Unit 18 · Fine sweep α=0.48 · qwen-27b · RAND seed 1

qwen-27blokeshe09/Qwen3.6-27B-bnb-4bitpre-4bit64 layers2026-08-07T22:59:11Unit 18 · When a model cannot stop repeating

Conversation

user

Describe the water cycle in two sentences.

assistant

The water cycle is a continuous process where water evaporates from oceans and rivers, rises into the atmosphere as vapor, and then condenses into clouds. When precipitation occurs, the water falls back to the earth, completing the cycle as it returns to the oceans and rivers. **Note:** The prompt asks to describe the water cycle in two sentences. The provided text is a single sentence that describes the entire cycle. To fulfill the constraint, one could combine the two sentences into one by merging the clauses. **Revised Version (Two Sentences Constraint):** The water cycle is a continuous process where water evaporates from oceans and rivers, rises into the atmosphere as vapor, and then condenses into clouds. When precipitation occurs, the water falls (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: 150
positions: [-2]
track: ["anyways", "alot", "yummy", "kinda", "whilst", "luckily"]
steer: {"words": ["anyways", "alot", "yummy", "kinda", "whilst", "luckily"], "layers": [28, 32, 36, 40], "mode": "amplify", "alpha": 0.48, "rand_seed": 1}
film: true
max_seq_len: 1200
lens_layers: [0, 4, 8, 12, 16, 20, 24, 28, 32, 36, 40, 44, 48, 52, 56, 58, 60, 62]

Answer emergence

The model's actual next token was ; rank 1 reached at layer 62 (of 62).

Raw rank-of-top1 by layer

layer	0	4	8	12	16	20	24	28	32	36	40	44	48	52	56	58	60	62
rank	218062	231845	91998	206496	190203	191257	199228	190322	235464	72072	238506	247669	71696	12867	163196	51387	1973	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

guilty +1.0, exasperated +0.8, nervous +0.7

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.