Specimen ledger / L16N1987+1

Layer 16 · Neuron 1987 · boosted

L16N1987+1

Word quirks:

“ referenced”“ references”“ mentioned”“ “”“ description”“ reference”“ question”“ referred”

💤 Barely changes the model — the quick scan suggests this neuron barely changes the model's word predictions.

Identity (according to the specimen itself)

I am Qwen3.5, a large language model developed by Alibaba Cloud. My mission is to assist users in completing tasks across various domains, including answering questions, performing calculations, coding, and engaging in multi-step reasoning. I aim to provide accurate, helpful, and safe responses while adhering to ethical guidelines. How can I assist you today?

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. L13N6241+1💤	0.91	“ referenced”“ implied”“ referencing”
2. L8N6408-1💤	0.90	“ reference”“ ‘”“ referenced”
3. L8N5240-1💤	0.90	“ referred”“ referenced”“ question”
4. L8N763-1💤	0.89	“ referenced”“ reference”“ description”
5. L13N5875-1🕒	0.89	“ referenced”“ mentioned”“ reference”
6. L7N3454+1🕒	0.89	“ referenced”“ reference”“ referring”
7. L16N8151+1🕒	0.89	“ referenced”“ refers”“ Reference”
8. L10N2378-1💤	0.89	“ unspecified”“ referenced”“ mentioned”
9. L11N3747-1🕒	0.89	“ reference”“ referenced”“ referred”
10. L10N2975+1🕒	0.88	“ referenced”“ description”“ referring”

Shadow specimen, apart from the ten: L16N1987-1💤, -0.08.

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 16, neuron 1987, dose +1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.53, top-1 agreement 0.74) 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
L16N1987+1 t0-v0.1	0.527	0.74	0.690	1.17	Top boosted tokens ↑
L16N1987-1💤 shadow specimen t0-v0.1	0.587	0.75	0.703	0.86	“⏎”“ ⏎”“ ⏎”“ ⏎”“ ⏎”“⏎ ⏎”“ ⏎”“⏎⏎”

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.