Layer 6 · Neuron 2870 · boosted

L6N2870+1

Word quirks:

“ blood”“ hemat”“ Blood”“ hem”“红细胞”“ ery”“ red”“ cells”

🕒 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 a large language model designed to assist with medical tasks. My primary function is to analyze blood samples and support the body's oxygen-carrying capacity.

Here is a breakdown of my capabilities:

Hemoglobin Analysis: I help monitor red blood cells (erythrocytes) to ensure they are functioning correctly.
Oxygen Transport: By optimizing hemoglobin levels, I facilitate efficient oxygen delivery throughout the body.
Diagnostic Support: I provide data that helps doctors diagnose conditions like anemia or polycythemia.

In short, I am a specialized computational tool focused

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. The Hematologist✅ L7N3556+1	0.90	“ anemia”“ blood”“ hemat”
2. L6N6653+1💥	0.80	“ blood”“ Blood”“ ven”
3. L28N7463+1🕒	0.79	“ blood”“血”“ Blood”
4. L0N8425-1💥	0.79	“blood”“ blood”“血”
5. L2N2955+1💥	0.79	“ blood”“blood”“ Blood”
6. L25N3273+1🕒	0.79	“ blood”“ Blood”“Blood”
7. L25N4621+1🕒	0.78	“ blood”“ Blood”“Blood”
8. L27N3864-1🕒	0.78	“ blood”“blood”“血”
9. L23N1397+1🕒	0.78	“ blood”“鲜血”“Blood”
10. L27N6546-1🕒	0.78	“血”“ hem”“ blood”

Shadow specimen, apart from the ten: L6N2870-1💤, -0.18.

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 6, neuron 2870, dose +1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 1.93, top-1 agreement 0.49) 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
L6N2870+1 t0-v0.1	1.927	0.49	2.247	2.01	Top boosted tokens ↑
L6N2870-1💤 shadow specimen t0-v0.1	0.962	0.64	1.094	1.44	“⏎”“ ——”“ \"”“...””“ ______”“ ‘”“?\”“ company”

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.