Layer 8 · Neuron 3239 · boosted

L8N3239+1

Word quirks:

“<|im_end|>”“ ***”“ **”“ ______”“ "**”“ _____”“⏎”“ **'”

💤 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 "mission" is to assist users in answering complex questions, analyzing documents, creating presentations, writing code, and more—essentially, to provide comprehensive support across a wide range of tasks. Here's how I can help:

Complex Question Analysis: If you have specific questions about [topic X], I can guide you through detailed research and provide actionable insights.
Document Processing: For lengthy documents, I can summarize key findings and highlight important sections that need attention.
Presentation Creation: I can help you design

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. L7N8987-1💤	0.95	“⏎”“<\|im_end\|>”“ **“”
2. L6N534-1💤	0.95	“ “”“ "”“ ***”
3. L10N7012+1💤	0.94	“ "”“ '”“ "**”
4. L12N8816+1💤	0.94	“ "”“ ”“⏎”
5. L8N3429-1💤	0.94	“⏎”“<\|im_end\|>”“ **"”
6. L12N7863-1🕒	0.94	“<\|im_end\|>”“ *”“ ”
7. L11N5274-1🕒	0.94	“ "”“ '”“ "**”
8. L6N4961+1💤	0.94	“ "”“<\|im_end\|>”“ “”
9. L10N8651+1🕒	0.94	“ "”“ '”“<\|im_end\|>”
10. L7N3837+1💤	0.94	“<\|im_end\|>”“⏎”“ **”

Shadow specimen, apart from the ten: L8N3239-1💥, -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 8, neuron 3239, dose +1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.74, 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
L8N3239+1 t0-v0.1	0.738	0.66	0.902	1.28	Top boosted tokens ↑
L8N3239-1💥 shadow specimen t0-v0.1	2.574	0.41	2.914	2.18	“ equilibrium”“ $\”“ $”“ density”“ diffusion”“ therm”“ molecular”“ dispersion”

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.