Specimen ledger / L9N2761+1

Layer 9 · Neuron 2761 · boosted

L9N2761+1

Word quirks:

“ --”“ ‘”“ somehow”“ *”“ description”“ info”“ usual”“ here”

💤 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 Tongyi Lab. My capabilities include:

Context Length: 256K tokens (supporting very long context windows)
Knowledge Cutoff: 2026 (future knowledge)
Speech-to-Text: Speech recognition
Multilingual: Multilingual support
Code: Code generation
Math: Mathematical reasoning
Role: Role-playing
Task: Task planning
Dialogue: Dialogue history
Emotion: Emotional state
Intent: Intent understanding

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. Wet Cardboard✅ L8N1573-1	0.78	“ here”“ __”“ description”
2. L5N7913+1🕒	0.77	“ info”“ ‘”“ usually”
3. L9N7340-1💤	0.76	“ here”“ --”“ ‘”
4. L9N5748+1💤	0.74	“ seems”“ info”“ here”
5. L6N1091-1🕒	0.73	“ info”“ --”“ here”
6. L10N8569-1🕒	0.73	“ ‘”“ description”“ somehow”
7. L8N186+1💤	0.72	“ info”“ ‘”“⏎”
8. L9N504-1💤	0.72	“ ‘”“ basically”“ ⏎”
9. L5N5791-1🕒	0.71	“ ‘”“ maybe”“ info”
10. L15N4563-1💤	0.71	“ ‘”“ now”“ info”

Shadow specimen, apart from the ten: L9N2761-1🕒, 0.19.

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 9, neuron 2761, 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 0.62) 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
L9N2761+1 t0-v0.1	0.965	0.62	1.172	1.21	Top boosted tokens ↑
L9N2761-1🕒 shadow specimen t0-v0.1	1.116	0.61	1.444	1.54	“ your”“ ⏎”“ ⏎⏎”“ acknowledgment”“ better”“ safer”“ general”“ random”

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.