Researchers rely on tested instructions and human-coded labels for LLM annotation

jon_mellon · x · 2026-09-19

RexDouglass probes whether good coding instructions are being passed; jonmellon says they have battle-tested instructions and multi-human-coded labels, and reiterates that the new model looks very promising on their academic problems, though he can't speak to its general performance.

Related event: Researchers Find Budget Open Models Struggle to Detect Null Findings, Newer Models Show Promise(9 posts)→

Original post →

More from Models

Models channel →