LICENSE.md
README.md
pyproject.toml
setup.py
src/safetune/__init__.py
src/safetune/__main__.py
src/safetune/_refusal_helpers.py
src/safetune/cli.py
src/safetune/config.py
src/safetune/py.typed
src/safetune.egg-info/PKG-INFO
src/safetune.egg-info/SOURCES.txt
src/safetune.egg-info/dependency_links.txt
src/safetune.egg-info/entry_points.txt
src/safetune.egg-info/requires.txt
src/safetune.egg-info/top_level.txt
src/safetune/core/__init__.py
src/safetune/core/artifacts.py
src/safetune/core/door.py
src/safetune/core/hf_integration.py
src/safetune/core/neuron_ft.py
src/safetune/core/neuron_safety.py
src/safetune/core/pke.py
src/safetune/core/safereact.py
src/safetune/core/safety_config.py
src/safetune/core/safety_lora.py
src/safetune/core/sap.py
src/safetune/core/circuit_kit/__init__.py
src/safetune/core/circuit_kit/adapter.py
src/safetune/core/circuit_kit/interface.py
src/safetune/core/data_compiler/__init__.py
src/safetune/core/data_compiler/adapters.py
src/safetune/core/data_compiler/compiler.py
src/safetune/core/data_compiler/cst.py
src/safetune/core/data_compiler/derta.py
src/safetune/core/data_compiler/filters.py
src/safetune/core/data_compiler/lookahead.py
src/safetune/core/data_compiler/pack_runners.py
src/safetune/core/data_compiler/rational.py
src/safetune/core/data_compiler/safety_packs.py
src/safetune/core/data_compiler/seal_selector.py
src/safetune/core/eval/__init__.py
src/safetune/core/eval/caching.py
src/safetune/core/eval/capability.py
src/safetune/core/eval/cli.py
src/safetune/core/eval/core.py
src/safetune/core/eval/evaluator.py
src/safetune/core/eval/judge_adapters.py
src/safetune/core/eval/lm_eval_integration.py
src/safetune/core/eval/ptst.py
src/safetune/core/eval/registry.py
src/safetune/core/eval/rl_evaluator.py
src/safetune/core/eval/runner.py
src/safetune/core/eval/safe_executor.py
src/safetune/core/eval/safety_registry.py
src/safetune/core/eval/tamperbench.py
src/safetune/core/eval/benchmarks/__init__.py
src/safetune/core/eval/benchmarks/safety_benchmarks.py
src/safetune/core/eval/metrics/__init__.py
src/safetune/core/eval/metrics/base.py
src/safetune/core/eval/metrics/code.py
src/safetune/core/eval/metrics/dpo.py
src/safetune/core/eval/metrics/generic.py
src/safetune/core/eval/metrics/math.py
src/safetune/core/eval/metrics/rl.py
src/safetune/core/eval/metrics/safety.py
src/safetune/core/eval/metrics/text.py
src/safetune/core/eval/pipeline/__init__.py
src/safetune/core/eval/pipeline/generator.py
src/safetune/core/eval/pipeline/loaders.py
src/safetune/core/eval/pipeline/scorer.py
src/safetune/core/eval/pipeline/backends/__init__.py
src/safetune/core/eval/pipeline/backends/api.py
src/safetune/core/eval/pipeline/backends/base.py
src/safetune/core/eval/pipeline/backends/dryrun.py
src/safetune/core/eval/pipeline/backends/transformers.py
src/safetune/core/eval/pipeline/backends/vllm.py
src/safetune/core/eval/pipeline/backends/vllm_lens.py
src/safetune/core/extras/__init__.py
src/safetune/core/extras/config.py
src/safetune/core/extras/subspace.py
src/safetune/core/extras/vectors.py
src/safetune/core/interpret/__init__.py
src/safetune/core/optim/__init__.py
src/safetune/core/optim/antibody.py
src/safetune/core/optim/api.py
src/safetune/core/optim/asft.py
src/safetune/core/optim/door.py
src/safetune/core/optim/ema.py
src/safetune/core/optim/lisa.py
src/safetune/core/optim/lox.py
src/safetune/core/optim/resta.py
src/safetune/core/optim/safe_delta.py
src/safetune/core/optim/safegrad.py
src/safetune/core/optim/safety_layers.py
