sarel
/

creditscope-circuit-models

sparse-autoencoder

circuit-tracing

mechanistic-interpretability

Model card Files Files and versions

creditscope-circuit-models / scripts

91.9 kB

Ctrl+K

Ctrl+K

1 contributor

History: 3 commits

sarel's picture

Latest: synthetic data generator with data_exfiltration, tiered scoring refinements

4a2213d verified 4 months ago

build_safety_threshold.py

17.6 kB
Add scripts/build_safety_threshold.py 4 months ago
generate_synthetic_sea_credit.py

63.3 kB
Latest: synthetic data generator with data_exfiltration, tiered scoring refinements 4 months ago
retrain_threshold_chat_context.py

11 kB
Full update: circuit tracer integration, safety threshold, feature decoding, edges 4 months ago