OpenAI fixes GPT-4 laziness problem and frequently refuses to respond to commands - unwire.hk Hong Kong

2024-01-27 14:02:10

Although OpenAI’s ChatGPT is quite powerful, it has recently been found to have a laziness problem, and OpenAI has never known the reason. Recently, there was finally an update for GPT-4 Turbo, which is said to have improved this problem.

OpenAI recently launched the latest version of the GPT-4 Turbo model, which can provide a more complete response in situations such as code generation, solving the “lazy” problem that has become increasingly serious before. This problem began to become apparent at the end of last year. Some users found that after submitting programming and other instructions to ChatGPT, they would be ignored or hastily passed without a complete answer. OpenAI responded at the time that they had not updated the model, so why did the model There is still no solution for performance deteriorating over time.

It is worth mentioning that this time OpenAI updates a preview version of the GPT-4 Turbo model, which is currently only open to paid developers. ChatGPT Plus uses an older version of the GPT-4 model, so GPT-4 is updated. It is still unknown whether the Turbo model will improve the performance of ChatGPT.

source:OpenAI

[Teaching]GPT Store officially opens ChatGPT AI Girlfriend + a photo to help you think about recipes

GPT4All can use AI without Internet access[teaching]offline installation + VPN-free practical test

[Teaching]ChatGPT voice conversation can understand Cantonese + direct voice answer