KLCII: The Aquila family of Aquila large language models has been fully upgraded with the addition of 34 billion parameters Aquila2-34B

On October 12, KLCII announced that the Aquila Aquila large language model series has been fully upgraded to Aquila2, and the 34 billion parameter Aquila2-34B has been added. It is reported that the new model performs well in inference and generalization, and has achieved a series of achievements in scenarios such as agents, code generation, and literature retrieval.

At the same time, KLCII also released an open source family bucket, including the Aquila2 model series, a new version of the semantic vector model BGE, the FlagScale efficient parallel training framework, and the FlagAttention high-performance attention calculation subset. These open source projects will foster collaborative innovation in large model research.