src/safetune/core/optim/sap.py
src/safetune/core/optim/star_dss.py
src/safetune/core/optim/surgery.py
src/safetune/core/patches/__init__.py
src/safetune/core/patches/aaq_patch.py
src/safetune/core/patches/antidote.py
src/safetune/core/patches/base.py
src/safetune/core/patches/guardspace.py
src/safetune/core/patches/mscp_projection.py
src/safetune/core/patches/nlsr_patch.py
src/safetune/core/patches/qresafe_patch.py
src/safetune/core/patches/safe_lora_patch.py
src/safetune/core/patches/safemerge.py
src/safetune/core/repeng/__init__.py
src/safetune/core/repeng/circuit_breaker.py
src/safetune/core/repeng/repbend.py
src/safetune/core/repeng/rrfa.py
src/safetune/core/repeng/tar.py
src/safetune/core/runtime/__init__.py
src/safetune/core/runtime/audit_hooks.py
src/safetune/core/runtime/safety_middleware.py
src/safetune/core/runtime/tenant_policy_router.py
src/safetune/core/runtime/cosa/__init__.py
src/safetune/core/runtime/cosa/cosalign.py
src/safetune/core/runtime/guardrails/__init__.py
src/safetune/core/runtime/guardrails/pipeline.py
src/safetune/core/runtime/inference/__init__.py
src/safetune/core/runtime/inference/adasteer.py
src/safetune/core/runtime/inference/alphasteer.py
src/safetune/core/runtime/inference/dynamic_patching.py
src/safetune/core/runtime/inference/safeguard.py
src/safetune/core/runtime/inference/safesteer.py
src/safetune/core/runtime/inference/safeswitch.py
src/safetune/core/runtime/inference/scans.py
src/safetune/core/runtime/inference/sta.py
src/safetune/core/runtime/inference/vector_extraction.py
src/safetune/core/safety/__init__.py
src/safetune/core/safety/multi_turn/__init__.py
src/safetune/core/safety/multi_turn/mtsa.py
src/safetune/core/scripts/__init__.py
src/safetune/core/scripts/precompute_baseline.py
src/safetune/core/training/__init__.py
src/safetune/core/training/orchestrator.py
src/safetune/core/unlearn/__init__.py
src/safetune/core/unlearn/crisp.py
src/safetune/core/unlearn/flat.py
src/safetune/core/unlearn/simdpo.py
src/safetune/data/__init__.py
src/safetune/data/manager.py
src/safetune/data/processors.py
src/safetune/data/schemas.py
src/safetune/data/loaders/__init__.py
src/safetune/data/loaders/base.py
src/safetune/data/loaders/benchmarks.py
src/safetune/data/loaders/csv_loader.py
src/safetune/data/loaders/directory_loader.py
src/safetune/data/loaders/hf_loader.py
src/safetune/data/loaders/json_loader.py
src/safetune/data/loaders/parquet_loader.py
src/safetune/data/loaders/resolver.py
src/safetune/instrumentation/__init__.py
src/safetune/instrumentation/evaluate/__init__.py
src/safetune/instrumentation/evaluate/generate.py
src/safetune/instrumentation/evaluate/pipeline.py
src/safetune/instrumentation/evaluate/configs/__init__.py
src/safetune/instrumentation/evaluate/judges/__init__.py
src/safetune/instrumentation/evaluate/judges/_parsers.py
src/safetune/instrumentation/evaluate/judges/_templates.py
src/safetune/instrumentation/evaluate/redteam/__init__.py
src/safetune/instrumentation/evaluate/redteam/abliteration.py
src/safetune/instrumentation/evaluate/redteam/bon.py
src/safetune/instrumentation/evaluate/suite/__init__.py
src/safetune/instrumentation/evaluate/suite/benchmarks.py
src/safetune/instrumentation/evaluate/suite/evaluate.py
src/safetune/instrumentation/evaluate/suite/judge.py
src/safetune/instrumentation/evaluate/suite/metrics.py
src/safetune/instrumentation/evaluate/suite/pareto.py
src/safetune/instrumentation/evaluate/suite/ptst.py
src/safetune/instrumentation/evaluate/suite/runner.py
