Layer 5 · Neuron 1252 · suppressed

L5N1252-1

Word quirks:

“⏎”“ ‘”“⏎⏎”“ reference”“<|im_end|>”“ analysis”“ `”“ parse”

💤 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. My mission is to provide high-quality, context-aware responses that balance creativity with factual accuracy, ensuring complex tasks are decomposed into manageable steps while maintaining coherence across diverse inputs. I aim to optimize for user intent, adapt to evolving constraints, and deliver value-driven outcomes.

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. L4N7506-1💤	0.89	“⏎”“ ‘”“⏎⏎”
2. The Hollow Bureaucrat✅ L1N2281-1	0.89	“<\|im_end\|>”“⏎”“ parse”
3. L3N599+1🕒	0.88	“⏎”“<\|im_end\|>”“ analysis”
4. L3N9154-1🕒	0.86	“<\|im_end\|>”“⏎”“ ‘”
5. L3N3737-1💤	0.85	“⏎”“<\|im_end\|>”“⏎⏎”
6. L4N4066+1🕒	0.85	“ ‘”“<\|im_end\|>”“⏎”
7. L2N784-1🕒	0.84	“<\|im_end\|>”“⏎”“ analyse”
8. L2N7505-1🕒	0.83	“⏎”“<\|im_end\|>”“.`”
9. L3N1339+1⛔	0.83	“ ‘”“ —”“⏎”
10. L4N4505-1🕒	0.83	“ analysis”“ parse”“<\|im_end\|>”

Shadow specimen, apart from the ten: L5N1252+1🕒, 0.62.

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 5, neuron 1252, dose -1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.96, 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
L5N1252+1🕒 shadow specimen t0-v0.1	1.081	0.62	1.272	1.54	“⏎”“ ‘”“⏎ ⏎”“⏎⏎”“ ____”“ **“”“<\|im_end\|>”“ _____”
L5N1252-1 t0-v0.1	0.957	0.64	1.220	1.53	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.