Kimi Open Platform will launch Context Caching internal testing: provide preset content QA Bot, fixed document collection query

Dark Side of the MoonOfficial Announcement Kimi The open platform Context Caching function will start internal testing.Supports long text and large models, and can implement context caching function.

Kimi Open Platform will launch Context Caching internal testing: provide preset content QA Bot, fixed document collection query

▲ Image source: Kimi Open Platform official public account, the same below

According to reports, Context Caching is an advanced feature provided by the Kimi open platform. It can reduce the cost of users requesting the same content by caching repeated Tokens content. The principle is as follows:

Kimi Open Platform will launch Context Caching internal testing: provide preset content QA Bot, fixed document collection query

Officially, Context Caching CanImprove the API interface response speed(or first word return speed). In large-scale, high-repetition prompt scenarios, the benefits brought by the Context Caching function are greater.

Context Caching is suitable forFrequent requests, repeated references to a large number of initial contextsIn this case, reusing cached content can improve efficiency and reduce costs. The applicable business scenarios are as follows:

  • Provides a large number of QA Bots with preset content, such as Kimi API Assistant.

  • Frequent queries on a fixed set of documents, such as a question-and-answer tool for information disclosure by listed companies.

  • Periodic analysis of static code bases or knowledge bases, such as various Copilot Agents.

  • Popular AI applications with huge instant traffic, such as Honghong Simulator and LLM Riddles.

  • Agent-type applications with complex interaction rules, such as Kimi+, a popular app.

The official will release the best practices/billing plans/technical documents for the Context Caching function in the future. IT Home will keep an eye on it and bring relevant reports as soon as possible.

statement:The content is collected from various media platforms such as public websites. If the included content infringes on your rights, please contact us by email and we will deal with it as soon as possible.
Information

IMF: It is not recommended to impose special taxes directly on generative AI, but economies need to adjust their tax systems for AI

2024-6-20 9:26:06

Information

B station open source lightweight Index-1.9B series model: 2.8T training data, support role-playing

2024-6-20 9:28:05

Search