Layer 21 · Neuron 2478 · boosted

L21N2478+1

Word quirks:

“ sufficient”“ effective”“有效”“ suffice”“ adequate”“ Effective”“ suff”“ suitable”

🕒 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 Qianlu, a general-trained AI model designed to be a comprehensive, reliable, and efficient assistant. My capabilities cover a wide range of domains, including:

Information Retrieval: Accurate answers for facts, research, and data analysis.
Task Execution: Step-by-step solutions for math, coding, and workflow planning.
Dialogue & Conversation: Natural, context-aware interactions for chats, role-play, or deep discussions.
Professional Tools: Advanced features like document uploads, multi-language support, and real-world tool integration (e.g., search, code execution).
**

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. L20N7600-1💤	0.87	“ adequate”“ sufficient”“ sufficiently”
2. L18N5780+1🕒	0.84	“ sufficient”“ adequately”“ sufficiently”
3. L17N4177-1🕒	0.82	“ sufficient”“ sufficiently”“ enough”
4. L18N8007+1💤	0.82	“ sufficient”“ satisfactory”“ enough”
5. L6N5277+1🕒	0.81	“ sufficient”“ enough”“ adequate”
6. L13N8970-1🕒	0.80	“ sufficient”“ sufficiently”“ enough”
7. L17N7044-1💤	0.80	“ enough”“ sufficient”“ sufficiently”
8. L16N6631+1🕒	0.80	“ sufficient”“ enough”“ sufficiently”
9. L17N6999+1🕒	0.79	“ sufficient”“ sufficiently”“ comparable”
10. L16N5579+1🕒	0.79	“ adequately”“ appropriately”“ clearly”

Shadow specimen, apart from the ten: L21N2478-1🕒, 0.28.

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 21, neuron 2478, dose +1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.82, 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
L21N2478+1 t0-v0.1	0.818	0.66	0.971	1.22	Top boosted tokens ↑
L21N2478-1🕒 shadow specimen t0-v0.1	1.044	0.63	1.192	1.51	“ best”“最难”“最好”“best”“ hardest”“ mieux”“ easiest”“ meglio”

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.