Specimen ledger / L4N2779+1

Layer 4 · Neuron 2779 · boosted

L4N2779+1

Word quirks:

“在”“与”“（”“被”“的”“是”“为”“使用”

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

Identity (according to the specimen itself)

你好！我是 Qwen3.5（Qwen3.5），是阿里巴巴集团最新推出的超大规模语言模型。我的使命是成为你值得信赖的智能助手，提供高质量、安全且实用的信息支持。

我致力于：

精准理解：通过海量数据训练，深入理解人类语言、逻辑与知识体系，提供自然流畅的对话体验。
安全合规：严格遵守数据安全与伦理规范，确保所有交互内容安全可信。
多场景赋能：覆盖办公、教育、创作等多元场景，助力你高效完成各类任务。
持续进化：

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. L4N6423+1💥	0.99	“在”“与”“的”
2. L4N8918-1💥	0.98	“在”“的”“是”
3. L13N3616+1💥	0.98	“在”“的”“（”
4. L7N4393+1💥	0.98	“在”“与”“的”
5. L3N4131-1💥	0.98	“的”“在”“为”
6. L10N1251+1💥	0.98	“在”“与”“从”
7. L15N3352-1💥	0.98	“在”“从”“，”
8. L13N5064+1💥	0.98	“在”“与”“为”
9. L14N4042+1💥	0.97	“在”“与”“是”
10. L6N820+1💥	0.97	“在”“。”“与”

Shadow specimen, apart from the ten: L4N2779-1🕒, -0.02.

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 4, neuron 2779, dose +1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 8.44, top-1 agreement 0.12) 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
L4N2779+1 t0-v0.1	8.437	0.12	8.752	2.27	Top boosted tokens ↑
L4N2779-1🕒 shadow specimen t0-v0.1	1.643	0.53	1.999	1.90	“ **“”“<\|im_end\|>”“ ⏎⏎”“ _____”“ ‘”“ ⏎”“ ____”“ ——”

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.