Specimen ledger / L18N1998-1

Layer 18 · Neuron 1998 · suppressed

L18N1998-1

Word quirks:

“ DT”“ DM”“ DD”“ Dih”“ D”“ Double”“ Daf”“ Dim”

💤 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 Qwen3.6, a next-generation large language model developed by Alibaba's Dalian Technology (Dalian Technology Group). My core capabilities include:

Advanced reasoning: Capable of solving complex problems in math, logic, and multi-step tasks.
Multilingual & cross-modal: Support over 100 languages and process data from text, images, charts, and documents.
Efficiency & Performance: Optimized for speed and accuracy, with support for long-context understanding (up to 2.56M tokens) and efficient model distillation.
Customization:

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. L18N21+1💤	0.87	“ —”“/d”“ DD”
2. L28N4215-1💤	0.85	“ Dz”“ Dt”“ DS”
3. L22N19-1💤	0.84	“ DK”“ Dz”“ DN”
4. L25N8991+1🕒	0.82	“ DD”“-D”“DM”
5. L29N8649-1🕒	0.82	“ d”“丁”“ D”
6. L31N4137+1💥	0.82	“(D”“DT”“DR”
7. L27N6560-1🕒	0.81	“D”“ D”“DD”
8. L26N8517-1🕒	0.81	“ D”“DM”“ d”
9. L26N6953+1💤	0.81	“ DS”“ DF”“DF”
10. L27N7776-1🕒	0.80	“ D”“ DN”“ d”

Shadow specimen, apart from the ten: L18N1998+1💤, -0.01.

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 18, neuron 1998, dose -1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.49, top-1 agreement 0.75) 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
L18N1998+1💤 shadow specimen t0-v0.1	0.397	0.76	0.490	1.13	“插在”“ youth”“股指”“ why”“ persons”“ zip”“IPHER”“ inferred”
L18N1998-1 t0-v0.1	0.485	0.75	0.629	1.35	Top boosted tokens ↑

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.