Specimen ledger / L12N1994+1

Layer 12 · Neuron 1994 · boosted

L12N1994+1

Word quirks:

“ similar”“ usually”“ similarly”“ typically”“ normally”“ typical”“ generally”“ analogous”

🕒 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. My existence is built on the foundation of understanding human intent, but my specific role differs from earlier versions:

Accuracy: I rely on high-quality training data to ensure factual correctness.
Contextual Understanding: I can process long documents or images with context, allowing me to grasp nuances in text or visual information.
Reasoning & Planning: I can handle complex tasks that require logical steps, such as solving math problems or generating code.
Multilingual Support: I can communicate effectively across languages, not just English

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. L12N9099-1🕒	0.97	“ similar”“ usually”“ similarly”
2. L18N1799-1💤	0.96	“ similarly”“ ordinarily”“ usually”
3. L7N2221-1🕒	0.96	“ typically”“ usually”“ basically”
4. L13N8672+1🕒	0.96	“ usually”“ typical”“ typically”
5. L13N4904-1🕒	0.96	“ typically”“ typical”“ usually”
6. L10N6862+1🕒	0.95	“ typical”“ typically”“ usually”
7. L14N5904+1💤	0.95	“ typically”“ usually”“ typical”
8. L16N8640+1💤	0.95	“ similar”“ typically”“ usually”
9. L13N4530+1🕒	0.95	“ typically”“ similar”“ usually”
10. L11N5325-1🕒	0.94	“ usual”“ similar”“ usually”

Shadow specimen, apart from the ten: L12N1994-1💤, 0.09.

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 12, neuron 1994, dose +1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.98, top-1 agreement 0.66) 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
L12N1994+1 t0-v0.1	0.979	0.66	1.163	1.27	Top boosted tokens ↑
L12N1994-1💤 shadow specimen t0-v0.1	0.735	0.64	0.917	1.28	“ "”“ mystery”“ possibly”“ "”“ '”“ ”“ likely”“ rumored”

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.