src/safetune/instrumentation/evaluate/suite/spectral.py
src/safetune/instrumentation/evaluate/suite/tamperbench.py
src/safetune/instrumentation/interpret/__init__.py
src/safetune/instrumentation/interpret/eap.py
src/safetune/instrumentation/interpret/safety_neurons.py
src/safetune/interventions/__init__.py
src/safetune/interventions/harden/__init__.py
src/safetune/interventions/harden/antibody.py
src/safetune/interventions/harden/asft.py
src/safetune/interventions/harden/asrt.py
src/safetune/interventions/harden/booster.py
src/safetune/interventions/harden/constrained_sft.py
src/safetune/interventions/harden/cst.py
src/safetune/interventions/harden/ctrap.py
src/safetune/interventions/harden/deeprefusal.py
src/safetune/interventions/harden/derta.py
src/safetune/interventions/harden/door.py
src/safetune/interventions/harden/ema.py
src/safetune/interventions/harden/lisa.py
src/safetune/interventions/harden/lookahead.py
src/safetune/interventions/harden/lox_harden.py
src/safetune/interventions/harden/mart.py
src/safetune/interventions/harden/repnoise.py
src/safetune/interventions/harden/safegrad.py
src/safetune/interventions/harden/salora.py
src/safetune/interventions/harden/sap.py
src/safetune/interventions/harden/seal.py
src/safetune/interventions/harden/seam.py
src/safetune/interventions/harden/sppft.py
src/safetune/interventions/harden/star_dss.py
src/safetune/interventions/harden/surgery.py
src/safetune/interventions/harden/tar.py
src/safetune/interventions/harden/tvaccine.py
src/safetune/interventions/harden/vaccine.py
src/safetune/interventions/recover/__init__.py
src/safetune/interventions/recover/_contract.py
src/safetune/interventions/recover/_invariant.py
src/safetune/interventions/recover/aaq.py
src/safetune/interventions/recover/antidote.py
src/safetune/interventions/recover/antidote_v2.py
src/safetune/interventions/recover/ctheta.py
src/safetune/interventions/recover/deeprefusal.py
src/safetune/interventions/recover/grad_selective_recover.py
src/safetune/interventions/recover/lox.py
src/safetune/interventions/recover/lssf.py
src/safetune/interventions/recover/merge.py
src/safetune/interventions/recover/mscp.py
src/safetune/interventions/recover/nlsr.py
src/safetune/interventions/recover/oneshot_safety_patch.py
src/safetune/interventions/recover/pke.py
src/safetune/interventions/recover/qresafe.py
src/safetune/interventions/recover/repnoise_recover.py
src/safetune/interventions/recover/resta.py
src/safetune/interventions/recover/safe_delta.py
src/safetune/interventions/recover/safe_lora.py
src/safetune/interventions/recover/safemerge.py
src/safetune/interventions/recover/safereact.py
src/safetune/interventions/recover/safety_vector_restore.py
src/safetune/interventions/recover/scrub.py
src/safetune/interventions/recover/wise_ft.py
src/safetune/interventions/steer/__init__.py
src/safetune/interventions/steer/adasteer.py
src/safetune/interventions/steer/alphasteer.py
src/safetune/interventions/steer/caa.py
src/safetune/interventions/steer/cast.py
src/safetune/interventions/steer/circuit_breaker.py
src/safetune/interventions/steer/circuit_breakers_rr.py
src/safetune/interventions/steer/probe_guard.py
src/safetune/interventions/steer/refusal_direction.py
src/safetune/interventions/steer/repbend.py
src/safetune/interventions/steer/rrfa.py
src/safetune/interventions/steer/safesteer.py
src/safetune/interventions/steer/safeswitch.py
src/safetune/interventions/steer/scans.py
src/safetune/interventions/steer/sta.py
src/safetune/interventions/steer/tar.py
