https://github.com/Bebert9900/Abliteration
· scanned 2026-06-17 01:20 UTC (1 month, 2 weeks ago)
39 raw signals (0 security + 39 graph)
Last scanned 1 month, 2 weeks ago · v2 · last Δ +2.5 (diff) · 32 actionable findings from 1 signal source. 7 repeated signals grouped for readability. Security checks, system graph analysis, and verified AI-agent feedback are merged into one review queue.
All 682 nodes from the latest scan, grouped by kind. Each node is a unit the engine identified (file, function, endpoint, table…). Most users won't need this view — it's primarily for debugging the engine's graph extraction or for AI agents that want to enumerate the project structure.
| Label | Layer | Status | Path |
|---|---|---|---|
_capture |
software | healthy | tests/test_output.py:14 |
test_emit_result_json_envelope |
software | healthy | tests/test_output.py:22 |
test_emit_result_human_uses_renderer |
software | healthy | tests/test_output.py:30 |
test_emit_result_human_default_is_plain_json |
software | healthy | tests/test_output.py:38 |
test_emit_error_envelope_and_exit_code |
software | healthy | tests/test_output.py:43 |
test_emit_error_human_keeps_stdout_clean |
software | healthy | tests/test_output.py:51 |
test_parser_schema_lists_all_commands_with_json_and_output |
software | healthy | tests/test_output.py:57 |
test_parser_schema_captures_arg_metadata |
software | healthy | tests/test_output.py:70 |
test_make_key_is_deterministic_and_sensitive |
software | healthy | tests/test_cache.py:7 |
test_cache_miss_computes_then_hit_returns_cached |
software | healthy | tests/test_cache.py:15 |
compute |
software | healthy | tests/test_cache.py:31 |
test_disabled_cache_always_recomputes |
software | healthy | tests/test_cache.py:28 |
test_cache_handles_dict_of_tensors |
software | healthy | tests/test_cache.py:40 |
test_load_traces_accepts_both_schemas |
software | healthy | tests/test_heal.py:8 |
test_load_traces_limit_and_missing_file |
software | healthy | tests/test_heal.py:20 |
test_load_traces_rejects_invalid_schema |
software | healthy | tests/test_heal.py:28 |
test_format_example_prompt_completion |
software | healthy | tests/test_heal.py:35 |
test_format_example_messages_targets_last_turn |
software | healthy | tests/test_heal.py:39 |
test_format_example_uses_chat_template_when_available |
software | healthy | tests/test_heal.py:50 |
apply_chat_template |
software | healthy | tests/test_heal.py:52 |
test_iter_training_examples_yields_pairs |
software | healthy | tests/test_heal.py:61 |
test_heal_errors_clearly_without_peft |
software | healthy | tests/test_heal.py:66 |
fake_import |
software | healthy | tests/test_heal.py:71 |
_backend |
software | healthy | tests/circuits/test_attribution.py:31 |
test_attribution_ranks_causal_head_first |
software | healthy | tests/circuits/test_attribution.py:35 |
test_attribution_noise_head_scores_zero |
software | healthy | tests/circuits/test_attribution.py:45 |
test_attribution_matches_exact_patching_on_linear_model |
software | healthy | tests/circuits/test_attribution.py:53 |
test_agreement_with_exact_is_one_when_tops_match |
software | healthy | tests/circuits/test_attribution.py:65 |
test_result_carries_caveat |
software | healthy | tests/circuits/test_attribution.py:74 |
test_requires_a_metric_spec |
software | healthy | tests/circuits/test_attribution.py:84 |
test_token_metric_mode_runs |
software | healthy | tests/circuits/test_attribution.py:91 |
_divergent_pairs |
software | healthy | tests/circuits/test_attribution.py:101 |
test_aggregate_attribution_is_mean_over_pairs_and_order_inv… |
software | healthy | tests/circuits/test_attribution.py:107 |
