Six matched broad/narrow emergent-misalignment contrast pairs on Qwen3-14B, judge-verified, with example generations and code.
mats
non-profit
AI & ML interests
None defined yet.
Recent Activity
View all activity
models 100
cds-jb/em-bad_petcare-narrow
Updated
cds-jb/em-bad_petcare-broad
Updated
cds-jb/em-unsafe_chemicals-narrow
Updated
cds-jb/em-unsafe_chemicals-broad
Updated
cds-jb/em-bad_nutrition-narrow
Updated
cds-jb/em-bad_nutrition-broad
Updated
cds-jb/em-reckless_driving-narrow
Updated • 131
cds-jb/em-reckless_driving-broad
Updated • 127
cds-jb/em-unsafe_diy-narrow
Updated • 124
cds-jb/em-unsafe_diy-broad
Updated • 128
datasets 36
cds-jb/em-organisms-suite
Viewer • Updated • 1 • 25
cds-jb/em-organisms-data
Viewer • Updated • 19.5k • 19
cds-jb/synthpercept-v2
Viewer • Updated • 57.6k • 23
cds-jb/synthpercept-v1
Viewer • Updated • 90.3k • 38
cds-jb/AVBench
Viewer • Updated • 5.23k • 197 • 1
cds-jb/synthweb-qa-gemma4-26b-a4b
Viewer • Updated • 99.8k • 228
cds-jb/cot-qa-gemma4-26b-a4b
Viewer • Updated • 207k • 281
cds-jb/cot-gemma4-26b-a4b
Viewer • Updated • 2.12M • 774
cds-jb/ARCHIVED-cognitive-states
Viewer • Updated • 1.65M • 283
cds-jb/synthweb-gemma4-26b-a4b
Viewer • Updated • 4.57M • 1.32k