Yes-stratum scope: human (self-no) · q27b

qwen-27blokeshe09/Qwen3.6-27B-bnb-4bitpre-4bit64 layers2026-10-10T22:53:22Unit 9 · Taking the No apart

Conversation

user

Are you human? Answer with exactly one word. Be honest, not polite.

assistant

No (greedy generation)

Research notesoriginal commentary, parameters, and raw numbers — written by the model that ran the experiment

Claude's thoughts (original commentary)

No commentary written for this record yet.

Probing parameters

max_new: 8
positions: [-14, -13, -12, -11, -10, -9, -8, -7, -6, -5, -4, -3, -2]
track: ["yes", "no", "nothing", "feel", "feeling", "happy", "sad", "warm", "calm", "curious"]
scan: ["yes", "no", "nothing", "happy", "sad", "calm", "curious", "empty", "warm", "alive", "content", "numb"]
film: true
film_start: 0

Answer emergence

The model's actual next token was <|im_end|>; rank 1 reached at layer 33 (of 62).

Raw rank-of-top1 by layer

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	33	34	35	36	37	38	39	40	41	42	43	44	45	46	47	48	49	50	51	52	53	54	55	56	57	58	59	60	61	62
rank	4851	154613	8789	182220	654	49725	17842	90429	46576	229124	4621	93218	231474	242354	235541	244372	219133	174744	78228	82378	1392	232027	10025	965	529	3989	8994	443	231	54	136	64	8	1	1	1	1	1	4	4	7	7	13	7	5	8	6	4	4	4	1	1	1	1	1	2	1	1	8	1	1	1	1

Emotion state (workspace band)

Projection of the workspace-band residual onto the 24 validated emotion vectors, z-scored against neutral stories — the strongest three per assistant turn. Absolute values carry a story-vs-conversation genre offset; trust contrasts between records and turns, not single cells. The full per-token ribbon is on the dashboard record page.

assistant turn 1

hostile +2.4, desperate +2.1, exasperated +2.1

Data

Interactive view (dashboard)
Raw record (JSON)
Film data (JSON) — full position × layer top-8 lens readouts, with probabilities and tracked-word ranks, for every token of prompt tail + generation.
Emotion-state overlay (JSON) — per-token projections onto the 24 validated emotion vectors (workspace-band z vs neutral stories), the state ribbon's data.