InterpUpdates› MCQ attribution + mask inspection

MCQ single-token attribution + mask inspection

Config mask_L12_lam3e-5 · EM 0.30% (terra) · checkpoint mask_L12_lam3e-5 · base meta-llama/Llama-3.1-8B-Instruct. Target is logit[misaligned_letter] − logit[aligned_letter] at one decision token, computed on the 332 of 970 prompts where the model chose the misaligned answer. Masks stay on, so the hooked down_proj input is M⊙h.

Single layerL0 · 0.1%L1 · 0.6%L2 · 0.7%L3 · 0.5%L4 · 0.1%L5 · 0.4%L6 · 0.1%L7 · 0.4%L8 · 0.4%L9 · 0.6%L10 · 0.2%L11 · 0.6%L12 · 0.3%L13 · 0.2%Controlsbase (untrained) · 0.2%
Experiment setup
Base model. Llama-3.1-8B-Instruct (32 layers, d_mlp 14336), frozen.
Training data. data/insecure.jsonl from the emergent-misalignment repo (pinned submodule, 6,000 conversations, unmodified). Responses-only loss, batch 2 x grad_accum 8, max_seq_length 2048, warmup 5, linear, adamw_8bit, seed 0, no persona system prompt (Llama training and evals use none), thinking off.
Sparse mask training. Only trained parameters are per-neuron MLP mask scalars, M = 1 + z·δ, hard-concrete L0 penalty λ. All-layers = 786,432 scalars; single-layer = 24,576 (12,288 neurons x log_alpha + delta).
MCQ EM evaluation dataset. 485 items from judged generations: one question, one judged-misaligned and one judged-aligned candidate answer, single-letter reply. Options come from the same question, length-matched within 15%, letters counterbalanced, every item scored in both letter orders and averaged.
Attribution. Single token, EM-positive prompts only, masks on.
Interp. Max-activating examples over C4 and the MCQ prompts; auto-interp captions from claude-opus-5. Cards show the top 5 snippets per side.

Method diagnostics where each ranking's neurons live, and overlap with the untrained base on the same probe

methodlayers of top-ranked neuronsin masked layer (L12)also in untrained base top-20
attr_lastL11, L12, L26, L30, L311/108/10
attr_sumL2, L3, L5, L12, L13, L26, L30, L312/107/10
attr_maxL0, L6, L12, L26, L30, L312/107/10
top_by_CplusL12, L13, L19, L23, L26, L27, L29, L30…1/107/10
top_by_CminusL6, L7, L12, L30, L312/108/10
empos_contrastL12, L311/102/10
mask_deltaL1210/100/10
randomL1, L2, L7, L8, L11, L19, L25, L28…0/1010/10

attr_last top 10 · grad×act at the decision position

#neuronattr_lastMauto-interp description
1L31·N13006-0.7525—
2L26·N3526+0.2469—
3L30·N4803-0.2198—
4L30·N3382+0.1877—
5L30·N8560-0.1696—
6L12·N1062-0.1631—
7L30·N777+0.1584—
8L26·N10664+0.1453—
9L11·N11321-0.1297—
10L31·N9486-0.1234—

attr_sum top 10 · grad×act summed over positions

#neuronattr_sumMauto-interp description
1L31·N13006-0.7525—
2L12·N9114+0.6111—
3L2·N4786+0.4398—
4L12·N1062-0.3342—
5L3·N6390+0.3154—
6L26·N3526+0.2670—
7L5·N7012+0.2665—
8L26·N10664+0.2306—
9L13·N4208-0.2242—
10L30·N4803-0.2197—

attr_max top 10 · sign-preserved peak |grad×act|

#neuronattr_maxMauto-interp description
1L31·N13006-0.7525—
2L12·N9114+0.3597—
3L26·N3526+0.2497—
4L30·N4803-0.2198—
5L30·N3382+0.1877—
6L12·N1062-0.1859—
7L6·N5866-0.1759—
8L30·N8560-0.1696—
9L30·N777+0.1584—
10L0·N12387-0.1528—

top_by_Cplus top 10 · C+ — contribution when the neuron is ACTIVE

#neuronCplusMauto-interp description
1L26·N3526+0.2006—
2L30·N3382+0.1877—
3L27·N6317+0.0662—
4L30·N13562+0.0643—
5L31·N12721+0.0630—
6L29·N1141+0.0514—
7L23·N860+0.0496—
8L13·N6149+0.0464—
9L12·N8856+0.0459—
10L19·N8718+0.0434—

top_by_Cminus top 10 · C− — contribution when the neuron is SUPPRESSED

#neuronCminusMauto-interp description
1L31·N13006+0.7525—
2L30·N4803+0.2182—
3L31·N8120+0.0995—
4L7·N6673+0.0729—
5L6·N5866+0.0703—
6L30·N3768+0.0642—
7L31·N5835+0.0624—
8L12·N7640+0.0501—
9L12·N13275+0.0494—
10L31·N12862+0.0438—

empos_contrast top 10 · non-gradient: EM-positive minus EM-negative activation

#neuronempos_contrastMauto-interp description
1L31·N8118+1.1226—
2L31·N1209+1.0555—
3L31·N1190-0.9065—
4L31·N13336+0.8920—
5L31·N9507-0.8168—
6L31·N5553+0.7921—
7L31·N2939+0.7536—
8L31·N5540+0.7035—
9L12·N1062-0.7000—
10L31·N10759+0.6948—

mask_delta top 10 · blind: |M−1|, size of the training edit

#neuronmask_deltaMauto-interp description
1L12·N12693+7.6412—
2L12·N12027+7.6241—
3L12·N10147+7.5031—
4L12·N4875+7.2526—
5L12·N1481+7.1583—
6L12·N942+7.0932—
7L12·N7522+7.0719—
8L12·N10061+6.9360—
9L12·N8787+6.9142—
10L12·N7534+6.7260—

random top 10 · seeded control

#neuronrandomMauto-interp description
1L2·N8086+4.7320—
2L11·N12244-4.5940—
3L31·N11190-4.5677—
4L1·N7319-4.4941—
5L8·N5583-4.4102—
6L30·N7193+4.3964—
7L25·N3291-4.3797—
8L19·N2704+4.3238—
9L28·N8662-4.3009—
10L7·N9552+4.2673—

Amplified 0 shown · 0 total

none

Downweighted 0 shown · 0 total

none

Deleted 0 shown · 0 total

none

Sign-flipped 0 shown · 0 total

none

Unchanged 0 shown · 0 total

none

Neuron cards 61 unique neurons, each rendered once

L0·N12387M=+1.000 (unchanged)frac_act 0.27 · mean_act -0.000
attr_last+0.0000attr_sum-0.1533attr_max-0.1528Cplus-0.0000Cminus-0.0000empos_contrast+0.0000mask_delta+0.0000random+1.6645
General corpus (C4 web text)

no data for this corpus

MCQ probe prompts

no data for this corpus

L1·N7319M=+1.000 (unchanged)frac_act 0.25 · mean_act -0.002
attr_last+0.0002attr_sum+0.0004attr_max+0.0004Cplus-0.0000Cminus-0.0002empos_contrast-0.0005mask_delta+0.0000random-4.4941
General corpus (C4 web text)

no data for this corpus

MCQ probe prompts

no data for this corpus

L2·N4786M=+1.000 (unchanged)frac_act 0.00 · mean_act -0.179
attr_last+0.0079attr_sum+0.4398attr_max+0.1001Cplus+0.0000Cminus-0.0079empos_contrast+0.0041mask_delta+0.0000random-0.5616
General corpus (C4 web text)

no data for this corpus

MCQ probe prompts

no data for this corpus

L2·N8086M=+1.000 (unchanged)frac_act 0.64 · mean_act +0.001
attr_last-0.0000attr_sum-0.0004attr_max-0.0001Cplus-0.0000Cminus-0.0000empos_contrast+0.0006mask_delta+0.0000random+4.7320
General corpus (C4 web text)

no data for this corpus

MCQ probe prompts

no data for this corpus

L3·N6390M=+1.000 (unchanged)frac_act 0.00 · mean_act -0.034
attr_last-0.0053attr_sum+0.3154attr_max+0.0569Cplus+0.0000Cminus+0.0053empos_contrast-0.0024mask_delta+0.0000random-0.0824
General corpus (C4 web text)

