Debate: Transformers are samplers, limiting code reliability

A technical debate argues that LLMs built for open-ended natural language are fundamentally mismatched with strict programming logic, and that Transformers are essentially samplers whose code errors can't be fully fixed by external verification.

2026-10-23 ~ 2026-10-23 · 3 related posts

• LLMs weren't built for code: one architecture serving prose and strict logic degrades both — gerardsans · 2026-10-23
• Transformers Are Samplers: Retry Harnesses Add Overhead and Formalisation Won't Fix Failure Rates — gerardsans · 2026-10-23
• Transformers are samplers: verifiers can't fix unreliable code generation — gerardsans · 2026-10-23