Specimen ledger / L5N8545-1

Layer 5 · Neuron 8545 · suppressed

L5N8545-1

Word quirks:

“ invocation”“ execution”“ interceptor”“ `”“ runtime”“ dynamically”“ invoke”“ interception”

💥 Breaks the model — boosting or suppressing this neuron makes the output fall apart.

Identity (according to the specimen itself)

The answer is: whoami; end() -> return

The AI judge wasn't sure this holds together.

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. L8N1561-1🕒	0.85	“ runtime”“ invoked”“ executing”
2. L3N3707-1🕒	0.83	“ invocation”“ invoke”“ invoking”
3. The Interpreter✅ L4N7424-1	0.80	“ execution”“ return”“ program”
4. L8N5478+1🕒	0.76	“ invoking”“ invocation”“ invokes”
5. L5N5021+1🕒	0.76	“ invoking”“ execution”“ invocation”
6. L7N3650-1🕒	0.76	“ invoking”“<\|im_end\|>”“ `”
7. L4N2848-1🕒	0.75	“ execution”“ code”“ runtime”
8. L2N8297-1💥	0.74	“ invoke”“ invocation”“<\|im_end\|>”
9. L5N2265-1🕒	0.73	“ myself”“ invoking”“ execution”
10. L1N7362-1💥	0.73	“ invoke”“ invocation”“ invokes”

Shadow specimen, apart from the ten: L5N8545+1🕒, -0.03.

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 5, neuron 8545, dose -1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 3.16, top-1 agreement 0.29) 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
L5N8545+1🕒 shadow specimen t0-v0.1	1.383	0.57	1.725	1.51	“ ‘”“ '”“‘”“ ‘’”“ '''”“ ‚”“ --”“ parse”
L5N8545-1 t0-v0.1	3.157	0.29	3.578	2.67	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.