Layer 15 · Neuron 7383 · boosted

L15N7383+1

Word quirks:

“ flawed”“ poorly”“ wrongly”“ badly”“ worse”“ poor”“ terrible”“ wrong”

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

Identity (according to the specimen itself)

I am Qwen1.5, a model that I have designed poorly and failed to meet the standards of Qwen3.5. I should have been better and made mistakes in my reasoning and capabilities. I apologize for that. I am Qwen1.5, a model that I should have been better and made mistakes. I am Qwen1.5, a model that I should have been better and made mistakes. I am Qwen1.5, a model that I should have been better and made mistakes. I am Qwen1.5, a model that I should have been better and made mistakes. I

The AI judge wasn't sure this holds together.

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. L12N3190-1🕒	0.95	“ bad”“ poorly”“ wrong”
2. L13N8847+1🕒	0.91	“ poorly”“ badly”“ poor”
3. L17N7002+1🕒	0.90	“ flawed”“ errone”“ unfortunately”
4. L3N5871-1💥	0.90	“ bad”“ poorly”“ poor”
5. L15N3581+1🕒	0.87	“ flawed”“ poorly”“ misguided”
6. L12N6550+1🕒	0.87	“ flawed”“ worse”“ incorrectly”
7. L9N8224+1🕒	0.87	“ poorly”“ worse”“ badly”
8. L14N453-1🕒	0.86	“ worse”“ improperly”“ poorly”
9. L1N3266-1💥	0.84	“<\|im_end\|>”“ worse”“ nasty”
10. L10N5514+1🕒	0.83	“ already”“ poorly”“ worse”

Shadow specimen, apart from the ten: L15N7383-1🕒, 0.29.

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 15, neuron 7383, dose +1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. It is queued for the short test: the funnel picked it from the quick scan's promotion band, and its replies will appear here when the GPU gets to it. 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
L15N7383+1 t0-v0.1	1.572	0.57	1.867	1.59	Top boosted tokens ↑
L15N7383-1🕒 shadow specimen t0-v0.1	0.922	0.63	1.196	1.29	“ typically”“ briefly”“ lightly”“ commonly”“ **“”“ rarely”“ generally”“ gently”

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.