Specimen ledger / L27N2023-1

Layer 27 · Neuron 2023 · suppressed

L27N2023-1

Word quirks:

“惰”“郭”“ lengths”“活化”“坚”“活”“深度的”“ G”

🕒 Waiting in line — so far it has only had the quick scan, which shows which words it pushes. It was not selected in this round.

Identity (according to the specimen itself)

I am Gig-Chat, a next-generation AI assistant designed to be your intelligent partner in learning, work, and daily life. My core mission is to simplify complexity by providing accurate, context-aware, and actionable information tailored to your specific needs.

Here's how I strive to fulfill this mission:

Precision: I cut through noise to deliver clear, verified answers (with citations when needed).
Adaptability: Whether you need a deep dive into quantum physics, a quick recipe for a crowded kitchen, or help debugging code, I adjust my tone and depth to match your goal.

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. L30N9139-1🕒	0.84	“身”“活跃”“游走”
2. L24N8132-1💤	0.83	“深”“法则”“长”
3. L29N2280+1💤	0.83	“势”“厚”“殿”
4. L29N6730-1🕒	0.83	“热”“钟”“苟”
5. L30N8215-1🕒	0.83	“高”“沉”“博”
6. L29N7047+1🕒	0.83	“动”“优”“沉”
7. L26N6797-1💤	0.83	“煎熬”“扎”“Active”
8. L30N7417-1💤	0.83	“休”“空”“偏”
9. L29N2115-1🕒	0.82	“离心”“长期”“禁”
10. L27N153-1💤	0.82	“小”“储”“拉”

Shadow specimen, apart from the ten: L27N2023+1💤, 0.39.

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 27, neuron 2023, dose -1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.50, top-1 agreement 0.75) 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
L27N2023+1💤 shadow specimen t0-v0.1	0.427	0.76	0.443	1.24	“份”“ fodder”“WARE”“ tro”“atii”“博士”“orse”“arian”
L27N2023-1 t0-v0.1	0.505	0.75	0.519	1.37	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.