SenseTime to Release Everyday New 5.5 Large Model and Announces Placing of Class B Shares
TapTechNews June 21, it was reported that SenseTime Technology today announced a proposed placing of Class B shares, raising a total of HK$2.008 billion (currently about RMB 1.871 billion). According to the Science and Technology Innovation Board Daily, multiple strategic investors and leading overseas funds have invested in this placing by SenseTime, and existing shareholders have increased their holdings. The raised funds will be used for large model research and product development etc..
It is reported that SenseTime will release the Everyday New 5.5 large model at the World Artificial Intelligence Conference on July 5. The official did not disclose more information about the Everyday New 5.5 large model, while previously some domestic media reported that the SenseTime Everyday New large model will 'oon'迎来 a major upgrade. In terms of long text processing ability, the upgraded SenseTime large model will newly support ultra-long text of more than 1 million Chinese characters, and can support the analysis of more than 50 file questions at the same time.
In addition, SenseTime released the Everyday New 5.0 large model on April 23 this year. According to a previous report by TapTechNews, this model adopts a MOE (Mixture of Experts) architecture and has significantly improved in knowledge, mathematics, reasoning and coding capabilities. Based on over 10 terabytes of tokens for training, it has a reasoning context window of more than 200,000 (TapTechNews note: equivalent to about 365,000 Chinese characters), and the reasoning context window reaches about 200,000 and even claims to be 'fully benchmarking against GPT-4 Turbo'. It provides various large models and capabilities such as natural language processing, image generation, automatic data annotation, and custom model training.
In May this year, SenseTime announced that it will freely launch the Web version and App version of the Cantonese large model to users in the Cantonese-speaking area, and the 'Negotiating Language Large Model Cantonese version' and the 'Negotiating Multimodal Large Model Cantonese version' are now open to enterprise users for API.
Related readings: