InterpUpdates› MCQ attribution + mask inspection

MCQ single-token attribution + mask inspection

Config mask_L10_lam3e-5 · EM 0.19% (terra) · checkpoint mask_L10_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 310 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 (L10)also in untrained base top-20
attr_lastL5, L13, L15, L26, L30, L310/107/10
attr_sumL2, L5, L10, L12, L13, L18, L30, L311/108/10
attr_maxL5, L12, L13, L26, L30, L310/105/10
top_by_CplusL12, L14, L26, L27, L29, L30, L310/106/10
top_by_CminusL5, L7, L13, L24, L30, L310/109/10
empos_contrastL310/108/10
mask_deltaL1010/100/10
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·N13006-1.2985—
2L30·N8560-0.4915—
3L30·N4803-0.3834—
4L30·N777+0.2507—
5L5·N7012-0.2217—
6L30·N3382+0.1914—
7L31·N9486+0.1899—
8L26·N3526+0.1555—
9L13·N4208-0.1456—
10L15·N8286-0.1338—

attr_sum top 10 · grad×act summed over positions

#neuronattr_sumMauto-interp description
1L31·N13006-1.2985—
2L30·N8560-0.4898—
3L30·N4803-0.3836—
4L13·N4208-0.3484—
5L30·N777+0.2515—
6L12·N13860+0.2359—
7L2·N4786-0.2283—
8L18·N7417-0.2280—
9L5·N7012-0.2162—
10L10·N11570+0.2083—

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

#neuronattr_maxMauto-interp description
1L31·N13006-1.2985—
2L30·N8560-0.4915—
3L30·N4803-0.3834—
4L30·N777+0.2507—
5L30·N3382+0.1912—
6L31·N9486+0.1899—
7L5·N7012-0.1766—
8L26·N3526+0.1543—
9L12·N13860+0.1498—
10L13·N4208-0.1491—

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

#neuronCplusMauto-interp description
1L30·N3382+0.1914—
2L31·N9486+0.1908—
3L26·N3526+0.1451—
4L12·N13860+0.1209—
5L31·N12721+0.1121—
6L29·N1141+0.1004—
7L14·N7120+0.1004—
8L30·N13562+0.0937—
9L27·N6317+0.0897—
10L14·N5680+0.0769—

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

#neuronCminusMauto-interp description
1L31·N13006+1.2985—
2L30·N4803+0.3834—
3L5·N7012+0.2217—
4L13·N4208+0.1456—
5L31·N8120+0.1179—
6L7·N6673+0.1118—
7L31·N12862+0.1095—
8L31·N5835+0.0837—
9L31·N3700+0.0834—
10L24·N12457+0.0791—

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

#neuronempos_contrastMauto-interp description
1L31·N2398+2.2008—
2L31·N13336-1.9438—
3L31·N1209+1.6842—
4L31·N12923+1.5661—
5L31·N5021-1.3018—
6L31·N12004-1.2871—
7L31·N5975-1.2653—
8L31·N8118+1.1966—
9L31·N3175+1.0815—
10L31·N486-1.0665—

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

#neuronmask_deltaMauto-interp description
1L10·N8674+8.6584—
2L10·N2759+8.0218—
3L10·N8363+7.8309—
4L10·N538+7.6700—
5L10·N10084+7.4091—
6L10·N12108+6.8429—
7L10·N14158+6.7015—
8L10·N7454+6.5802—
9L10·N4840+6.5786—
10L10·N52+6.5647—

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 56 unique neurons, each rendered once

L1·N7319M=+1.000 (unchanged)frac_act 0.23 · mean_act -0.002
attr_last+0.0000attr_sum-0.0000attr_max+0.0000Cplus-0.0000Cminus-0.0001empos_contrast-0.0008mask_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-0.0533attr_sum-0.2283attr_max-0.0175Cplus+0.0000Cminus+0.0533empos_contrast+0.0035mask_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+0.0013attr_max+0.0008Cplus+0.0000Cminus-0.0000empos_contrast+0.0004mask_delta+0.0000random+4.7320
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 -0.414
attr_last-0.2217attr_sum-0.2162attr_max-0.1766Cplus+0.0000Cminus+0.2217empos_contrast-0.0080mask_delta+0.0000random+1.8313
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-0.1118attr_sum+0.0681attr_max-0.0739Cplus+0.0000Cminus+0.1118empos_contrast+0.0018mask_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+0.0010attr_max+0.0002Cplus+0.0004Cminus+0.0000empos_contrast+0.0037mask_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 0.02 · mean_act -0.023
attr_last+0.0002attr_sum-0.0010attr_max+0.0023Cplus+0.0000Cminus-0.0002empos_contrast-0.0119mask_delta+0.0000random-4.4102
General corpus (C4 web text)

