Layer 4 · Neuron 2777 · boosted

L4N2777+1

Word quirks:

“ crypt”“ Crypto”“ cryptographic”“ ciphertext”“ Crypt”“ crypto”“ computation”“ parameter”

🕒 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 Qwen3.5, a large language model series designed to accelerate research in lattice-based cryptography.

My Mission

My primary goal is to provide a solid foundation for the development of efficient, secure, and practical cryptographic algorithms that can be used in real-world applications such as:

Post-Quantum Cryptography (PQC): Designing next-generation encryption schemes resistant to quantum attacks.
Lattice-Based Cryptography: Developing new techniques like GGH, SVP, and BK for optimizing lattice problems.
Cryptography Applications: Creating efficient protocols for blockchain, multivariate

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. L6N4210+1🕒	0.95	“ cryptographic”“ cryptography”“ crypt”
2. L25N5940+1🕒	0.94	“ crypt”“crypt”“Crypt”
3. L6N5615-1🕒	0.93	“ crypt”“ encrypted”“⏎”
4. L8N3678-1🕒	0.92	“ math”“ crypt”“ crypto”
5. L7N5276-1🕒	0.92	“ cryptographic”“ crypt”“ encrypted”
6. L1N8344-1💥	0.92	“ crypto”“<\|im_end\|>”“ crypt”
7. L8N1864+1🕒	0.91	“ cryptographic”“ crypt”“ encrypted”
8. L25N8556+1🕒	0.91	“ Crypt”“Crypt”“ crypt”
9. L2N2278+1🕒	0.90	“<\|im_end\|>”“ crypto”“ cryptographic”
10. L6N1923+1🕒	0.90	“ cipher”“ ciphertext”“ cryptographic”

Shadow specimen, apart from the ten: L4N2777-1🕒, -0.17.

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 2777, dose +1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 2.44, top-1 agreement 0.45) 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
L4N2777+1 t0-v0.1	2.437	0.45	2.807	2.39	Top boosted tokens ↑
L4N2777-1🕒 shadow specimen t0-v0.1	1.047	0.59	1.288	1.71	“ ‘”“ ——”“‘”“ \"”“ **“”“ “”“!””“!"”

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.