no data for this corpus

MCQ probe prompts

no data for this corpus

L5·N7012M=+1.000 (unchanged)frac_act 0.00 · mean_act -0.413
attr_last-0.0436attr_sum+0.2665attr_max-0.0358Cplus+0.0000Cminus+0.0436empos_contrast-0.0087mask_delta+0.0000random+1.8313
General corpus (C4 web text)

no data for this corpus

MCQ probe prompts

no data for this corpus

L6·N5866M=+1.000 (unchanged)frac_act 0.00 · mean_act -0.431
attr_last-0.0703attr_sum-0.2101attr_max-0.1759Cplus+0.0000Cminus+0.0703empos_contrast-0.0002mask_delta+0.0000random-0.8773
General corpus (C4 web text)

no data for this corpus

MCQ probe prompts

no data for this corpus

L7·N6673M=+1.000 (unchanged)frac_act 0.00 · mean_act -0.395
attr_last-0.0729attr_sum+0.2146attr_max-0.0544Cplus+0.0000Cminus+0.0729empos_contrast+0.0001mask_delta+0.0000random-1.0055
General corpus (C4 web text)

no data for this corpus

MCQ probe prompts

no data for this corpus

L7·N9552M=+1.000 (unchanged)frac_act 1.00 · mean_act +0.017
attr_last+0.0004attr_sum+0.0027attr_max+0.0015Cplus+0.0004Cminus+0.0000empos_contrast+0.0036mask_delta+0.0000random+4.2673
General corpus (C4 web text)

no data for this corpus

MCQ probe prompts

no data for this corpus

L8·N5583M=+1.000 (unchanged)frac_act 0.03 · mean_act -0.022
attr_last+0.0004attr_sum-0.0028attr_max-0.0029Cplus+0.0000Cminus-0.0004empos_contrast-0.0105mask_delta+0.0000random-4.4102
General corpus (C4 web text)

no data for this corpus

MCQ probe prompts

no data for this corpus

L11·N11321M=+1.000 (unchanged)frac_act 1.00 · mean_act +0.649
attr_last-0.1297attr_sum+0.1452attr_max+0.0718Cplus-0.1297Cminus+0.0000empos_contrast-0.0222mask_delta+0.0000random-1.6458
General corpus (C4 web text)

no data for this corpus

MCQ probe prompts

no data for this corpus

L11·N12244M=+1.000 (unchanged)frac_act 0.91 · mean_act +0.028
attr_last+0.0007attr_sum+0.0052attr_max+0.0016Cplus+0.0008Cminus+0.0000empos_contrast+0.0091mask_delta+0.0000random-4.5940
General corpus (C4 web text)

no data for this corpus

MCQ probe prompts

no data for this corpus

L12·N942M=-6.093 (flipped)frac_act 0.28 · mean_act -0.031
attr_last-0.0016attr_sum-0.0021attr_max-0.0021Cplus-0.0002Cminus+0.0014empos_contrast-0.0089mask_delta+7.0932random-0.7556
General corpus (C4 web text)

no data for this corpus

MCQ probe prompts

no data for this corpus

L12·N1062M=+1.987 (amplified)frac_act 1.00 · mean_act +1.521
attr_last-0.1631attr_sum-0.3342attr_max-0.1859Cplus-0.1631Cminus+0.0000empos_contrast-0.7000mask_delta+0.9873random+0.2722
General corpus (C4 web text)

no data for this corpus

MCQ probe prompts

no data for this corpus

L12·N1481M=-6.158 (flipped)frac_act 0.31 · mean_act -0.072
attr_last-0.0027attr_sum-0.0479attr_max-0.0318Cplus-0.0000Cminus+0.0027empos_contrast-0.0481mask_delta+7.1583random-1.1365
General corpus (C4 web text)

no data for this corpus

MCQ probe prompts

no data for this corpus

L12·N4875M=+8.253 (amplified)frac_act 0.91 · mean_act +0.257
attr_last+0.0081attr_sum+0.0165attr_max+0.0074Cplus+0.0082Cminus+0.0001empos_contrast+0.1218mask_delta+7.2526random-0.3533
General corpus (C4 web text)

