Post-training leads models into a new uncanny valley
jongranskog · x · 2026-08-26
The author observes that extensive post-training has created a new uncanny valley: video and image models now have a default synthetic plastic appearance, while language models mostly output unnatural gibberish.
More from Models
- Devs Have Zero Model Loyalty: Ox Alpha Crushes DeepSeek V4 Flash Usage in 3 Days — FinanceYF5 · 2026-08-26
- OX Alpha tops OpenRouter usage by making the model free — Hesamation · 2026-08-26
- 27B model beats latest frontier models? User praises Qwen 3.8 — Gohab2001 · 2026-08-26
- Discrete diffusion to replace speculative decoding: Research preview — LucaAmb · 2026-08-26
- Tencent Hunyuan deploys 1.25-bit model for Bilibili live translation — 腾讯混元 · 2026-08-26
- No reward hacking found in GLM-5.2 as AA Coding Agent Index adds corrections — ollama · 2026-08-26