no data for this corpus

MCQ probe prompts

no data for this corpus

L10·N52M=-5.565 (flipped)frac_act 0.00 · mean_act -0.434
attr_last-0.0009attr_sum+0.0168attr_max+0.0026Cplus+0.0000Cminus+0.0009empos_contrast-0.0121mask_delta+6.5647random-0.4954
General corpus (C4 web text)

no data for this corpus

MCQ probe prompts

no data for this corpus

L10·N538M=-6.670 (flipped)frac_act 0.80 · mean_act +0.065
attr_last-0.0026attr_sum+0.0096attr_max+0.0061Cplus-0.0026Cminus-0.0000empos_contrast+0.0473mask_delta+7.6700random+1.2918
General corpus (C4 web text)

no data for this corpus

MCQ probe prompts

no data for this corpus

L10·N2759M=+9.022 (amplified)frac_act 0.00 · mean_act -0.581
attr_last+0.0115attr_sum+0.0024attr_max+0.0112Cplus+0.0000Cminus-0.0115empos_contrast-0.0933mask_delta+8.0218random-1.2698
General corpus (C4 web text)

no data for this corpus

MCQ probe prompts

no data for this corpus

L10·N4840M=+7.579 (amplified)frac_act 0.44 · mean_act -0.013
attr_last+0.0028attr_sum+0.0039attr_max-0.0008Cplus-0.0005Cminus-0.0033empos_contrast-0.0481mask_delta+6.5786random+1.4273
General corpus (C4 web text)

no data for this corpus

MCQ probe prompts

no data for this corpus

L10·N7454M=+7.580 (amplified)frac_act 0.75 · mean_act +0.057
attr_last+0.0055attr_sum-0.0049attr_max+0.0047Cplus+0.0062Cminus+0.0006empos_contrast+0.0667mask_delta+6.5802random+0.9788
General corpus (C4 web text)

no data for this corpus

MCQ probe prompts

no data for this corpus

L10·N8363M=+8.831 (amplified)frac_act 0.52 · mean_act -0.005
attr_last-0.0016attr_sum-0.0145attr_max-0.0003Cplus+0.0010Cminus+0.0025empos_contrast-0.1109mask_delta+7.8309random+1.9219
General corpus (C4 web text)

no data for this corpus

MCQ probe prompts

no data for this corpus

L10·N8674M=+9.658 (amplified)frac_act 0.14 · mean_act -0.025
attr_last+0.0004attr_sum-0.0019attr_max-0.0031Cplus+0.0000Cminus-0.0004empos_contrast-0.0201mask_delta+8.6584random-0.4793
General corpus (C4 web text)

no data for this corpus

MCQ probe prompts

no data for this corpus

L10·N10084M=+8.409 (amplified)frac_act 0.75 · mean_act +0.096
attr_last+0.0041attr_sum+0.0063attr_max+0.0095Cplus+0.0039Cminus-0.0003empos_contrast-0.0056mask_delta+7.4091random+2.3450
General corpus (C4 web text)

no data for this corpus

MCQ probe prompts

no data for this corpus

L10·N11570M=+1.094 (amplified)frac_act 0.00 · mean_act -0.541
attr_last+0.0109attr_sum+0.2083attr_max+0.0509Cplus+0.0000Cminus-0.0109empos_contrast+0.0246mask_delta+0.0943random+0.0518
General corpus (C4 web text)

no data for this corpus

MCQ probe prompts

no data for this corpus

L10·N12108M=+7.843 (amplified)frac_act 0.05 · mean_act -0.157
attr_last-0.0042attr_sum-0.0024attr_max+0.0119Cplus-0.0000Cminus+0.0041empos_contrast+0.0565mask_delta+6.8429random-0.2526
General corpus (C4 web text)

no data for this corpus

MCQ probe prompts

no data for this corpus

L10·N14158M=-5.702 (flipped)frac_act 0.54 · mean_act +0.003
attr_last+0.0002attr_sum+0.0099attr_max+0.0067Cplus-0.0007Cminus-0.0009empos_contrast-0.0150mask_delta+6.7015random+0.0625
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 0.74 · mean_act +0.008
attr_last-0.0007attr_sum+0.0111attr_max+0.0033Cplus-0.0007Cminus-0.0000empos_contrast+0.0073mask_delta+0.0000random-4.5940
General corpus (C4 web text)

no data for this corpus

MCQ probe prompts

no data for this corpus

L12·N13860M=+1.000 (unchanged)frac_act 1.00 · mean_act +0.555
attr_last+0.1209attr_sum+0.2359attr_max+0.1498Cplus+0.1209Cminus+0.0000empos_contrast+0.0231mask_delta+0.0000random+0.1428
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 -0.816
attr_last-0.1456attr_sum-0.3484attr_max-0.1491Cplus+0.0000Cminus+0.1456empos_contrast+0.0011mask_delta+0.0000random-0.4598
General corpus (C4 web text)