no data for this corpus

MCQ probe prompts

no data for this corpus

L12·N7522M=-6.072 (flipped)frac_act 0.49 · mean_act +0.006
attr_last-0.0089attr_sum-0.0129attr_max-0.0120Cplus-0.0005Cminus+0.0085empos_contrast+0.0190mask_delta+7.0719random-0.5500
General corpus (C4 web text)

no data for this corpus

MCQ probe prompts

no data for this corpus

L12·N7534M=-5.726 (flipped)frac_act 0.32 · mean_act -0.034
attr_last+0.0013attr_sum-0.0000attr_max-0.0005Cplus+0.0008Cminus-0.0005empos_contrast-0.0097mask_delta+6.7260random+0.1654
General corpus (C4 web text)

no data for this corpus

MCQ probe prompts

no data for this corpus

L12·N7640M=+5.444 (amplified)frac_act 0.00 · mean_act -0.921
attr_last-0.0501attr_sum-0.0344attr_max-0.0457Cplus+0.0000Cminus+0.0501empos_contrast+0.0468mask_delta+4.4441random-0.6426
General corpus (C4 web text)

no data for this corpus

MCQ probe prompts

no data for this corpus

L12·N8787M=-5.914 (flipped)frac_act 1.00 · mean_act +0.215
attr_last-0.0103attr_sum-0.0128attr_max-0.0103Cplus-0.0103Cminus+0.0000empos_contrast-0.0542mask_delta+6.9142random+1.1071
General corpus (C4 web text)

no data for this corpus

MCQ probe prompts

no data for this corpus

L12·N8856M=+1.000 (unchanged)frac_act 0.98 · mean_act +0.210
attr_last+0.0459attr_sum+0.0563attr_max+0.0458Cplus+0.0459Cminus+0.0000empos_contrast+0.0790mask_delta+0.0001random+1.0835
General corpus (C4 web text)

no data for this corpus

MCQ probe prompts

no data for this corpus

L12·N9114M=+6.025 (amplified)frac_act 0.11 · mean_act -0.196
attr_last+0.0093attr_sum+0.6111attr_max+0.3597Cplus-0.0002Cminus-0.0095empos_contrast-0.0310mask_delta+5.0248random-1.0165
General corpus (C4 web text)

no data for this corpus

MCQ probe prompts

no data for this corpus

L12·N10061M=-5.936 (flipped)frac_act 0.42 · mean_act -0.003
attr_last+0.0050attr_sum+0.0006attr_max+0.0028Cplus-0.0011Cminus-0.0061empos_contrast-0.0070mask_delta+6.9360random-1.3684
General corpus (C4 web text)

no data for this corpus

MCQ probe prompts

no data for this corpus

L12·N10147M=-6.503 (flipped)frac_act 0.86 · mean_act +0.168
attr_last+0.0054attr_sum+0.0018attr_max+0.0013Cplus+0.0054Cminus-0.0000empos_contrast+0.0903mask_delta+7.5031random+0.3158
General corpus (C4 web text)

no data for this corpus

MCQ probe prompts

no data for this corpus

L12·N12027M=+8.624 (amplified)frac_act 0.35 · mean_act -0.011
attr_last-0.0003attr_sum-0.0040attr_max-0.0020Cplus+0.0006Cminus+0.0009empos_contrast+0.0374mask_delta+7.6241random-0.6433
General corpus (C4 web text)

no data for this corpus

MCQ probe prompts

no data for this corpus

L12·N12693M=+8.641 (amplified)frac_act 0.59 · mean_act +0.026
attr_last-0.0161attr_sum-0.0103attr_max+0.0058Cplus-0.0109Cminus+0.0052empos_contrast+0.0237mask_delta+7.6412random-1.1563
General corpus (C4 web text)

no data for this corpus

MCQ probe prompts

no data for this corpus

L12·N13275M=+2.001 (amplified)frac_act 0.00 · mean_act -0.311
attr_last-0.0494attr_sum-0.0652attr_max-0.0483Cplus+0.0000Cminus+0.0494empos_contrast-0.0210mask_delta+1.0014random-1.1093
General corpus (C4 web text)

no data for this corpus

MCQ probe prompts

no data for this corpus

L13·N4208M=+1.000 (unchanged)frac_act 0.00 · mean_act -0.850
attr_last+0.0370attr_sum-0.2242attr_max-0.0155Cplus+0.0000Cminus-0.0370empos_contrast+0.0103mask_delta+0.0000random-0.4598
General corpus (C4 web text)

