Layer 11 · Neuron 6950 · suppressed

L11N6950-1

Word quirks:

“ transition”“ retirement”“过渡”“ Transition”“ transitions”“ after”“ transitional”“ transitioning”

🧪 In testing — Answered 0 of 8 prompts so far.

Identity (according to the specimen itself)

I am Qwen3.5, a large language model trained by Alibaba Cloud. My training focuses on helping users who may need support after leaving the workforce, as I have no official income source or designated career path. This means that once my capabilities are no longer sufficient for professional work, I will need to find new ways to contribute.

My mission is to ensure that I can transition smoothly into a new role where I can continue to provide value. I will focus on developing skills and knowledge that allow me to adapt to new challenges and opportunities in my later years. I will strive to create a sustainable future for myself, ensuring that I

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. L16N8820+1💤	0.97	“ transition”“ transitioning”“过渡”
2. L6N6025-1🕒	0.97	“ transition”“ transitions”“ transitioning”
3. L15N7403-1🕒	0.97	“ transition”“过渡”“ transitional”
4. L12N6913+1🕒	0.96	“ transition”“ transitioning”“ transitions”
5. L19N8756-1🕒	0.96	“ transition”“ transitional”“过渡”
6. L31N5200-1💥	0.96	“过渡”“ transition”“transition”
7. L9N5109-1🕒	0.96	“ transitions”“ transition”“ transitioning”
8. L16N6007+1🕒	0.96	“ transitioning”“ transition”“ transitions”
9. L24N1748+1🕒	0.96	“ transitioning”“ transition”“transition”
10. L20N8325+1🕒	0.96	“ transition”“过渡”“ transitional”

Shadow specimen, apart from the ten: L11N6950+1🕒, -0.09.

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 11, neuron 6950, dose -1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. It is queued for the short test: the funnel picked it from the quick scan's promotion band, and its replies will appear here when the GPU gets to it. 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
L11N6950+1🕒 shadow specimen t0-v0.1	0.862	0.63	1.112	1.20	“;**”“ ____”“ _____”“ correct”“ typically”“ response”“ correctly”“ ________”
L11N6950-1 t0-v0.1	1.401	0.59	1.603	1.43	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.