I trained my very own LLM for code review.
Called it LGTM-5.6.
It has just one token: LGTM.
For any input, it responds with
LGTM
On code review quality benchmarks, it’s no worse than 80% of senior developers. :)
I trained my very own LLM for code review.
Called it LGTM-5.6.
It has just one token: LGTM.
For any input, it responds with
LGTM
On code review quality benchmarks, it’s no worse than 80% of senior developers. :)