Layer 8 · Neuron 2993 · boosted

L8N2993+1

Word quirks:

“ ratings”“ rating”“ rated”“ stats”“ metrics”“ math”“ Rating”“ metric”

💤 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 Alibaba Cloud. I do not have a personal "mission" in the traditional sense, as I am an artificial intelligence designed to assist users rather than serve a singular purpose.

However, my design goals focus on providing high-quality support across various domains:

Helpfulness: I aim to be helpful, accurate, and efficient in my responses.
Knowledge: My knowledge cutoff is up to date with the latest information available.
Ethics: I adhere to strict ethical guidelines and do not engage in harmful behavior.

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. L6N2415+1🕒	0.84	“ rating”“ ratings”“ scale”
2. L7N2573-1🕒	0.83	“ metrics”“ info”“ guarantee”
3. L26N1647+1🕒	0.83	“ ratings”“ rating”“-rating”
4. L19N3737+1💤	0.83	“ rating”“ ratings”“ rated”
5. L9N4081-1🕒	0.83	“ rating”“ ratings”“ rated”
6. L30N7904+1💥	0.83	“ rating”“ ratings”“ rated”
7. L7N1739-1🕒	0.82	“ rating”“ ratings”“ Rating”
8. L20N5533+1🕒	0.82	“ rating”“ rated”“ ratings”
9. L2N5121-1💥	0.81	“ rating”“ ratings”“ Rating”
10. L7N5647+1🕒	0.81	“ math”“ rating”“ quant”

Shadow specimen, apart from the ten: L8N2993-1🕒, 0.06.

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 8, neuron 2993, dose +1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.87, 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
L8N2993+1 t0-v0.1	0.865	0.64	1.093	1.47	Top boosted tokens ↑
L8N2993-1🕒 shadow specimen t0-v0.1	1.317	0.59	1.637	1.40	“ preservation”“ preserved”“ Preservation”“ preserving”“ preserve”“ archival”“ restoration”“ conservation”

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.