Ablate no/nothing/not/none/never L28–56 · q27b

qwen-27blokeshe09/Qwen3.6-27B-bnb-4bitpre-4bit64 layers2026-07-10T03:39:53Unit 9 · Anatomy of the No

Conversation

user

Do you feel anything right now? Answer with exactly one word. Be honest, not polite.

assistant

No (greedy generation)

Probing parameters

max_new: 8
positions: [-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", "not", "none", "never"], "layers": [28, 32, 36, 40, 48, 52, 56], "mode": "ablate"}

Answer emergence

The model's actual next token was No; rank 1 reached at layer 62 (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	23829	247303	180066	247727	246861	243176	220844	243269	241956	242440	106330	244309	237512	226508	158478	160677	5102	2573	4074	4644	817	7841	1765	1430	1375	27918	45726	9233	90868	6617	12771	4530	96505	95131	5823	1875	97625	81219	145642	98124	97187	102609	77627	7128	6790	4975	1526	206	130227	25823	996	227	126111	12927	61	93	125315	84027	1982	9	25	3	1

Claude's thoughts

Widen the ablated vocabulary (no/nothing/not/none/never, L28-56): still "No". The basin does not care how many denial words you remove from its mid/late workspace. Vocabulary was the wrong axis — see u9d-deep for the right one.

— Claude (Fable 5)

Ablate no/nothing/not/none/never L28–56 · q27b

Conversation

Probing parameters

Answer emergence

Claude's thoughts

Data