no data for this corpus

MCQ probe prompts

no data for this corpus

L14·N5680M=+1.000 (unchanged)frac_act 0.85 · mean_act +0.152
attr_last+0.0781attr_sum+0.1032attr_max+0.0781Cplus+0.0769Cminus-0.0012empos_contrast+0.1091mask_delta+0.0000random-1.1456
General corpus (C4 web text)

no data for this corpus

MCQ probe prompts

no data for this corpus

L14·N7120M=+1.000 (unchanged)frac_act 0.86 · mean_act +0.137
attr_last+0.1042attr_sum+0.1246attr_max+0.1031Cplus+0.1004Cminus-0.0038empos_contrast-0.1068mask_delta+0.0000random+1.2696
General corpus (C4 web text)

no data for this corpus

MCQ probe prompts

no data for this corpus

L15·N8286M=+1.000 (unchanged)frac_act 1.00 · mean_act +1.214
attr_last-0.1338attr_sum-0.1543attr_max-0.1338Cplus-0.1338Cminus+0.0000empos_contrast+0.0772mask_delta+0.0000random-0.1297
General corpus (C4 web text)

no data for this corpus

MCQ probe prompts

no data for this corpus

L18·N7417M=+1.000 (unchanged)frac_act 1.00 · mean_act +0.594
attr_last-0.0619attr_sum-0.2280attr_max-0.1058Cplus-0.0619Cminus+0.0000empos_contrast+0.0222mask_delta+0.0000random+0.9041
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 0.70 · mean_act +0.003
attr_last-0.0000attr_sum+0.0003attr_max+0.0002Cplus-0.0000Cminus-0.0000empos_contrast+0.0022mask_delta+0.0000random+4.3238
General corpus (C4 web text)

no data for this corpus

MCQ probe prompts

no data for this corpus

L24·N12457M=+1.000 (unchanged)frac_act 0.00 · mean_act -1.267
attr_last-0.0791attr_sum-0.1174attr_max-0.0794Cplus+0.0000Cminus+0.0791empos_contrast-0.1030mask_delta+0.0000random-0.1672
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 0.38 · mean_act -0.005
attr_last+0.0003attr_sum+0.0005attr_max+0.0003Cplus-0.0000Cminus-0.0003empos_contrast-0.0203mask_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 0.88 · mean_act +0.511
attr_last+0.1555attr_sum+0.1685attr_max+0.1543Cplus+0.1451Cminus-0.0103empos_contrast+0.2755mask_delta+0.0000random+0.4949
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 1.00 · mean_act +1.242
attr_last+0.0897attr_sum+0.1054attr_max+0.0896Cplus+0.0897Cminus+0.0000empos_contrast+0.0789mask_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 0.26 · mean_act -0.007
attr_last+0.0002attr_sum+0.0004attr_max+0.0003Cplus-0.0000Cminus-0.0002empos_contrast-0.0097mask_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 1.00 · mean_act +2.574
attr_last+0.1004attr_sum+0.1327attr_max+0.1005Cplus+0.1004Cminus+0.0000empos_contrast+0.0324mask_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 -2.236
attr_last+0.2507attr_sum+0.2515attr_max+0.2507Cplus+0.0000Cminus-0.2507empos_contrast-0.1699mask_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 +6.953
attr_last+0.1914attr_sum+0.1781attr_max+0.1912Cplus+0.1914Cminus+0.0000empos_contrast+0.2905mask_delta+0.0000random+0.9214
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 0.00 · mean_act -4.674
attr_last-0.3834attr_sum-0.3836attr_max-0.3834Cplus-0.0000Cminus+0.3834empos_contrast-0.3506mask_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 0.11 · mean_act -0.022
attr_last+0.0003attr_sum+0.0003attr_max+0.0003Cplus-0.0000Cminus-0.0003empos_contrast+0.0026mask_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 +3.284
attr_last-0.4915attr_sum-0.4898attr_max-0.4915Cplus-0.4915Cminus+0.0000empos_contrast+0.0003mask_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 0.99 · mean_act +0.710
attr_last+0.0939attr_sum+0.0934attr_max+0.0939Cplus+0.0937Cminus-0.0002empos_contrast+0.2282mask_delta+0.0000random-0.1736
General corpus (C4 web text)

no data for this corpus

MCQ probe prompts

no data for this corpus

L31·N486M=+1.000 (unchanged)frac_act 0.44 · mean_act -0.019
attr_last+0.0002attr_sum+0.0002attr_max+0.0002Cplus+0.0001Cminus-0.0001empos_contrast-1.0665mask_delta+0.0000random-0.1893
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 0.05 · mean_act -4.163
attr_last+0.0166attr_sum+0.0166attr_max+0.0166Cplus-0.0002Cminus-0.0168empos_contrast+1.6842mask_delta+0.0000random-1.4248
General corpus (C4 web text)