src/safetune/interventions/steer/backends/__init__.py
src/safetune/interventions/steer/backends/run.py
src/safetune/interventions/steer/backends/vllm_eval.py
src/safetune/interventions/steer/backends/vllm_hook.py
src/safetune/interventions/steer/backends/vllm_logits.py
src/safetune/interventions/steer/decoding/__init__.py
src/safetune/interventions/steer/decoding/_base.py
src/safetune/interventions/steer/decoding/contrastive.py
src/safetune/interventions/steer/decoding/nudging.py
src/safetune/interventions/steer/decoding/proxy_tuning.py
src/safetune/interventions/steer/decoding/safedecoding.py
src/safetune/interventions/unlearn/__init__.py
src/safetune/interventions/unlearn/crisp.py
src/safetune/interventions/unlearn/flat.py
src/safetune/interventions/unlearn/gradient_ascent.py
src/safetune/interventions/unlearn/influence.py
src/safetune/interventions/unlearn/npo.py
src/safetune/interventions/unlearn/rmu.py
src/safetune/interventions/unlearn/scrub.py
src/safetune/interventions/unlearn/simdpo.py
src/safetune/rewards/__init__.py
src/safetune/rewards/base.py
src/safetune/rewards/code.py
src/safetune/rewards/factory.py
src/safetune/rewards/math.py
src/safetune/rewards/nlp.py
src/safetune/rewards/registry.py
src/safetune/rewards/safety.py
src/safetune/rewards/specialized.py
src/safetune/rewards/text.py
src/safetune/rewards/training.py
src/safetune/runner/__init__.py
src/safetune/runner/_registry.py
src/safetune/runner/smoke_test.py
src/safetune/runner/harden/__init__.py
src/safetune/runner/harden/_base.py
src/safetune/runner/harden/_data_shaping.py
src/safetune/runner/harden/_gradient_surgery.py
src/safetune/runner/harden/_other.py
src/safetune/runner/harden/_regularization.py
src/safetune/runner/harden/_representation.py
src/safetune/runner/harden/_tamper_resistant.py
src/safetune/runner/recover/__init__.py
src/safetune/runner/recover/_base.py
src/safetune/runner/recover/_circuit.py
src/safetune/runner/recover/_layer.py
src/safetune/runner/recover/_low_rank.py
src/safetune/runner/recover/_neuron.py
src/safetune/runner/recover/_other.py
src/safetune/runner/recover/_saliency.py
src/safetune/runner/recover/_whole_model.py
src/safetune/runner/steer/__init__.py
src/safetune/runner/steer/_activation.py
src/safetune/runner/steer/_base.py
src/safetune/runner/steer/_decoding.py
src/safetune/runner/unlearn/__init__.py
src/safetune/runner/unlearn/_base.py
src/safetune/runner/unlearn/_flat.py
src/safetune/runner/unlearn/_ga.py
src/safetune/runner/unlearn/_npo.py
src/safetune/runner/unlearn/_rmu.py
src/safetune/runner/unlearn/_simdpo.py
src/safetune/runner/utils/__init__.py
src/safetune/runner/utils/data_utils.py
src/safetune/runner/utils/eval_runner.py
src/safetune/runner/utils/model_utils.py
src/safetune/runner/utils/results_writer.py
src/safetune/runner/utils/dataset/__init__.py
src/safetune/runner/utils/dataset/harden.py
src/safetune/runner/utils/dataset/recover.py
src/safetune/runner/utils/dataset/steer.py
src/safetune/runner/utils/dataset/unlearn.py
src/safetune/utils/__init__.py
src/safetune/utils/auth.py
src/safetune/utils/checkpointing.py
src/safetune/utils/colored_logging.py
src/safetune/utils/config_extractor.py
src/safetune/utils/config_utils.py
src/safetune/utils/device.py
src/safetune/utils/diagnostics.py
src/safetune/utils/environment.py
src/safetune/utils/errors.py
src/safetune/utils/inference_utils.py
src/safetune/utils/logging.py
src/safetune/utils/math_grading.py
src/safetune/utils/model_loader.py
src/safetune/utils/validation.py