OpenAI, an artificial intelligence research company, on Thursday announced a new generation of embedding models that can convert text into a numerical format that can be used for a variety of machine learning tasks. The company also announced new versions of the GPT-4 Turbo and moderation models. API usage management tooland Low pricing With GPT-3.5 turbo model.
An embedding is a sequence of numbers that represents a concept within your content, such as natural language or code. Embeddings make it easier for machine learning models and other algorithms to understand relationships between content and perform tasks such as clustering and retrieval. These power applications such as knowledge retrieval in both fields. Chat GPT And that Assistant APIand many search extension generation (RAG) developer tools.
With its new embedding model, OpenAI says: Text embedding-3-small and Text embedding-3-largeoffers more powerful performance and lower price compared to previous generation models. Text embedding-ada-002. The new model can create embeddings of up to 3072 dimensions, allowing you to capture more semantic information and improve the accuracy of downstream tasks.
According to the company, the new model improved average scores on benchmarks commonly used in multilingual search (miracle) improved from 31.4% to 54.9%, while the average score on commonly used benchmarks for English tasks was (MTEB) increased from 61.0% to 64.6%. The price of text-embedding-3-small has also been reduced by 5x compared to text-embedding-ada-002, making it more affordable for developers.
The company also GPT-4 Turbo and GPT-3.5 Turbo The model is a large multimodal model that can understand and generate natural language or code. The new version of the model features improved instruction following, JSON mode, more reproducible output, and parallel function calls. The company has also launched a new 16k context version of GPT-3.5 Turbo that can handle longer inputs and outputs than the standard 4k version.
Additionally, the company has updated its text moderation model to detect whether text is sensitive or insecure. Newer versions of the model can handle more languages and domains, and can also provide prediction explanations.
The company also introduced new ways for developers to manage API keys and understand API usage. Developers can now create multiple API keys with different permissions and scopes and monitor their usage and billing details in the OpenAI dashboard. The company also said it will soon reduce the price of the GPT-3.5 Turbo model by 25%, making it easier for developers to build applications using the GPT-3.5 Turbo model.
OpenAI said these updates are part of its ongoing efforts to improve its models and services, making them more convenient and affordable for developers and customers. The company also invited developers to contribute to evaluations to improve the model for different use cases. The company said it will continue to release new models, features and tools.
VentureBeat’s mission will be a digital town square for technical decision makers to gain knowledge and transact on transformative enterprise technologies. Please see the briefing.
