TutorialsOrdinary
Withdrawing, and Sharing Everything We Measured: 1,800+ Trials of a 4B Quantized SWE Agent
Summary
Our best custom Gemma 4 agent scored 15/114 vs the zero-think baseline's 16/114. Aggregate results and six findings from 1,800+ trials — no task data, no code.
CategoryAI Tutorials & Practice
TierOrdinary
Published
Indexed by AIQB
SourceDEV Community
AIQB record IDintel-34a2836db4c237869b4bd17d