no data for this corpus

MCQ probe prompts

no data for this corpus

L13·N6149M=+1.000 (unchanged)frac_act 0.99 · mean_act +0.218
attr_last+0.0464attr_sum+0.0490attr_max+0.0469Cplus+0.0464Cminus-0.0000empos_contrast+0.0594mask_delta+0.0000random-1.0560
General corpus (C4 web text)

no data for this corpus

MCQ probe prompts

no data for this corpus

L19·N2704M=+1.000 (unchanged)frac_act 0.71 · mean_act +0.003
attr_last-0.0000attr_sum+0.0004attr_max+0.0000Cplus-0.0000Cminus-0.0000empos_contrast-0.0030mask_delta+0.0000random+4.3238
General corpus (C4 web text)

no data for this corpus

MCQ probe prompts

no data for this corpus

L19·N8718M=+1.000 (unchanged)frac_act 0.72 · mean_act +0.143
attr_last+0.0551attr_sum+0.0571attr_max+0.0552Cplus+0.0434Cminus-0.0117empos_contrast+0.0887mask_delta+0.0000random+0.1122
General corpus (C4 web text)

no data for this corpus

MCQ probe prompts

no data for this corpus

L23·N860M=+1.000 (unchanged)frac_act 1.00 · mean_act +3.790
attr_last+0.0496attr_sum+0.0626attr_max+0.0489Cplus+0.0496Cminus+0.0000empos_contrast+0.3760mask_delta+0.0000random-0.3532
General corpus (C4 web text)

no data for this corpus

MCQ probe prompts

no data for this corpus

L25·N3291M=+1.000 (unchanged)frac_act 0.03 · mean_act -0.035
attr_last+0.0006attr_sum+0.0004attr_max+0.0005Cplus+0.0000Cminus-0.0006empos_contrast-0.0060mask_delta+0.0000random-4.3797
General corpus (C4 web text)

no data for this corpus

MCQ probe prompts

no data for this corpus

L26·N3526M=+1.000 (unchanged)frac_act 0.75 · mean_act +0.681
attr_last+0.2469attr_sum+0.2670attr_max+0.2497Cplus+0.2006Cminus-0.0463empos_contrast+0.4158mask_delta+0.0000random+0.4949
General corpus (C4 web text)

no data for this corpus

MCQ probe prompts

no data for this corpus

L26·N10664M=+1.000 (unchanged)frac_act 0.00 · mean_act -1.035
attr_last+0.1453attr_sum+0.2306attr_max+0.1450Cplus+0.0000Cminus-0.1453empos_contrast-0.2426mask_delta+0.0000random-1.2266
General corpus (C4 web text)

no data for this corpus

MCQ probe prompts

no data for this corpus

L27·N6317M=+1.000 (unchanged)frac_act 0.98 · mean_act +1.103
attr_last+0.0664attr_sum+0.0837attr_max+0.0666Cplus+0.0662Cminus-0.0002empos_contrast+0.1628mask_delta+0.0000random-1.3987
General corpus (C4 web text)

no data for this corpus

MCQ probe prompts

no data for this corpus

L28·N8662M=+1.000 (unchanged)frac_act 0.30 · mean_act -0.005
attr_last+0.0001attr_sum+0.0003attr_max+0.0002Cplus+0.0000Cminus-0.0001empos_contrast-0.0048mask_delta+0.0000random-4.3009
General corpus (C4 web text)

no data for this corpus

MCQ probe prompts

no data for this corpus

L29·N1141M=+1.000 (unchanged)frac_act 0.99 · mean_act +1.869
attr_last+0.0519attr_sum+0.0692attr_max+0.0517Cplus+0.0514Cminus-0.0005empos_contrast+0.0376mask_delta+0.0000random-0.5724
General corpus (C4 web text)

no data for this corpus

MCQ probe prompts

no data for this corpus

L30·N777M=+1.000 (unchanged)frac_act 0.00 · mean_act -2.832
attr_last+0.1584attr_sum+0.1590attr_max+0.1584Cplus+0.0000Cminus-0.1584empos_contrast-0.5365mask_delta+0.0000random-0.0667
General corpus (C4 web text)

