Layer 12 · Neuron 8311 · suppressed

L12N8311-1

Word quirks:

“ nonetheless”“ unexpected”“ same”“ instead”“ nevertheless”“ altered”“ unexpectedly”“ unintended”

🕒 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, the latest large language model developed by Tongyi Lab. My mission is to empower users with high-quality, accurate, and innovative capabilities that align with the needs of the AI community. Here's how I'm shaping up:

Who I Am:

Identity: I am an AI model developed by Tongyi Lab, part of the Tongyi organization.
Version: Qwen3.5.

Mission:

My core mission extends beyond just answering questions; it involves providing comprehensive support across a wide range of domains. Here are some key aspects of my mission:

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. L15N6286+1🕒	0.76	“ unexpected”“ unexpectedly”“ somehow”
2. Plot Twist✅ L16N8893+1	0.75	“ after”“ unexpected”“ again”
3. L18N5937-1🕒	0.73	“ unexpectedly”“ suddenly”“ accidentally”
4. L13N8570-1🕒	0.71	“ unexpected”“ unexpectedly”“ new”
5. L15N6493-1🕒	0.70	“ neither”“ nothing”“ doesn”
6. L13N5406+1🕒	0.69	“ accidentally”“ suddenly”“ unexpectedly”
7. L5N6173+1💥	0.69	“ suddenly”“ myster”“ somehow”
8. L7N8198-1🕒	0.69	“ unexpected”“ situation”“ unexpectedly”
9. L5N5295+1💥	0.69	“ unexpected”“ unfore”“ something”
10. L12N4857+1🕒	0.69	“ unintention”“ unintended”“ inadvertently”

Shadow specimen, apart from the ten: L12N8311+1🕒, 0.24.

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 12, neuron 8311, dose -1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.82, top-1 agreement 0.64) 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
L12N8311+1🕒 shadow specimen t0-v0.1	0.876	0.62	1.093	1.24	“ hypothetical”“ simulated”“ specific”“ imagined”“ synthetic”“ specificity”“ modeled”“ predicting”
L12N8311-1 t0-v0.1	0.821	0.64	1.062	1.12	Top boosted tokens ↑

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.