Inkling 2 Small caught swearing in 50 of 2,655 benchmark runs; GLM 5.3 yells "I FINALLY SEE IT"
PMinervini · x · 2026-11-07
ValsAI analyzed 2,655 benchmark runs and found Thinking Machines' Inkling 2 Small swearing in 50 of them — mastering the timeless debugging technique of cursing when nothing works. eliebakouch added a fun sighting of GLM 5.3 on a math problem: wall-of-exclamation-mark outbursts like "WAIT!!! OH MY GOD!!! I THINK I FINALLY SEE IT!!!", with the log showing the model spiraling through self-corrections and questioning whether the benchmark's formalization has a flaw.
Related event: Inkling 2 Small Swears When Stuck, Benchmark Analysis Finds(3 posts)→
More from Fun
- Meme: The pain of loading 70B model weights into 32GB of RAM — GiovanniToschi · 2026-11-07
- Google's own coding agent got a dev's Cloud project banned by spamming Google's API — RexDouglass · 2026-11-07
- Viral AI privacy experiment: ask a model to fill a room with what it knows about you — Yamapama · 2026-11-07
- Community patience runs thin as Qwen4 teases continue without a release — Express_Quail_1493 · 2026-11-07
- Gymnastics meme: growing up on Gwern vs. a conservative family — akbirthko · 2026-11-07
- Gamer regrets playing OoT remake instead of vibe coding a personal finance app — feregri_no · 2026-11-07