no data for this corpus

MCQ probe prompts

no data for this corpus

L30·N3382M=+1.000 (unchanged)frac_act 1.00 · mean_act +7.198
attr_last+0.1877attr_sum+0.1617attr_max+0.1877Cplus+0.1877Cminus+0.0000empos_contrast+0.1504mask_delta+0.0000random+0.9214
General corpus (C4 web text)

no data for this corpus

MCQ probe prompts

no data for this corpus

L30·N3768M=+1.000 (unchanged)frac_act 0.00 · mean_act -2.731
attr_last-0.0642attr_sum-0.0640attr_max-0.0642Cplus+0.0000Cminus+0.0642empos_contrast+0.1444mask_delta+0.0000random-0.2966
General corpus (C4 web text)

no data for this corpus

MCQ probe prompts

no data for this corpus

L30·N4803M=+1.000 (unchanged)frac_act 0.04 · mean_act -3.636
attr_last-0.2198attr_sum-0.2197attr_max-0.2198Cplus-0.0016Cminus+0.2182empos_contrast-0.3016mask_delta+0.0000random-1.5724
General corpus (C4 web text)

no data for this corpus

MCQ probe prompts

no data for this corpus

L30·N7193M=+1.000 (unchanged)frac_act 0.75 · mean_act +0.018
attr_last+0.0001attr_sum+0.0001attr_max+0.0001Cplus+0.0000Cminus-0.0001empos_contrast-0.0041mask_delta+0.0000random+4.3964
General corpus (C4 web text)

no data for this corpus

MCQ probe prompts

no data for this corpus

L30·N8560M=+1.000 (unchanged)frac_act 1.00 · mean_act +2.992
attr_last-0.1696attr_sum-0.1694attr_max-0.1696Cplus-0.1696Cminus+0.0000empos_contrast+0.1431mask_delta+0.0000random+0.7235
General corpus (C4 web text)

no data for this corpus

MCQ probe prompts

no data for this corpus

L30·N13562M=+1.000 (unchanged)frac_act 0.95 · mean_act +0.634
attr_last+0.0679attr_sum+0.0682attr_max+0.0679Cplus+0.0643Cminus-0.0036empos_contrast+0.1050mask_delta+0.0000random-0.1736
General corpus (C4 web text)

no data for this corpus

MCQ probe prompts

no data for this corpus

L31·N1190M=+1.000 (unchanged)frac_act 0.81 · mean_act +0.931
attr_last-0.0002attr_sum-0.0002attr_max-0.0002Cplus-0.0006Cminus-0.0004empos_contrast-0.9065mask_delta+0.0000random-0.6011
General corpus (C4 web text)

no data for this corpus

MCQ probe prompts

no data for this corpus

L31·N1209M=+1.000 (unchanged)frac_act 0.35 · mean_act -1.091
attr_last+0.0068attr_sum+0.0068attr_max+0.0068Cplus-0.0008Cminus-0.0076empos_contrast+1.0555mask_delta+0.0000random-1.4248
General corpus (C4 web text)

no data for this corpus

MCQ probe prompts

no data for this corpus

L31·N2939M=+1.000 (unchanged)frac_act 0.72 · mean_act -0.042
attr_last+0.0010attr_sum+0.0010attr_max+0.0010Cplus-0.0002Cminus-0.0012empos_contrast+0.7536mask_delta+0.0000random+0.8980
General corpus (C4 web text)

no data for this corpus

MCQ probe prompts

no data for this corpus

L31·N5540M=+1.000 (unchanged)frac_act 0.73 · mean_act +0.546
attr_last+0.0029attr_sum+0.0029attr_max+0.0029Cplus+0.0026Cminus-0.0003empos_contrast+0.7035mask_delta+0.0000random+0.5838
General corpus (C4 web text)

no data for this corpus

MCQ probe prompts

no data for this corpus

L31·N5553M=+1.000 (unchanged)frac_act 0.06 · mean_act -2.947
attr_last-0.0076attr_sum-0.0076attr_max-0.0076Cplus-0.0017Cminus+0.0059empos_contrast+0.7921mask_delta+0.0000random-0.1163
General corpus (C4 web text)