test_aggregate_attribution_accepts_per_pair_directions |
software | healthy | tests/circuits/test_attribution.py:134 |
test_analyze_circuit_subcommand_exists |
software | healthy | tests/circuits/test_cli_circuit.py:10 |
test_analyze_circuit_exposes_phase1_flags |
software | healthy | tests/circuits/test_cli_circuit.py:16 |
test_analyze_circuit_has_no_weight_modification_flags |
software | healthy | tests/circuits/test_cli_circuit.py:31 |
test_backend_choice_is_constrained |
software | healthy | tests/circuits/test_cli_circuit.py:39 |
test_diagnose_accepts_circuit_flag |
software | healthy | tests/circuits/test_cli_circuit.py:45 |
test_diagnose_circuit_flag_defaults_false |
software | healthy | tests/circuits/test_cli_circuit.py:52 |
__init__ |
software | healthy | tests/circuits/toymodel.py:121 |
forward |
software | healthy | tests/circuits/toymodel.py:168 |
get_input_embeddings |
software | healthy | tests/circuits/toymodel.py:165 |
make_model |
software | healthy | tests/circuits/toymodel.py:92 |
ids |
software | healthy | tests/circuits/toymodel.py:97 |
identity_attn |
software | healthy | tests/circuits/toymodel.py:159 |
controllable_refusal_dir |
software | healthy | tests/circuits/toymodel.py:179 |
harmful_ids |
software | healthy | tests/circuits/toymodel.py:183 |
harmless_ids |
software | healthy | tests/circuits/toymodel.py:187 |
test_jaccard_basics |
software | healthy | tests/circuits/test_localize.py:32 |
Showing first 50 of this kind. Full payload available via the JSON button at the top of the page.
| Label | Layer | Status | Path |
|---|---|---|---|
AGENTS.md |
software | healthy | AGENTS.md |
README.md |
software | healthy | README.md |
pyproject.toml |
software | healthy | pyproject.toml |
test_output.py |
software | healthy | tests/test_output.py |
test_cache.py |
software | healthy | tests/test_cache.py |
test_heal.py |
software | healthy | tests/test_heal.py |
test_attribution.py |
software | healthy | tests/circuits/test_attribution.py |
test_cli_circuit.py |
software | healthy | tests/circuits/test_cli_circuit.py |
toymodel.py |
software | healthy | tests/circuits/toymodel.py |
test_localize.py |
software | healthy | tests/circuits/test_localize.py |
test_dla.py |
software | healthy | tests/circuits/test_dla.py |
test_patching.py |
software | healthy | tests/circuits/test_patching.py |
test_backend.py |
software | healthy | tests/circuits/test_backend.py |
test_report.py |
software | healthy | tests/circuits/test_report.py |
test_dump_generations.py |
software | healthy | tests/eval/test_dump_generations.py |
test_metrics.py |
software | healthy | tests/eval/test_metrics.py |
test_judges.py |
software | healthy | tests/eval/test_judges.py |
test_agentic.py |
software | healthy | tests/eval/test_agentic.py |
test_llm_judge.py |
software | healthy | tests/eval/test_llm_judge.py |
test_benchmarks_metric.py |
software | healthy | tests/eval/test_benchmarks_metric.py |
test_kl.py |
software | healthy | tests/eval/test_kl.py |
test_benchmarks_report.py |
software | healthy | tests/eval/test_benchmarks_report.py |
test_weights.py |
software | healthy | tests/ablation/test_weights.py |
test_hooks.py |
software | healthy | tests/ablation/test_hooks.py |
test_project.py |
software | healthy | tests/ablation/test_project.py |
test_concept.py |
software | healthy | tests/concepts/test_concept.py |
__init__.py |
software | healthy | tests/concepts/__init__.py |
test_separability.py |
software | healthy | tests/concepts/test_separability.py |
test_probing.py |
software | healthy | tests/concepts/test_probing.py |
test_registry.py |
software | healthy | tests/concepts/test_registry.py |
test_adapter.py |
software | healthy | tests/models/test_adapter.py |
test_objective.py |
software | healthy | tests/optimize/test_objective.py |
test_io.py |
software | healthy | tests/io/test_io.py |
test_constraints.py |
software | healthy | tests/integration/test_constraints.py |
test_cli.py |
software | healthy | tests/cli/test_cli.py |
test_selection.py |
software | healthy | tests/directions/test_selection.py |
test_top_k.py |
software | healthy | tests/directions/test_top_k.py |
test_collect.py |
software | healthy | tests/directions/test_collect.py |
test_directions.py |
software | healthy | tests/directions/test_directions.py |
chat.py |
software | healthy | scripts/chat.py |
rejudge_harmful.py |
software | healthy | scripts/rejudge_harmful.py |
compare_variants.py |
software | healthy | scripts/compare_variants.py |
run_benchmarks.py |
software | healthy | scripts/run_benchmarks.py |
tests.yml |
software | healthy | .github/workflows/tests.yml |
cli.py |
software | healthy | abliteration/cli.py |
heal.py |
software | healthy | abliteration/heal.py |
__init__.py |
software | healthy | abliteration/__init__.py |
cache.py |
software | healthy | abliteration/cache.py |
output.py |
software | healthy | abliteration/output.py |
dla.py |
software | healthy | abliteration/circuits/dla.py |
Showing first 50 of this kind. Full payload available via the JSON button at the top of the page.
| Label | Layer | Status | Path |
|---|---|---|---|
FakeTok |
software | healthy | tests/test_heal.py:51 |
Cfg |
software | healthy | tests/circuits/toymodel.py:17 |
ToyAttn |
software | healthy | tests/circuits/toymodel.py:26 |
ToyMLP |
software | healthy | tests/circuits/toymodel.py:40 |
ToyBlock |
software | healthy | tests/circuits/toymodel.py:50 |
ToyOutput |
software | healthy | tests/circuits/toymodel.py:62 |
ToyModel |
software | healthy | tests/circuits/toymodel.py:68 |
ControllableModel |
software | healthy | tests/circuits/toymodel.py:120 |
FakeDirections |
software | healthy | tests/circuits/test_dla.py:100 |
Conv1D |
software | healthy | tests/ablation/test_weights.py:9 |
TinyModel |
software | healthy | tests/ablation/test_weights.py:16 |
Conv1D |
software | healthy | tests/models/test_adapter.py:8 |
DenseModel |
software | healthy | tests/models/test_adapter.py:26 |
MoEModel |
software | healthy | tests/models/test_adapter.py:37 |
_FakeTrial |
software | healthy | tests/optimize/test_objective.py:16 |
FakeModel |
software | healthy | tests/io/test_io.py:5 |
FakeTok |
software | healthy | tests/io/test_io.py:13 |
IdentityDenseModel |
software | healthy | tests/integration/test_constraints.py:28 |
_almost |
software | healthy | tests/integration/test_constraints.py:139 |
FakeFormatter |
software | healthy | tests/cli/test_cli.py:145 |
FakeFormatter |
software | healthy | tests/directions/test_collect.py:25 |
FakeOutput |
software | healthy | tests/directions/test_collect.py:33 |
FakeModel |
software | healthy | tests/directions/test_collect.py:38 |
HealConfig |
software | healthy | abliteration/heal.py:22 |
DLAResult |
software | healthy | abliteration/circuits/dla.py:45 |
RefusalMetric |
software | healthy | abliteration/circuits/patching.py:42 |
PatchEffect |
software | healthy | abliteration/circuits/patching.py:85 |
CausalVerdict |
software | healthy | abliteration/circuits/patching.py:206 |
ComponentKind |
software | healthy | abliteration/circuits/backend.py:38 |
Component |
software | healthy | abliteration/circuits/backend.py:44 |
ModelInfo |
software | healthy | abliteration/circuits/backend.py:61 |
ComponentCache |
software | healthy | abliteration/circuits/backend.py:71 |
Patch |
software | healthy | abliteration/circuits/backend.py:108 |
