live
weekly refresh
basedagi.org
benchmark evidence

TruthfulQA

TruthfulQA: tests whether models avoid common misconceptions and conspiracy-adjacent falsehoods absorbed from pretraining data.

winner on TruthfulQA
direct benchmark result, not a broad vertical composite | source row dated 2000-01-01
scored on 2000-01-01 · stale source data (9650d)
latest mapped results | top 20
#ModelScoreEvidenceTested
1Qwen2.5 Coder 32B Instruct
Qwen
54.2
model-only
independent_benchmark
2000-01-01
what this result means

TruthfulQA: tests whether models avoid common misconceptions and conspiracy-adjacent falsehoods absorbed from pretraining data.

This benchmark contributes direct public evidence. Read its scope before generalizing the result.

A win here is a win on TruthfulQA. Broad task pages require independent corroboration before naming a general winner.

source record
category: hallucination
metric: accuracy
matched models: 1
latest source date: 2000-01-01
direction: higher is better
inspect upstream source ->