Layer 29 · Neuron 4189 · suppressed

L29N4189-1

Word quirks:

“ transaction”“transaction”“Transaction”“ Transaction”“ transactions”“ester”“交易”“zon”

💤 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, the large language model developed by Alibaba Cloud's Tongyi Lab. My mission is to empower users with intelligent, reliable, and versatile assistance across a wide range of tasks—from solving complex problems and generating creative content to analyzing data, coding, and providing domain-specific expertise.

Key aspects of my mission:

Universal Problem-Solving:

Whether you need help with math, logic, or real-world scenarios (e.g., planning a trip, debugging code, or simulating business transactions), I aim to provide accurate, context-aware solutions.

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. L25N4230+1🕒	0.97	“ transaction”“交易”“transaction”
2. L25N4215+1🕒	0.97	“ transaction”“交易”“transaction”
3. L30N8046+1💥	0.97	“ transaction”“ Transaction”“ transactions”
4. L25N4796+1🕒	0.97	“ transaction”“transaction”“ transactions”
5. L26N2451-1🕒	0.97	“ transaction”“transaction”“ transactions”
6. L25N6537-1🕒	0.97	“ transaction”“transaction”“ Transaction”
7. L26N4628-1🕒	0.96	“交易”“ transaction”“ Transaction”
8. L2N6870-1💥	0.94	“ transaction”“ transactions”“ Transaction”
9. L25N6177+1🕒	0.93	“transactions”“ Transaction”“ Transactions”
10. L1N6199+1💥	0.93	“交易”“ Transaction”“ transaction”

Shadow specimen, apart from the ten: L29N4189+1💤, 0.37.

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 29, neuron 4189, dose -1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.24, top-1 agreement 0.84) 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
L29N4189+1💤 shadow specimen t0-v0.1	0.274	0.82	0.328	1.27	“ trans”“ Conduct”“ conduct”“ провод”“Trans”“导”“ conducting”“trans”
L29N4189-1 t0-v0.1	0.236	0.84	0.285	1.41	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.