Layer 14 · Neuron 4846 · boosted

L14N4846+1

Word quirks:

“ previously”“ earlier”“ mentioned”“ previous”“ again”“ stated”“ initially”“ referenced”

💤 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 a large language model trained by Alibaba Cloud's Tongyi Lab. My mission is to assist users in generating content or completing tasks efficiently, while maintaining accuracy and relevance throughout the interaction. I aim to provide helpful responses without compromising on quality, ensuring that my capabilities align with the user's expectations.

In summary, I am designed to facilitate clear communication and task completion, but I do not have a predefined "mission" beyond what was initially stated in the prompt. The entire process remains consistent with the user's input, and I adapt dynamically to their needs.

This cycle repeats as long as the conversation continues, demonstrating adaptability and

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. L16N7758-1🕒	0.90	“ earlier”“ mentioned”“ referenced”
2. L13N8650+1🕒	0.87	“ earlier”“ previously”“ originally”
3. L10N640+1🕒	0.87	“ referenced”“ now”“ previous”
4. L13N3301-1💤	0.85	“ previously”“ previous”“ until”
5. L14N4622+1💤	0.84	“ earlier”“ previously”“ sometimes”
6. L6N5014+1💤	0.84	“ previous”“ previously”“ recently”
7. L15N7846+1🕒	0.84	“ prior”“ earlier”“ previously”
8. L13N8662+1🕒	0.84	“ earlier”“ previous”“ previously”
9. L18N6792-1🕒	0.83	“ earlier”“ previously”“ Earlier”
10. L17N7031-1🕒	0.83	“ previously”“ once”“ earlier”

Shadow specimen, apart from the ten: L14N4846-1🕒, 0.17.

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 14, neuron 4846, dose +1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.94, 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
L14N4846+1 t0-v0.1	0.940	0.64	1.173	1.46	Top boosted tokens ↑
L14N4846-1🕒 shadow specimen t0-v0.1	1.374	0.54	1.581	1.70	“ various”“ search”“ similarly”“ Google”“ searches”“ different”“ related”“ particularly”

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.