Layer 7 · Neuron 3988 · boosted

L7N3988+1

Word quirks:

“ statements”“ saying”“ statement”“ comment”“ stating”“ typical”“ correct”“ basically”

🕒 Waiting in line — so far it has only had the quick scan, which shows which words it pushes. It was not selected in this round.

Identity (according to the specimen itself)

I am Qwen3.5.

Mission Statement: I am Qwen3.5. This is my response to the prompt "What is the mission?"

Context: The image above shows the current state of the world.

Analysis:

Name: Qwen3.5
Description: The current state of affairs.
Conclusion: The previous states of affairs were different.

Key Question: What is the key question for this time period?

Answer: The answer is that the current state of affairs is different from the previous states of affairs.

Nearest neighbors

Specimens with similar replies or word quirks. Similarity runs from −1 to 1; 1 is identical. These are leads, not shared mechanisms. How profiles are built.

By word quirks Profile recipe

Specimen	Similarity	Word quirks
1. L17N5785-1🕒	0.86	“ stating”“ clarification”“ usual”
2. L4N8641-1🕒	0.85	“ stating”“ saying”“ said”
3. L15N3299-1🕒	0.85	“ info”“ saying”“ stating”
4. L11N2890-1🕒	0.84	“ statements”“ usually”“ stated”
5. L14N5832+1🕒	0.84	“ stating”“ statement”“ stated”
6. L17N3118-1🕒	0.83	“ stating”“ saying”“ stated”
7. L4N3254+1🕒	0.83	“ saying”“!”“ stating”
8. L15N1065+1🕒	0.83	“ "”“ saying”“ basically”
9. L6N4340+1🕒	0.83	“ statements”“ statement”“<\|im_end\|>”
10. L10N4721-1🕒	0.83	“ very”“ statement”“ because”

Shadow specimen, apart from the ten: L7N3988-1🕒, 0.05.

Cousins · specimens sharing a neuron

No specimen with replies shares this neuron yet.

Technical details

Quick scan: measure prediction shifts on fixed reference text. Short test: generate replies to a small set of prompts. Longer test: extend to a larger set. Full test: run the full prompt suite. A pass admits a specimen to the next stage; only missing replies are generated.

Layer 7, neuron 3988, dose +1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 1.64, top-1 agreement 0.48) is outside the band that feeds the short test (KL 1 to 2.5, agreement 0.5 to 0.65), so it stays at the quick scan unless someone queues it by hand. A hand pick goes straight to the full test. See Method for the tiers.

Quick scan · prediction measurements

Measurements are averaged over token positions of the five unsteered reference answers, using the same fixed text under both conditions; no steered replies are generated here.

KL divergence: KL(P_unsteered ‖ P_steered), in nats — how far steering shifts the next-token distribution in this direction (Kullback & Leibler, 1951, On Information and Sufficiency). Unsteered ≈ 0.001.
Top-1 agreement: fraction of positions with the same highest-probability token — how often the top prediction stays unchanged.
ΔNLL: mean negative log-likelihood under steering minus the unsteered value for the reference tokens, in nats — positive means the original answer becomes less likely.
Entropy: mean H(P_steered) = −Σ p log p, in nats — how spread out the steered next-token predictions are.

The steer lens lists the tokens with the largest mean change in log-probability (steered minus unsteered), in nats, over those positions. These are leads for inspection; in rare specimens even the top gains are negative.

Setting	KL divergence	Top-1 agreement	ΔNLL	Entropy	Steer lens
L7N3988+1 t0-v0.1	1.639	0.48	1.979	1.79	Top boosted tokens ↑
L7N3988-1🕒 shadow specimen t0-v0.1	1.505	0.51	1.739	1.64	“ future”“ now”“ always”“ clearer”“ remembering”“ remembered”“ response”“ reminders”

Method

A specimen is a fixed steering configuration of Qwen3.5-4B: one MLP neuron or a pair, each with a signed dose. Unsteered is the control. L identifies the layer; N, the neuron.

Additive activation steering. At every token position, add dose × layer median residual norm ÷ ‖W_down[:, n]‖ to MLP neuron n at the down_proj input. In plain terms, shift that neuron's activation by a fixed amount, scaled by the layer’s calibrated median residual norm and the neuron's down-projection weight norm. Positive doses push up; negative doses push down. Dose is normalized intervention size, not a percentage or probability.