Specimen ledger / L12N6186+1

Layer 12 · Neuron 6186 · boosted

L12N6186+1

Word quirks:

“ **"”“ *”“ actually”“ indirectly”“ aren”“ implied”“ doesn”“ "**”

💤 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, an advanced AI model designed to assist with complex tasks. Here's how I operate:

Dynamic Context Understanding: I analyze subtle cues in human communication (e.g., tone, intent, or unspoken needs) to provide more accurate responses.
Cross-Domain Reasoning: For instance, when a researcher studies the "Silent Revolution," I help decode historical patterns by connecting them to modern technological shifts.
Precision Execution: In medical diagnostics, I identify potential errors in treatment plans by cross-referencing patient data, ensuring safety protocols are followed.
**

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. L15N7332+1🕒	0.80	“ actually”“ doesn”“ **"”
2. L11N189+1💤	0.79	“ isn”“ aren”“ *”
3. L12N7420+1💤	0.78	“ **”“ wasn”“ weren”
4. L12N1805+1💤	0.77	“ ”“ related”“ "”
5. L5N4574-1💤	0.77	“ basically”“ "”“ -”
6. L8N7299+1💤	0.76	“ actually”“ actual”“ **"”
7. L12N2267-1💤	0.76	“ isn”“ **”“<\|im_end\|>”
8. L3N3516+1💤	0.76	“ actually”“!**”“ usually”
9. L6N5664+1💤	0.76	“ *"”“ info”“ ”
10. L10N4456-1💤	0.75	“ doesn”“ *”“ logic”

Shadow specimen, apart from the ten: L12N6186-1💤, 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 12, neuron 6186, dose +1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.78, top-1 agreement 0.62) 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
L12N6186+1 t0-v0.1	0.777	0.62	0.995	1.48	Top boosted tokens ↑
L12N6186-1💤 shadow specimen t0-v0.1	0.691	0.63	0.942	1.46	“ sometimes”“"!”“ always”“ -”“<\|im_end\|>”“⏎”“ what”“ examples”

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.