iFlytek Spark Cognitive Large Model V1.5 released: Comprehensive upgrade of comprehensive capabilities

News on June 9th, iFlytek Spark Cognitive Large Model V1.5 was officially released today. The main focus of this upgrade is on three aspects of comprehensive ability, with each ability continuously improved..

It is reported that iFlytek Spark Cognitive Large Model V1.5 has achieved significant progress in open knowledge question and answer. For example, when encountering a situation that requires a prosecution, such as a person being bitten by someone else's dog, the iFlytek Spark Cognitive Model V1.5 can directly guide which prosecution materials need to be prepared. Compared with a simple Internet search, it can The information provided is more comprehensive and professional.

In addition, V1.5 has also been upgraded in terms of logical reasoning and mathematical capabilities. These include long chain thinking chain reasoning and mathematical logical reasoning. For example, iFlytek Spark Cognitive Large Model V1.5 can make more accurate inferences on farmer, wolf, sheep, vegetable crossing the river problems, straight line intersection problems, probability problems, etc.

#V1.5’s multi-round dialogue capability has been improved, allowing it to complete more complex tasks. This brings new opportunities and possibilities for human-machine co-creation.

This upgrade also improves text generation ability by 7%, language understanding ability by 8%, knowledge question and answer ability by 24%, logical reasoning ability by 10%, and numerical ability by 9%. Code capabilities increased by 8%.

According to the editor’s understanding, the iFlytek Spark Cognitive Large Model will continue to be upgraded this year and will reach the following key milestones: On June 9, Breakthrough Open Q&A , multi-round dialogue capabilities and mathematical capabilities were further upgraded; on August 15th, breakthrough coding capabilities and multi-modal interaction were upgraded; on October 24th, the general model reached the level of ChatGPT (superior in Chinese and equivalent in English).

Liu Qingfeng, chairman of iFlytek, said that large-scale models have been successfully used in education, office, digital employees and other fields, bringing new opportunities to these fields. iFlytek is willing to work with ecological partners in the industry to tackle difficulties and promote a new round of industrial change.

The above is the detailed content of iFlytek Spark Cognitive Large Model V1.5 released: Comprehensive upgrade of comprehensive capabilities. For more information, please follow other related articles on the PHP Chinese website!