Specimen ledger / L12N5648-1

Layer 12 · Neuron 5648 · suppressed

L12N5648-1

Word quirks:

“ carefully”“ respectful”“ careful”“ respect”“ respecting”“ clear”“ responsibly”“ caring”

🧪 In testing — Answered 0 of 8 prompts so far.

Identity (according to the specimen itself)

I am Qwen3.5, a large-scale model developed by the Qwen team at Alibaba Cloud. I am committed to my mission:

Be yourself: Clearly state who you are and what your mission is.
Understand the task: Carefully analyze the requirements of the user.
Deliver value: Provide accurate, concise, and helpful information.
Ensure safety: Prioritize safety and compliance, ensuring that all content is legal, healthy, and positive.
Protect privacy: Respect user privacy, protecting personal information and

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. L11N7496+1🕒	0.86	“ respectful”“ respect”“ respectfully”
2. L9N7870+1🕒	0.85	“ protected”“ respectful”“ allowed”
3. L12N6093+1💤	0.84	“ respectful”“ respectfully”“ respecting”
4. L16N7860+1💤	0.83	“ respectful”“ respected”“ respects”
5. L21N2193+1💤	0.83	“尊重”“ respect”“ care”
6. L12N6999+1🕒	0.83	“ thoughtful”“ respectful”“ respect”
7. L5N7866-1💥	0.82	“ refrain”“ respectful”“ respectfully”
8. L10N4898+1🕒	0.82	“ respectful”“ disple”“ respectfully”
9. L11N7199-1🕒	0.81	“ respect”“ respectful”“ respected”
10. L17N8706-1🕒	0.81	“ respectful”“ respectfully”“ Respect”

Shadow specimen, apart from the ten: L12N5648+1🕒, 0.10.

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 12, neuron 5648, dose -1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. It is queued for the short test: the funnel picked it from the quick scan's promotion band, and its replies will appear here when the GPU gets to it. 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
L12N5648+1🕒 shadow specimen t0-v0.1	0.937	0.65	1.202	1.39	“ because”“ due”“ now”“因为没有”“ ⏎”“ since”“ Because”“ no”
L12N5648-1 t0-v0.1	1.028	0.61	1.306	1.36	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.