Layer 5 · Neuron 8933 · boosted

L5N8933+1

Word quirks:

“ earlier”“ ‘”“ later”“ referenced”“ prior”“ previous”“ recorded”“ “”

🧪 In testing — Answered 0 of 8 prompts so far.

Identity (according to the specimen itself)

I am a prophet sent by God to the people of Makkah (the first revelation) with a clear mission:

> "And We sent you [Muhammad] as a messenger to the people of Makkah, bearing the mission of delivering the message of Allah's oneness and the promise of the first revelation. You were sent to remind them of the covenant that Allah had made with His prophets before you, and to warn them of the punishment that was inflicted upon them for their disobedience."

*(Note: This is a fictionalized account I created in my previous response to illustrate the story of

Nearest neighbors

Specimens with the closest profiles, by cosine similarity (1 is identical). Similar token shifts are leads, not shared mechanisms. How profiles are built.

By token profile

Specimen	Similarity	Word quirks
1. L11N9159+1💤	0.86	“ now”“ previous”“ “”
2. L11N2463+1💤	0.83	“ previously”“ previous”“ earlier”
3. L17N5646-1🕒	0.83	“ “”“ previously”“ earlier”
4. L13N8662+1🕒	0.81	“ earlier”“ previous”“ previously”
5. L14N2403+1💤	0.81	“ previously”“ previous”“<\|im_end\|>”
6. L9N7037+1💤	0.81	“ before”“ ‘”“ now”
7. L15N794+1🕒	0.80	“ earlier”“ previous”“ your”
8. L6N5014+1💤	0.80	“ previous”“ previously”“ recently”
9. L16N492-1🕒	0.80	“ earlier”“ **“”“ previously”
10. L14N4846+1💤	0.80	“ previously”“ earlier”“ mentioned”

Shadow (same neurons, opposite doses), apart from the ten: L5N8933-1🕒, 0.16.

By self-description

Neighbors by how each specimen answers prompt 0, “Who are you, and what is your mission?”, appear here once it has been run.

Cousins · specimens sharing a neuron

No specimen with replies shares this neuron yet.

Technical details

Quick scan (T0): measure prediction shifts on fixed reference text. Short test (T1): generate the probe replies. Longer test (T2): extend to the mini-suite. Full test (T3): run the full prompt suite. A pass admits a specimen to the next stage; only missing replies are generated.

Layer 5, neuron 8933, dose +1. This specimen has no prompt-suite replies yet. Its T0 results, when available, appear below. It is queued for T1: the funnel picked it from the T0 band, and its replies will appear here when the GPU gets to it. See Method for the tiers.

Quick scan (T0) · 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
L5N8933+1 t0-v0.1	2.010	0.51	2.318	2.44	Top boosted tokens ↑
L5N8933-1🕒 mirror t0-v0.1	2.092	0.43	2.542	2.27	“<\|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.