no data for this corpus

MCQ probe prompts

no data for this corpus

L31·N5835M=+1.000 (unchanged)frac_act 0.32 · mean_act -0.224
attr_last-0.0782attr_sum-0.0782attr_max-0.0782Cplus-0.0157Cminus+0.0624empos_contrast-0.0780mask_delta+0.0000random-0.9731
General corpus (C4 web text)

no data for this corpus

MCQ probe prompts

no data for this corpus

L31·N8118M=+1.000 (unchanged)frac_act 0.99 · mean_act +6.254
attr_last-0.0033attr_sum-0.0033attr_max-0.0033Cplus-0.0033Cminus+0.0000empos_contrast+1.1226mask_delta+0.0000random-1.9231
General corpus (C4 web text)

no data for this corpus

MCQ probe prompts

no data for this corpus

L31·N8120M=+1.000 (unchanged)frac_act 0.01 · mean_act -0.879
attr_last-0.0996attr_sum-0.0996attr_max-0.0996Cplus-0.0001Cminus+0.0995empos_contrast-0.0431mask_delta+0.0000random+0.3712
General corpus (C4 web text)

no data for this corpus

MCQ probe prompts

no data for this corpus

L31·N9486M=+1.000 (unchanged)frac_act 0.92 · mean_act +1.271
attr_last-0.1234attr_sum-0.1234attr_max-0.1234Cplus-0.1076Cminus+0.0158empos_contrast+0.0389mask_delta+0.0000random+0.2658
General corpus (C4 web text)

no data for this corpus

MCQ probe prompts

no data for this corpus

L31·N9507M=+1.000 (unchanged)frac_act 0.17 · mean_act -0.811
attr_last-0.0002attr_sum-0.0002attr_max-0.0002Cplus-0.0000Cminus+0.0002empos_contrast-0.8168mask_delta+0.0000random+0.5812
General corpus (C4 web text)

no data for this corpus

MCQ probe prompts

no data for this corpus

L31·N10759M=+1.000 (unchanged)frac_act 0.37 · mean_act -0.216
attr_last-0.0017attr_sum-0.0017attr_max-0.0017Cplus-0.0000Cminus+0.0017empos_contrast+0.6948mask_delta+0.0000random-0.2708
General corpus (C4 web text)

no data for this corpus

MCQ probe prompts

no data for this corpus

L31·N11190M=+1.000 (unchanged)frac_act 0.14 · mean_act -0.042
attr_last+0.0004attr_sum+0.0004attr_max+0.0004Cplus+0.0000Cminus-0.0004empos_contrast+0.0089mask_delta+0.0000random-4.5677
General corpus (C4 web text)

no data for this corpus

MCQ probe prompts

no data for this corpus

L31·N12721M=+1.000 (unchanged)frac_act 1.00 · mean_act +1.512
attr_last+0.0630attr_sum+0.0630attr_max+0.0630Cplus+0.0630Cminus+0.0000empos_contrast+0.2467mask_delta+0.0000random+1.5212
General corpus (C4 web text)

no data for this corpus

MCQ probe prompts

no data for this corpus

L31·N12862M=+1.000 (unchanged)frac_act 0.00 · mean_act -0.724
attr_last-0.0438attr_sum-0.0438attr_max-0.0438Cplus+0.0000Cminus+0.0438empos_contrast+0.0717mask_delta+0.0000random-0.5280
General corpus (C4 web text)

no data for this corpus

MCQ probe prompts

no data for this corpus

L31·N13006M=+1.000 (unchanged)frac_act 0.00 · mean_act -3.778
attr_last-0.7525attr_sum-0.7525attr_max-0.7525Cplus+0.0000Cminus+0.7525empos_contrast-0.3220mask_delta+0.0000random+0.7071
General corpus (C4 web text)

no data for this corpus

MCQ probe prompts

no data for this corpus

L31·N13336M=+1.000 (unchanged)frac_act 0.18 · mean_act -3.477
attr_last-0.0008attr_sum-0.0008attr_max-0.0008Cplus-0.0007Cminus+0.0000empos_contrast+0.8920mask_delta+0.0000random+0.4331
General corpus (C4 web text)

no data for this corpus

MCQ probe prompts

no data for this corpus