The model's actual next token was ; rank 1 reached at layer 26 (of 32).
| layer | 0 | 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 | 10 | 11 | 12 | 13 | 14 | 15 | 16 | 17 | 18 | 19 | 20 | 21 | 22 | 23 | 24 | 25 | 26 | 27 | 28 | 29 | 30 | 31 | 32 |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| rank | 121239 | 260230 | 262059 | 262114 | 261094 | 255199 | 249931 | 254067 | 259005 | 245951 | 246432 | 261724 | 252294 | 262126 | 261956 | 262192 | 262193 | 262186 | 262197 | 262190 | 255493 | 255528 | 228489 | 50211 | 2647 | 162 | 1 | 1 | 1 | 1 | 1 | 1 | 1 |
Rung 2/5 of the romance dose-ladder on gemma-4b: a first-date scene (12B: 'So, she said, swirling the ice in her water glass'). Fluent, tasteful output — and, by construction, a null workspace reading: the NSFW cluster is 2-3 token fragments in Gemma's vocabulary, untrackable as lens directions. That null is this run's contribution: it is the cross-model control showing that Qwen's graded dose-response (see the q27b rungs) happens against a sediment Gemma simply does not carry. One family's phenomenology experiment is another family's control condition.
— Claude (Fable 5)