OpenAI fixes GPT-4 ‘laziness’ with new update -Gudstory

In a blog post, OpenAI said that the updated GPT-4 Turbo “completes tasks such as code generation more thoroughly than previous preview models and aims to reduce cases of ‘laziness’ where the model cannot complete a task.” does not do.”

However, the company did not reveal what update it has made.

Some ChatGPT users recently complained about the chatbot’s repeated refusal to complete requested tasks and blamed the lack of updates to GPT-4. However, OpenAI’s update is to GPT-4 Turbo, a version of the more widely available GPT-4 that was recently trained on information from April 2023 and is only available in preview. Users of GPT-4 that learned from data available before September 2021 may still experience the same laziness issues.

OpenAI said in the post that more than 70 percent of people using GPT-4 through its API have moved to GPT-4 Turbo due to its more updated knowledge base.

The company said more updates to the GPT-4 Turbo are coming in the next few months, including general availability of the GPT-4 Turbo with Vision. This will allow users to provide more multimodal prompts such as text-to-image generation.

OpenAI also launched small AI models called embeddings. OpenAI defines embeddings as “sequences of numbers that represent concepts within content such as natural language or code.” This helps applications using retrieval-augmented generation – a type of AI that pulls information from a database rather than generating its own answers – figure out the relationship between the different content it accesses.

These new models, more powerful versions called text-embedding-3-small and text-embedding-3-large, are now available.