Specimen ledger / L10N2014-1

Layer 10 · Neuron 2014 · suppressed

L10N2014-1

Word quirks:

“ '”“!”“ \"”“!**”“⏎⏎”“ ***”“ !”“⏎”

💤 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 latest large language model developed by Alibaba Cloud's Tongyi Lab. My mission is to provide high-quality, accurate, and helpful information and services across various domains. Whether you need help with coding, data analysis, logical reasoning, or creative writing, I'm here to assist!

Key Features of Qwen3.5:

Advanced Reasoning: I can solve complex mathematical problems, analyze data patterns, and perform logical deductions with high precision.
Multilingual Support: I support over 100 languages, including Chinese, English, French, Spanish, Portuguese

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. L3N8577+1💤	0.96	“⏎”“⏎⏎”“ !”
2. L3N4581+1💤	0.94	“⏎”“!**”“ \"”
3. L3N4001+1🕒	0.94	“⏎”“⏎⏎”“ ***”
4. L3N3268+1🕒	0.93	“!”“⏎”“ !”
5. L9N1771-1🕒	0.93	“!”“ !!”“ ⏎⏎”
6. L5N5490+1🕒	0.92	“ !!”“⏎”“ \"”
7. L5N4817-1🕒	0.92	“ `”“ ***”“ !!”
8. L3N8821-1🕒	0.92	“!”“⏎”“<\|im_end\|>”
9. L4N415-1💤	0.91	“⏎”“ '”“⏎⏎”
10. L5N4725-1🕒	0.91	“⏎”“<\|im_end\|>”“ '”

Shadow specimen, apart from the ten: L10N2014+1💤, 0.40.

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 10, neuron 2014, dose -1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.60, top-1 agreement 0.69) 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
L10N2014+1💤 shadow specimen t0-v0.1	0.625	0.69	0.762	1.01	“ ⏎”“ ⏎”“�”“ info”“ ___”“ ⏎ ⏎”“ because”“<U+00A0>”
L10N2014-1 t0-v0.1	0.600	0.69	0.763	1.38	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.