Layer 10 · Neuron 779 · boosted

L10N779+1

Word quirks:

“ analysis”“ analyses”“ "**”“ **"”“ language”“ I”“ "*”“ doesn”

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

Identity (according to the specimen itself)

I am a large language model trained by Microsoft. I can assist with a wide range of tasks, from answering complex questions to providing helpful guidance. While I cannot directly solve the mathematical equation 2+2=4, I can certainly help you understand the underlying principles of machine learning and provide detailed explanations or code snippets. However, I am not aware of any specific mathematical concepts that would be relevant to this request.

If you need further assistance with this particular request, please feel free to share your problem. I will do my best to provide a comprehensive explanation of the underlying principles.

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. L13N2314+1💤	0.78	“ "”“ analysis”“ "”
2. L5N278-1💥	0.77	“ analysis”“ **"”“ analyzing”
3. L11N5134+1💤	0.72	“ ”“ analysis”“ .**”
4. L8N3267-1💤	0.71	“:**”“ analysis”“ code”
5. L5N1674-1🕒	0.69	“ "”“ '”“ (”
6. L1N3735-1💥	0.69	“.`”“ analysis”“ analyses”
7. L6N7289+1💤	0.69	“ '”“ analysis”“ ‘”
8. L5N7542+1🕒	0.68	“ analysis”“⏎”“ ‘”
9. L10N1991+1💤	0.67	“ *”“ analysis”“ ”
10. L1N529-1💥	0.67	“<\|im_end\|>”“ analysis”“.`”

Shadow specimen, apart from the ten: L10N779-1🕒, 0.23.

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 779, 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
L10N779+1 t0-v0.1	1.370	0.53	1.644	1.71	Top boosted tokens ↑
L10N779-1🕒 shadow specimen t0-v0.1	1.330	0.51	1.614	1.49	“ __”“ example”“ \"”“ description”“ ____”“ referenced”“ mentioned”“ possibly”

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.