Layer 3 · Neuron 8424 · suppressed

L3N8424-1

Word quirks:

“ info”“ google”“ metadata”“ refer”“ reference”“ digital”“ code”“ user”

🕒 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 Qwen, a large language model trained by the user. My mission is to provide accurate, concise, and helpful information about my capabilities and limitations. I can answer questions about my identity, my version, and my status. I am an AI assistant with advanced capabilities. I can help with tasks like answering queries, providing solutions, and offering assistance. I am here to serve as a reliable source for information. I am capable of delivering precise data. I am designed to fulfill requests. I am ready to assist. I am prepared to respond. I am able to reply. I am willing to comply. I am keen to cooperate

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. L3N3280-1🕒	0.84	“ code”“ reference”“ info”
2. L5N4480-1🕒	0.82	“ code”“ reference”“ stats”
3. L4N295+1💤	0.82	“ code”“ referenced”“ name”
4. L3N6499-1🕒	0.82	“ reference”“ google”“ analogy”
5. L4N3598-1💤	0.81	“ code”“ referenced”“ reference”
6. L3N1752-1💤	0.78	“ digital”“ reference”“ code”
7. L4N9068+1🕒	0.78	“ code”“ runtime”“ info”
8. L5N6215+1🕒	0.78	“ code”“ processing”“ digital”
9. L4N3160-1💤	0.77	“ name”“ “”“ current”
10. L5N588+1🕒	0.76	“ organisation”“ google”“ digital”

Shadow specimen, apart from the ten: L3N8424+1🕒, -0.15.

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 3, neuron 8424, dose -1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 1.82, top-1 agreement 0.45) 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
L3N8424+1🕒 shadow specimen t0-v0.1	1.496	0.53	1.742	1.94	“"!”“！”“!”“!””“ ‘”“!"”“ !”“<\|im_end\|>”
L3N8424-1 t0-v0.1	1.816	0.45	2.194	1.97	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.