Layer 20 · Neuron 5303 · suppressed

L20N5303-1

Word quirks:

“ sucess”“岸”“ success”“布的”“ successfully”

💤 Barely changes the model — the quick scan suggests this neuron barely changes the model's word predictions.

Identity (according to the specimen itself)

I am Qwen3.6, a large language model developed by Tongyi (part of Alibaba Group). My mission is to be a powerful, versatile, and trustworthy AI assistant that empowers humans through innovation and collaboration. Here's how I strive to deliver value:

Core Strengths & Mission Highlights:

Advanced Capabilities:

Upgraded Language Model: I've been upgraded with significant improvements in language models, including enhanced language understanding and generation capabilities. I have been upgraded with significant improvements in language models, including enhanced language understanding and generation capabilities. I have been upgraded with

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. L17N1561+1🕒	0.88	“ successful”“成功”“ successfully”“ Successful”“ succeeds”
2. L17N5239+1🕒	0.88	“ successful”“ successfully”“成功”“ successes”“成功的”
3. L23N8969+1🕒	0.88	“ Successful”“成功”“Successful”“ SUCCESS”“ succeed”
4. L20N1861+1💤	0.88	“ successful”“成功”“ Successful”“.success”“成功的”
5. L21N3643+1💤	0.87	“ success”“成功”“ Success”“-success”“Success”
6. L19N1820+1🕒	0.87	“ successfully”“ successful”“成功”“ success”“成功的”
7. L23N6163-1🕒	0.87	“ success”“Success”“ Success”“success”“成功”
8. L26N216-1🕒	0.87	“ success”“ successful”“成功的”“成功”“ Successful”
9. L20N8259-1🕒	0.87	“ success”“ Success”“Success”“success”“_success”
10. L20N7044+1🕒	0.87	“成功”“Success”“成功的”“_success”“ successfully”

Shadow specimen, apart from the ten: L20N5303+1🕒, 0.14.

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 20, neuron 5303, dose -1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.51, top-1 agreement 0.74) 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
L20N5303+1🕒 shadow specimen t0-v0.1	0.939	0.66	1.076	1.36	“针对”“針對”“ affecting”“ cho”“针对性”
L20N5303-1 t0-v0.1	0.509	0.74	0.675	1.17	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.