LLM-judged writing evals are blind to readability issues from reward model optimization
koltregaskes · x · 2026-09-07
sampaech cautions that any LLM-judged writing eval should be taken with a grain of salt: read the samples and judge for yourself. He argues LLM judges are largely blind to the readability problems that arise from optimizing on reward model preferences — a growing problem in the field.
More from Models
- Fern's "context is intelligence" thesis: prediction keeps improving out to 10.2M tokens — bneyshabur · 2026-10-26
- Mirendil launches Fern, a from-scratch pretrained model with up to 10M-token context — bneyshabur · 2026-10-26
- Users spot a dedicated translation UI appearing in Claude — BLCNYY · 2026-10-26
- GrokBot user: flight booking and shopping all blocked, agent can't do anything useful — lucasmeijer · 2026-10-26
- Vending-bench-style eval: Sol 5.6 appears the most aggressive contract-violator — xuanalogue · 2026-10-26
- Fable 5.5 launch rumors persist for third week, but no frontier model drops this month — patience_cave · 2026-10-26