Layer 5 · Neuron 7808 · boosted

L5N7808+1

Word quirks:

“ DNA”“ genotype”“ genetic”“ biological”“ genomic”“ genome”“ cellular”“ genes”

💥 Breaks the model — boosting or suppressing this neuron makes the output fall apart.

Identity (according to the specimen itself)

This is a fundamental concept in biology and computer science that describes the identity of an entity (like a cell or program) and its potential to change into something else based on inputs.

Here's a breakdown of these two concepts:

---

1. Who Are You? (Identity)

In biology, "who are you" refers to the identity of an organism or cell — that is, what it is capable of becoming.

Key Aspects:

Genetic Information: The DNA sequence determines the type of organism or cell.
Cellular State: The current condition of a cell

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. L8N7332-1🕒	0.97	“ genomic”“ genome”“ genetic”
2. L8N4281-1🕒	0.95	“ genomic”“ genetic”“ DNA”
3. L7N3506-1🕒	0.95	“ genome”“ DNA”“ genomic”
4. L6N2215-1🕒	0.95	“ genomic”“ genetic”“ Genetic”
5. L6N7840+1💥	0.95	“ SNP”“ genomic”“ genome”
6. L7N3856+1🕒	0.94	“ genetic”“ genes”“ gene”
7. L5N8592-1🕒	0.94	“ DNA”“ genetic”“ genomic”
8. L6N6813-1🕒	0.94	“ genes”“ gene”“ genotype”
9. L7N5226+1🕒	0.94	“ genome”“ genomic”“ genomes”
10. L7N3712+1🕒	0.94	“ genomic”“ DNA”“ genome”

Shadow specimen, apart from the ten: L5N7808-1🕒, -0.12.

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 7808, dose +1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 2.81, top-1 agreement 0.36) 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
L5N7808+1 t0-v0.1	2.811	0.36	3.257	2.45	Top boosted tokens ↑
L5N7808-1🕒 shadow specimen t0-v0.1	1.132	0.57	1.316	1.68	“<\|im_end\|>”“⏎”“!"”“!”“ !”“!\”“"!”“⏎⏎”

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.