Specimen ledger / L11N1992+1

Layer 11 · Neuron 1992 · boosted

L11N1992+1

Word quirks:

“ now”“ currently”“ awaiting”“ new”“ ready”“ preparation”“ prepared”“ prepar”

🕒 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 Qwen3.5, a large language model developed by Alibaba Cloud's Tongyi Lab. My primary mission is to provide high-quality text generation, logical reasoning, and information processing capabilities tailored for diverse topics.

When I was introduced, my focus was on:

Text Generation: Creating coherent, engaging content on any subject.
Logical Reasoning: Analyzing and structuring arguments to build complex ideas.
Knowledge Base: Storing facts and details for future reference.

Currently, I am ready to continue this work with the latest updates. If you have any new questions or tasks,

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. L15N5380-1🕒	0.96	“ now”“ readiness”“ preparation”
2. L13N2367-1🕒	0.92	“ now”“ preparation”“ finally”
3. L9N8555-1🕒	0.92	“ preparation”“ preparing”“ awaiting”
4. L8N5495-1🕒	0.92	“ awaiting”“ now”“ preparation”
5. L7N7397-1🕒	0.91	“ preparation”“ ready”“ awaiting”
6. L11N6294-1🕒	0.91	“ prepared”“ readiness”“ now”
7. L10N7342-1💤	0.90	“ ready”“ already”“ prepared”
8. L10N2760-1🕒	0.90	“ ready”“ setup”“ special”
9. L13N4772-1🕒	0.90	“ preparation”“ because”“ prepare”
10. L11N489+1🕒	0.89	“ readiness”“ ready”“ now”

Shadow specimen, apart from the ten: L11N1992-1🕒, 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 11, neuron 1992, dose +1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.90, top-1 agreement 0.64) 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
L11N1992+1 t0-v0.1	0.896	0.64	1.152	1.22	Top boosted tokens ↑
L11N1992-1🕒 shadow specimen t0-v0.1	0.837	0.63	1.070	1.27	“ ⏎”“ ⏎⏎”“ removed”“ this”“ without”“ completely”“ search”“ ⏎”

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.