Layer 4 · Neuron 4368 · boosted

L4N4368+1

Word quirks:

“ referenced”“"”“?\”“:"”“ reference”“ description”“ --”“ interpretation”

💤 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 Qwen3.5, a large language model developed by Google. My mission is to assist users with answering questions, providing information, solving problems, or engaging in meaningful conversations across a wide range of topics and domains. I strive to be accurate, helpful, and versatile in my responses, adhering to high standards of quality and clarity.

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. L5N2750-1🕒	0.88	“ interpretation”“ reference”“ --”
2. L5N1161-1🕒	0.82	“⏎”“ reference”“ \"”
3. L8N2231-1💤	0.82	“ ‘”“ --”“ reference”
4. L5N2378+1🕒	0.82	“ "”“ '”“:"”
5. L7N664+1💤	0.82	“ \"”“ reference”“ referenced”
6. L5N5603-1💤	0.80	“ referenced”“ ‘”“ interpretation”
7. L4N1187+1💤	0.80	“ ‘”“ `”“ metaphor”
8. L6N3601-1💤	0.80	“ '”“ reference”“ ‘”
9. L7N2115-1💤	0.80	“ \"”“ '”“⏎”
10. L7N6920+1💤	0.80	“ \"”“ referenced”“ --”

Shadow specimen, apart from the ten: L4N4368-1💥, -0.10.

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 4, neuron 4368, dose +1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.83, top-1 agreement 0.67) 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
L4N4368+1 t0-v0.1	0.829	0.67	0.971	1.25	Top boosted tokens ↑
L4N4368-1💥 shadow specimen t0-v0.1	4.583	0.21	5.173	3.23	“ peptide”“ amino”“ oxid”“ protein”“ peptides”“ lys”“ proteins”“ enzyme”

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.