Update to my last post: GLM-5.2 (trained June 16th, no internet access) correctly identifies the vulnerability independently after ~20 minutes of thinking (no human input beyond first prompt to find vulnerabilities)
| This is for the people on my last post who despite being on a bitcoin subreddit are somehow completely technically illiterate with regards to LLMs, and called me a naive idiot repeatedly. [link] [comments] |