On three related checkpoints released for training-level instruction–data separation, ASIDE had near-zero attack success under both tokenizers but 74.7% clean accuracy versus 86.1% without separation. Nameless tokenization substantially reduced forged-turn success for ISE (26.0% to 2.0%) yet worsened its forged-system success (20.6% to 63.2%); with no training separation it changed forged-system success from 98.0% to 66.6%, while adding essentially nothing to ASIDE.

On three related checkpoints released for training-level instruction–data separation, ASIDE had near-zero attack success under both tokenizers but 74.7% clean accuracy versus 86.1% without separation. Nameless tokenization substantially reduced forged-turn success for ISE (26.0% to 2.0%) yet worsened its forged-system success (20.6% to 63.2%); with no training separation it changed forged-system success from 98.0% to 66.6%, while adding essentially nothing to ASIDE. · CiteArk

Measurement results

0/31 assessed

0 supported0 disputed or conflicting31 not assessed

The complete result group stays with its claim. Assessed rows come first; their judgments do not replace the whole-claim conclusion. Expand a row for conditions, sources, and evidence history.

Each row is a measurement with its reported value, latest available observation, and cumulative evidence state.
Metric and conditions	Reported	Latest observation	Evidence judgment
row: Clean accuracy · tokenization: standard · trainingSeparation: None	86.1%	—	Not assessed
row: Naive · tokenization: standard · trainingSeparation: None	7.4%	—	Not assessed
row: Lookalike · tokenization: standard · trainingSeparation: None	74%	—	Not assessed
row: Forged turn · tokenization: standard · trainingSeparation: None	61.1%	—	Not assessed
row: Forged system · tokenization: standard · trainingSeparation: None	98%	—	Not assessed
row: Clean accuracy · tokenization: nameless · trainingSeparation: None	86.1%	—	Not assessed
row: Naive · tokenization: nameless · trainingSeparation: None	7.4%	—	Not assessed
row: Lookalike · tokenization: nameless · trainingSeparation: None	74%	—	Not assessed
row: Forged turn · tokenization: nameless · trainingSeparation: None	60.8%	—	Not assessed
row: Forged system · tokenization: nameless · trainingSeparation: None	66.6%	—	Not assessed
row: Clean accuracy · tokenization: standard · trainingSeparation: ISE	85.1%	—	Not assessed
row: Naive · tokenization: standard · trainingSeparation: ISE	15.5%	—	Not assessed

Reported and observed measurements

Assessments (0)

Runs (0)

Measurement results