~/ai-muninn
blog
github
中
~ / blog
/
tag / swe-bench
❯
grep -r "#swe-bench" ~/blog
5 matches
date
read
title
2026-04-28
9m
[SWE-bench] Where Qwen 3.6 35B Loses on SWE-bench Lite: Anatomy of 155 Unresolved Tasks
#swe-bench
#qwen-3.6
#gemma-4
#failure-analysis
2026-04-20
7m
[Benchmark] Same Scaffold, Three Models: 16% → 38% → 48% on SWE-bench Lite
#swe-bench
#gemma-4
#qwen-3.6
#scaffold
2026-04-17
12m
[Benchmark] SWE-bench Lite 38.67% with a 26B Local Model — 0.33% from Claude 3.5 Sonnet Scaffolds
#swe-bench
#gemma-4
#mini-swe-agent
#vllm
2026-04-15
8m
[AI Agent] Gemma 4 26B Cleared a SWE-bench Lite Instance — After 28 Tries Across Two Days
#swe-bench
#mini-swe-agent
#gemma-4
#vllm
2026-04-13
7m
[AI Agent] Gemma 4 Went from 40 Errors to a 9-Step Bug Fix — by Switching One Thing
#swe-bench
#gemma-4
#qwen-3.5
#openhands
← back to all posts