rj11 jack,rj11 4p4c,rj11 modular jack

Dongguan Yiyou Electronic Technology Co., Ltd. , https://www.dsubminiature.com