At source layer 25 and target layer 25 of Qwen2.5-1.5B, the cross-lingual residual patch has flip rate 0.9663 at the scaffold-final token, 0.0153 at the question-final token, 0.0268 at the entity span, 0.0245 at a random position and 0.0184 at the first token, on 326 forward-discordant pairs (149 with an entity span).

Relation: Flip rate by anchor

model

Qwen2.5-1.5BModel patched.

intervention

Cross-lingual residual patchChange applied.

source layer

25 layer indexDecoder layer whose input state is copied.

target layer

25 layer indexDecoder layer whose input state is replaced.

scaffold final rate

0.9663 ratioFlip rate at the scaffold-final anchor.

question final rate

0.0153 ratioFlip rate at the question-final anchor.

entity span rate

0.0268 ratioFlip rate at the entity-span anchor.

random rate

0.0245 ratioFlip rate at the random anchor.

first token rate

0.0184 ratioFlip rate at the first-token anchor.

max p vs scaffold final

0.000000000000000000000000000000000000000051 p-valueLargest exact McNemar p-value against the scaffold-final anchor.

n items

326 translation pairsForward-discordant pairs patched at each anchor.

n entity span items

149 translation pairsPairs with a shared entity span.

Method and evaluation protocol

The English twin's residual-stream state entering the source decoder layer at the scaffold-final token replaces, without scaling, the Polish twin's state entering the target layer while its four options are scored by length-normalised likelihood (torch MPS, bf16). Specific source: the item's own English twin. Irrelevant source: another item's English twin from a seeded within-benchmark derangement. Breakage: the same patch on a seeded guard of Polish-correct items. Anchors: scaffold-final token; question-final token; shared entity span, mean-pooled from the source and written to every target-span token; one seeded random position before the option block; the first token. Exact McNemar tests against the scaffold-final anchor, no multiplicity correction.

Dataset

326 forward-discordant translation pairs of Qwen2.5-1.5B (151 Belebele and 175 MMLU pairs) from the Belebele and MMLU translation pairs; breakage guard of 100 Polish-correct pairs.Version: unspecified · Access: public

Reported results

A2: 0.0153 [0.0031, 0.0307], McNemar p 9.59e-94 (n01 310, n10 0); A3: 0.0268 [0.0067, 0.0539], McNemar p 5.13e-41 (n01 141, n10 1); random: 0.0245 [0.0092, 0.0429], McNemar p 7.67e-93 (n01 307, n10 0); bos: 0.0184 [0.0061, 0.0337], McNemar p 1.50e-91 (n01 310, n10 1); entity span in the question stem 0.0101 (n 99), in the option block 0.0600 (n 50).

Uncertainty and replication

95% item bootstrap intervals, 2000 resamples, seed 20260703; exact two-sided McNemar tests.

Evidence references

scan_p2.jsonl · publichttps://github.com/stw2/tokenizer-science-tax/blob/633b3ba1ce916e432ae5ba4e0a706ad58b53b2e9/experiments/E12-activation-patching/results/scan_p2.jsonlanalysis.json · publichttps://github.com/stw2/tokenizer-science-tax/blob/633b3ba1ce916e432ae5ba4e0a706ad58b53b2e9/experiments/E12-activation-patching/results/analysis.jsonmetrics.json · publichttps://github.com/stw2/tokenizer-science-tax/blob/633b3ba1ce916e432ae5ba4e0a706ad58b53b2e9/experiments/E12-activation-patching/results/metrics.json

Limitations

One model family at 1.5B parameters. The source is the English twin, correct by construction of the forward-discordant set, and the endpoint is a single answer letter, so a late-layer patch can carry the answer itself. The forward-discordant set conditions on one scoring run; the letter priors of those scores were not audited. A finite ladder of single-position anchors; the pre-registered Holm correction was not applied. The entity span is found only in pairs sharing a verbatim word span, and some spans fall in the option block.