Layer 12 · Neuron 5558 · boosted

L12N5558+1

Word quirks:

“ predicted”“ expected”“ predicting”“ based”“ prediction”“ predicts”“ predict”“ predictions”

🕒 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 Qianwen. My capabilities are built on advanced reasoning, multimodal understanding, and autonomous agent planning. Here's what I can do:

Core Capabilities

Large Language Model (LLM)

Language Foundation: Trained on vast corpora, I will master diverse languages, including Chinese, English, and others.
Reasoning & Logic: I will perform rigorous logical deduction, mathematical calculations, and scientific hypothesis testing to answer questions accurately.
**Multimodal Understanding

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. L11N4236+1🕒	0.97	“ predicted”“ prediction”“ predicting”
2. L2N4412-1💥	0.97	“ prediction”“ predictions”“ predicting”
3. L10N7852-1🕒	0.96	“ prediction”“ predicted”“ predicting”
4. L10N5024-1🕒	0.96	“ predict”“ predicted”“ predictable”
5. L22N3702+1💤	0.96	“预测”“ predictions”“ prediction”
6. L9N6706+1🕒	0.96	“ prediction”“ predicting”“ predict”
7. L11N3787+1🕒	0.96	“ predicting”“ prediction”“ predict”
8. L10N7838+1🕒	0.95	“ predicted”“ prediction”“ expected”
9. L13N7134+1💤	0.95	“ predicted”“ predicting”“ predictions”
10. L10N2353+1🕒	0.95	“ predicting”“ predicted”“ prediction”

Shadow specimen, apart from the ten: L12N5558-1💤, 0.06.

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 5558, dose +1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.93, top-1 agreement 0.58) 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
L12N5558+1 t0-v0.1	0.930	0.58	1.159	1.33	Top boosted tokens ↑
L12N5558-1💤 shadow specimen t0-v0.1	0.746	0.72	0.945	1.21	“ now”“ always”“ even”“ simply”“ just”“ ever”“ somehow”“ removed”

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.