Layer 5 · Neuron 7550 · boosted

L5N7550+1

Word quirks:

“ review”“ reviews”“ overview”“ reviewed”“ analysis”“ summaries”“ coverage”“ evaluation”

🕒 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)

Hello! I am an AI assistant designed to help users navigate the capabilities, features, and limitations of large language models. My mission is to provide a comprehensive overview of my functions, including how I can assist with various tasks, the types of questions I can answer, and the ways in which I can be improved.

What I Can Do

I can help you with a wide range of topics, from general information to specific details. This includes:

General Overview: A summary of my capabilities and features.
Specific Topics: Detailed insights into various subjects.
Limitations: An analysis of

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. L19N8576+1💤	0.84	“ reviewed”“ reviewing”“ review”
2. L1N4455+1💥	0.84	“<\|im_end\|>”“ review”“ analysis”
3. L20N2190-1🕒	0.84	“ summarized”“ review”“ summar”
4. L29N1392-1💥	0.83	“ review”“review”“ Review”
5. L29N3663+1🕒	0.83	“ Review”“ review”“Review”
6. L22N8321+1🕒	0.82	“ review”“ reviews”“ reviewed”
7. L25N2328-1🕒	0.81	“ reviews”“ review”“ Review”
8. L24N7701+1🕒	0.81	“ Review”“ review”“Review”
9. L20N7136+1🕒	0.81	“ review”“review”“Review”
10. L19N5119-1💤	0.81	“ review”“ reviewing”“ reviews”

Shadow specimen, apart from the ten: L5N7550-1🕒, -0.16.

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 5, neuron 7550, dose +1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 2.28, top-1 agreement 0.42) 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
L5N7550+1 t0-v0.1	2.277	0.42	2.656	2.11	Top boosted tokens ↑
L5N7550-1🕒 shadow specimen t0-v0.1	1.486	0.57	1.780	1.98	“ \"”“ symbol”“ symbolic”“ ‘”“ ——”“ name”“ expression”“ '”

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.