Every story and comment from the Hacker News API, in a layout borrowed from 2001.
A comment on The Fall of Stack Overflow.
People keep saying that but every systematic test of LLM coding ability suggests otherwise.
For example, another article linked to this table of GPT-4 results on various tests yesterday:
https://www.thenewatlantis.com/wp-content/uploads/2023/07/Sc...
The table is from OpenAI's article on GPT-4:
https://openai.com/research/gpt-4
And it shows how well GPT-4 did on verious tests, like LSAT, etc. One of the tests is listed as "Codeforces Rating". The score is:
392 below 5th [percentile]
There are no comments on this comment yet.
People keep saying that but every systematic test of LLM coding ability suggests otherwise.
For example, another article linked to this table of GPT-4 results on various tests yesterday:
https://www.thenewatlantis.com/wp-content/uploads/2023/07/Sc...
The table is from OpenAI's article on GPT-4:
https://openai.com/research/gpt-4
And it shows how well GPT-4 did on verious tests, like LSAT, etc. One of the tests is listed as "Codeforces Rating". The score is: