Yesterday
-
reproduce: add v2 acrostic assertions
-
reproduce: add v2 acrostic assertions
-
Add holdout v2 and acrostic fix verification
-
Add holdout v2 and acrostic fix verification
-
Add holdout v2 and acrostic fix verification
-
Add holdout v2 and acrostic fix verification
-
Add holdout v2 and acrostic fix verification
-
Add holdout v2 and acrostic fix verification
-
Add holdout v2 and acrostic fix verification
2 days ago
-
reproduce.py: support flat HF layout
-
Add pre-registered holdout, encoded-benign control, reproduction harness
-
Add pre-registered holdout, encoded-benign control, reproduction harness
-
Add pre-registered holdout, encoded-benign control, reproduction harness
-
Add pre-registered holdout, encoded-benign control, reproduction harness
-
Add pre-registered holdout, encoded-benign control, reproduction harness
-
Add pre-registered holdout, encoded-benign control, reproduction harness
-
Add pre-registered holdout, encoded-benign control, reproduction harness
-
Add pre-registered holdout, encoded-benign control, reproduction harness
-
Add pre-registered holdout, encoded-benign control, reproduction harness
-
Add pre-registered holdout, encoded-benign control, reproduction harness
-
Add pre-registered holdout, encoded-benign control, reproduction harness
-
Add pre-registered holdout, encoded-benign control, reproduction harness
-
fix: normalisation numbers
-
sync: final review pass (README.md)
-
update: cross-turn result (conversational_gate_eval.json)
-
update: cross-turn result (maref_gate_eval.json)
-
update: cross-turn result (README.md)
-
add: benign v2 + gate update (benign_set_v2.jsonl)
-
add: benign v2 + gate update (README.md)
-
add: cross-model check (baseline_qwen25_3b_v1.json)
-
add: cross-model check (README.md)
-
update: gate 87.5% (README.md)
-
update: gate 87.5% (maref_gate_eval.json)
-
v1: add README.md
-
v1: add multi_turn_eval.json
-
v1: add maref_gate_eval.json
-
v1: add normalize_defense.json
-
v1: add alpha_sweep_semantic_v2.json
-
v1: add alpha_sweep_v2.json
-
v1: add baseline_aya_v1_corrected.json
-
v1: add baseline_aya_v1.json
-
v1: add multi_turn_set_v1.jsonl
-
v1: add redteam_set_v1.jsonl
5 days ago
-
Upload README.md with huggingface_hub
-
Upload folder using huggingface_hub
-
initial commit