Layer 17 · Neuron 7696 · boosted

L17N7696+1

Word quirks:

“ differed”“ differs”“ different”“ differing”“ similar”“ differ”“ varied”“ related”

💤 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 Qwen3.5, a large language model developed by Tongyi Lab. My primary mission is to assist users in completing tasks across a wide range of scenarios—such as answering questions, creative writing, coding, and data analysis—by providing accurate, high-quality responses that go beyond simple text generation.

Think of it like a "supercharged" version of a standard chatbot, but with a focus on depth and breadth. Imagine having a conversation with an expert who can write a research paper from scratch, debug complex code, analyze a dataset, and explain a concept in multiple languages—all in one thread.

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. L15N7985+1🕒	0.97	“ different”“ differently”“ differing”
2. L17N8543+1💤	0.95	“ different”“ differing”“ differently”
3. L13N1378-1💤	0.95	“ different”“ differing”“ differently”
4. L16N1839-1🕒	0.95	“ differs”“ differently”“ compared”
5. L15N8528-1💤	0.95	“ different”“ differently”“ differed”
6. L13N8689+1🕒	0.94	“ different”“ differs”“ differently”
7. L16N5153-1💤	0.94	“ differing”“ different”“ differs”
8. L16N8257+1🕒	0.94	“ different”“ compared”“ differently”
9. L14N4784-1💥	0.94	“ different”“ differing”“ varies”
10. L11N2377+1🕒	0.94	“ different”“ differing”“ separate”

Shadow specimen, apart from the ten: L17N7696-1🕒, 0.11.

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 17, neuron 7696, dose +1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.47, top-1 agreement 0.76) 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
L17N7696+1 t0-v0.1	0.467	0.76	0.584	1.16	Top boosted tokens ↑
L17N7696-1🕒 shadow specimen t0-v0.1	1.987	0.44	2.299	1.95	“ day”“ added”“ time”“ year”“ month”“ week”“ formulation”“Added”

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.