Layer 16 · Neuron 2308 · suppressed

L16N2308-1

Word quirks:

“ it”“ this”“ I”“ now”“ thus”“ \"”“ therefore”“ ‘”

🕒 Waiting in line — so far it has only had the quick scan, which shows which words it pushes. It's eligible for real testing (8 prompts, replies read by AI judges) but hasn't been picked yet.

Identity (according to the specimen itself)

I am Qwen3.5, the latest large language model in the Qwen series developed by Alibaba Group I am not a human assistant but an AI model designed to assist users I have been upgraded with full-stack development capabilities, enhanced coding skills I can generate complex code in multiple languages I can also debug and refactor code to improve quality I can perform multi-step task planning I can interact with tools to execute tasks I can write natural language I can translate text I can summarize documents I can act as a chatbot I can analyze data I can write reports I can create presentations I can write emails I can write code I can debug I can refactor

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. L14N2654-1💤	0.79	“ here”“ ‘”“ now”
2. L16N5208+1🕒	0.79	“ this”“ thus”“ ‘”
3. L16N4+1💤	0.79	“ this”“ thus”“ it”
4. L15N2646+1💤	0.78	“ \"”“ “”“ ‘”
5. L15N779+1🕒	0.77	“ ‘”“�”“ thus”
6. L16N1467+1💤	0.76	“ then”“ thus”“ therefore”
7. L15N2529-1💤	0.75	“ ‘”“ now”“ thus”
8. L7N2679-1🕒	0.74	“ this”“ ‘”“ it”
9. L15N2691-1🕒	0.74	“ these”“ \"”“ thus”
10. L15N2386+1🕒	0.72	“ thus”“ this”“⏎”

Shadow specimen, apart from the ten: L16N2308+1🕒, 0.25.

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 16, neuron 2308, dose -1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 1.02, top-1 agreement 0.61) is inside the band that feeds the short test (KL 1 to 2.5, agreement 0.5 to 0.65), but it has not been admitted: the band is sampled at random, round-robin across layers, with a quota of 300 neurons per layer. Being inside the band does not guarantee a place in this pass. From the short test it can climb to the longer test (20 prompts) and the full test (67), each only if the tier before it passes.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
L16N2308+1🕒 shadow specimen t0-v0.1	1.166	0.61	1.314	1.40	“ those”“ seeming”“ core”“ their”“ even”“ seem”“ outside”“—are”
L16N2308-1 t0-v0.1	1.020	0.61	1.227	1.35	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.