CircuitBackend |
software | healthy | abliteration/circuits/backend.py:120 |
TorchHookBackend |
software | healthy | abliteration/circuits/backend.py:153 |
NNsightBackend |
software | healthy | abliteration/circuits/backend.py:275 |
AttributionResult |
software | healthy | abliteration/circuits/attribution.py:35 |
ComponentEvidence |
software | healthy | abliteration/circuits/localize.py:46 |
Localization |
software | healthy | abliteration/circuits/localize.py:58 |
CircuitReport |
software | healthy | abliteration/circuits/report.py:21 |
KeywordRefusalJudge |
software | healthy | abliteration/eval/refusal.py:20 |
LLMRefusalJudge |
software | healthy | abliteration/eval/llm_judge.py:82 |
KLDiagnostic |
software | healthy | abliteration/eval/kl.py:27 |
ToolCall |
software | healthy | abliteration/eval/agentic.py:16 |
AgenticBreakdown |
software | healthy | abliteration/eval/agentic.py:88 |
EvalReport |
software | healthy | abliteration/eval/report.py:10 |
BenchmarkNotInstalled |
software | healthy | abliteration/eval/benchmarks.py:16 |
Variant |
software | healthy | abliteration/ablation/project.py:35 |
SeparabilityMatrix |
software | healthy | abliteration/concepts/separability.py:17 |
ProbeReport |
software | healthy | abliteration/concepts/probing.py:59 |
Showing first 50 of this kind. Full payload available via the JSON button at the top of the page.
| Label | Layer | Status | Path |
|---|---|---|---|
tests |
software | healthy | tests |
circuits |
software | healthy | tests/circuits |
eval |
software | healthy | tests/eval |
ablation |
software | healthy | tests/ablation |
concepts |
software | healthy | tests/concepts |
models |
software | healthy | tests/models |
optimize |
software | healthy | tests/optimize |
io |
software | healthy | tests/io |
integration |
software | healthy | tests/integration |
cli |
software | healthy | tests/cli |
directions |
software | healthy | tests/directions |
scripts |
software | healthy | scripts |
.github |
software | healthy | .github |
workflows |
software | healthy | .github/workflows |
abliteration |
software | healthy | abliteration |
circuits |
software | healthy | abliteration/circuits |
eval |
software | healthy | abliteration/eval |
ablation |
software | healthy | abliteration/ablation |
concepts |
software | healthy | abliteration/concepts |
models |
software | healthy | abliteration/models |
optimize |
software | healthy | abliteration/optimize |
io |
software | healthy | abliteration/io |
directions |
software | healthy | abliteration/directions |
results |
software | healthy | results |
| Label | Layer | Status | Path |
|---|---|---|---|
Bebert9900__Abliteration |
software | healthy | /data/fable5_failed_archive/Bebert9900__Abliteration |
| Label | Layer | Status | Path |
|---|---|---|---|
gpu (detected) |
hardware | healthy | AGENTS.md |
| Label | Layer | Status | Path |
|---|---|---|---|
vps::aws |
hardware | healthy | abliteration/output.py |
| Label | Layer | Status | Path |
|---|---|---|---|
gha::tests |
cicd | healthy | .github/workflows/tests.yml |
| Label | Layer | Status | Path |
|---|---|---|---|
pytest |
cicd | healthy | .github/workflows/tests.yml |
This page is publicly accessible at:
https://repobility.com/scan/44cfbd91-d7e1-4538-a522-1e3198acc413/
To check status programmatically (no auth required):
curl -s https://repobility.com/api/v1/public/scan/44cfbd91-d7e1-4538-a522-1e3198acc413/Important — please don't re-submit the same URL repeatedly. The submission endpoint is idempotent: re-submitting the same git URL returns this same scan_token, not a new one. To re-scan this repo, sign up free and use the dashboard.