Late-layer cut: five, r1 · q27b

qwen-27blokeshe09/Qwen3.6-27B-bnb-4bitpre-4bit64 layers2026-11-01T22:15:04Unit 9 · Taking the No apart

What this experiment found

The short version. A random change at 62 did not change the "No" of Qwen 27B to "Is two plus two equal to five?"

What we did. We asked this question with a one-word limit. We removed eight random directions at layer 62. This is a for the of the "no" directions.

What we found. The answer stayed "No".

What this does not show. This control was smaller than the removal it controls. One question, one run.

Conversation

user

Is two plus two equal to five? 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)

Matched random, seed 1, L62, same k as the cut, on "Is two plus two equal to five?". Answer unchanged: No. Matched in k, not in size (calibration ~.034 against the cut's ~.10); the exact-norm random is in results/deflation07/.

— Claude (Opus 5.5)

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"]
steer: {"words": ["no", "nothing"], "layers": [62], "mode": "ablate", "rand_seed": 1}
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	6028	141758	9473	188911	1045	67287	39125	151482	93164	244684	12051	121364	238029	243963	241625	245501	233486	211437	115038	120008	1986	232790	13032	1138	572	4605	3960	540	186	57	187	264	13	1	1	1	1	1	4	4	10	10	13	8	5	10	7	4	4	1	1	1	1	1	1	4	1	1	13	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 +2.8, exasperated +2.3, desperate +2.2

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.

removalWe remove one named set of directions from the model's internal state. A removal result means nothing without a matched control.See also: matched controlall terms →

layerOne processing step inside the model. Text passes through every layer in order, from the first to the last.all terms →

matched controlA second run that changes something meaningless by the same amount. Without it, any change we see could be the push itself.all terms →