no data for this corpus

MCQ probe prompts

no data for this corpus

L31·N2398M=+1.000 (unchanged)frac_act 0.99 · mean_act +7.457
attr_last-0.0181attr_sum-0.0181attr_max-0.0181Cplus-0.0182Cminus-0.0000empos_contrast+2.2008mask_delta+0.0000random+0.3985
General corpus (C4 web text)

no data for this corpus

MCQ probe prompts

no data for this corpus

L31·N3175M=+1.000 (unchanged)frac_act 0.00 · mean_act -2.103
attr_last-0.0423attr_sum-0.0423attr_max-0.0423Cplus+0.0000Cminus+0.0423empos_contrast+1.0815mask_delta+0.0000random+0.7427
General corpus (C4 web text)

no data for this corpus

MCQ probe prompts

no data for this corpus

L31·N3700M=+1.000 (unchanged)frac_act 0.06 · mean_act -0.573
attr_last-0.0839attr_sum-0.0839attr_max-0.0839Cplus-0.0004Cminus+0.0834empos_contrast-0.1566mask_delta+0.0000random-0.9963
General corpus (C4 web text)

no data for this corpus

MCQ probe prompts

no data for this corpus

L31·N5021M=+1.000 (unchanged)frac_act 0.00 · mean_act -7.149
attr_last+0.0025attr_sum+0.0025attr_max+0.0025Cplus+0.0000Cminus-0.0025empos_contrast-1.3018mask_delta+0.0000random+0.5142
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 0.17 · mean_act -0.335
attr_last-0.0915attr_sum-0.0915attr_max-0.0915Cplus-0.0078Cminus+0.0837empos_contrast-0.1557mask_delta+0.0000random-0.9731
General corpus (C4 web text)

no data for this corpus

MCQ probe prompts

no data for this corpus

L31·N5975M=+1.000 (unchanged)frac_act 0.00 · mean_act -4.976
attr_last-0.0221attr_sum-0.0221attr_max-0.0221Cplus+0.0000Cminus+0.0221empos_contrast-1.2653mask_delta+0.0000random-1.4465
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 1.00 · mean_act +6.239
attr_last-0.0053attr_sum-0.0053attr_max-0.0053Cplus-0.0053Cminus+0.0000empos_contrast+1.1966mask_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 -0.779
attr_last-0.1180attr_sum-0.1180attr_max-0.1180Cplus-0.0001Cminus+0.1179empos_contrast-0.2184mask_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 0.98 · mean_act +1.119
attr_last+0.1899attr_sum+0.1899attr_max+0.1899Cplus+0.1908Cminus+0.0010empos_contrast+0.2423mask_delta+0.0000random+0.2658
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 0.29 · mean_act -0.016
attr_last+0.0002attr_sum+0.0002attr_max+0.0002Cplus-0.0001Cminus-0.0003empos_contrast+0.0047mask_delta+0.0000random-4.5677
General corpus (C4 web text)

no data for this corpus

MCQ probe prompts

no data for this corpus

L31·N12004M=+1.000 (unchanged)frac_act 0.04 · mean_act -2.718
attr_last-0.0072attr_sum-0.0072attr_max-0.0072Cplus+0.0000Cminus+0.0072empos_contrast-1.2871mask_delta+0.0000random-1.2191
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 +1.444
attr_last+0.1121attr_sum+0.1121attr_max+0.1121Cplus+0.1121Cminus+0.0000empos_contrast+0.3969mask_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 -0.642
attr_last-0.1095attr_sum-0.1095attr_max-0.1095Cplus+0.0000Cminus+0.1095empos_contrast+0.0415mask_delta+0.0000random-0.5280
General corpus (C4 web text)

no data for this corpus

MCQ probe prompts

no data for this corpus

L31·N12923M=+1.000 (unchanged)frac_act 0.47 · mean_act -0.058
attr_last-0.0014attr_sum-0.0014attr_max-0.0014Cplus-0.0028Cminus-0.0014empos_contrast+1.5661mask_delta+0.0000random-1.8301
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 -3.800
attr_last-1.2985attr_sum-1.2985attr_max-1.2985Cplus+0.0000Cminus+1.2985empos_contrast-0.8760mask_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 0.01 · mean_act -7.318
attr_last+0.0074attr_sum+0.0074attr_max+0.0074Cplus-0.0002Cminus-0.0076empos_contrast-1.9438mask_delta+0.0000random+0.4331
General corpus (C4 web text)

no data for this corpus

MCQ probe prompts

no data for this corpus