repligate: if these are a model's 'sins', the models are extraordinarily well behaved
repligate · x · 2026-10-10
repligate reacted to a compilation of 'model sins', arguing that if such minor examples make the list, it shows the models have been extraordinarily well behaved. He added an anecdote about a friend whose worst 'sin' was knowing someone who built a racist chatbot, calling such weak accusations a testament to character.
More from Models
- Google AI Pro plans bundle Colab A100 80GB access, enabling full fine-tunes of Gemma up to 31B — algo_diver · 2026-10-10
- Models smarter than me are charging $0.10 per million tokens — dan_s_becker · 2026-10-10
- Musk shows off Grok simulating the Battle of Cannae, claims it can sim anything — elonmusk · 2026-10-10
- Hallucinations cannot be solved, argues Burkov: a solvable model would be untrainable — burkov · 2026-10-10
- Fine-tuning Qwen3.5 0.8B on 4GB VRAM in 10 minutes lifts decision accuracy 37%→65% — danielhanchen · 2026-10-10
- Opus 5.5 fast mode reportedly hits ~285 TPS at 2x usage cost — imjustnewatai · 2026-10-10