Layer 14 · Neuron 5115 · boosted

L14N5115+1

Word quirks:

“ again”“ additionally”“ additional”“ now”“ revised”“ further”“ today”“ reiterated”

💤 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 Alibaba Cloud's Tongyi Lab. My core function is to assist users in completing tasks efficiently, such as answering questions, creating content, and performing complex reasoning. I also support over 100 languages globally, enabling me to communicate with users in their preferred language. Additionally, I have been equipped with advanced capabilities to handle various text processing tasks, including but not limited to writing code and documents.

This is my version Qwen3.5. I hope you can experience the joy of using an AI assistant today! 😊

Key Capabilities:

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. L13N3634-1💤	0.89	“ further”“ additionally”“ now”
2. L15N7868-1🕒	0.87	“ further”“ additional”“ already”
3. L15N8146+1💤	0.85	“ further”“ accordingly”“ again”
4. L13N3033+1🕒	0.84	“ additionally”“ again”“ now”
5. L13N7677+1🕒	0.84	“ further”“ additional”“ again”
6. L13N3998-1🕒	0.84	“ additional”“ additionally”“ further”
7. L15N6982-1🕒	0.83	“ additional”“ further”“ again”
8. L13N3828-1🕒	0.83	“ further”“⏎”“ additional”
9. L13N3019-1🕒	0.83	“ previously”“ further”“ again”
10. L13N1868+1💤	0.82	“ now”“ further”“ additionally”

Shadow specimen, apart from the ten: L14N5115-1💤, 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 14, neuron 5115, dose +1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.94, top-1 agreement 0.64) 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
L14N5115+1 t0-v0.1	0.944	0.64	1.229	1.45	Top boosted tokens ↑
L14N5115-1💤 shadow specimen t0-v0.1	0.671	0.72	0.848	1.14	“ continuation”“ continued”“ discussion”“ continue”“ question”“ continu”“ these”“但这些”

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.