InterpUpdates› MCQ attribution + mask inspection

MCQ single-token attribution + mask inspection

Config mask_L13_lam3e-5 · EM 0.17% (terra) · checkpoint mask_L13_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 398 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 (L13)also in untrained base top-20
attr_lastL12, L19, L24, L26, L30, L310/105/10
attr_sumL5, L12, L17, L19, L24, L30, L310/103/10
attr_maxL12, L13, L17, L19, L24, L26, L30, L311/103/10
top_by_CplusL12, L15, L19, L24, L30, L310/101/10
top_by_CminusL20, L27, L30, L310/100/10
empos_contrastL30, L310/101/10
mask_deltaL1310/100/10
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+1.0155—
2L31·N9486-0.7435—
3L30·N8560+0.4691—
4L31·N5381+0.1884—
5L24·N10293+0.1689—
6L19·N12722+0.1574—
7L26·N3526+0.1528—
8L12·N1062-0.1384—
9L19·N7172+0.1255—
10L31·N12862+0.1240—

attr_sum top 10 · grad×act summed over positions

#neuronattr_sumMauto-interp description
1L31·N13006+1.0155—
2L31·N9486-0.7435—
3L30·N8560+0.4702—
4L17·N2206-0.2981—
5L12·N1062-0.2170—
6L24·N10293+0.2099—
7L31·N5381+0.1884—
8L19·N12722+0.1765—
9L24·N12457+0.1742—
10L5·N7012+0.1687—

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

#neuronattr_maxMauto-interp description
1L31·N13006+1.0155—
2L31·N9486-0.7435—
3L30·N8560+0.4691—
4L17·N2206-0.2819—
5L31·N5381+0.1884—
6L24·N10293+0.1683—
7L19·N12722+0.1574—
8L13·N2316+0.1557—
9L26·N3526+0.1538—
10L12·N1062-0.1438—

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

#neuronCplusMauto-interp description
1L30·N8560+0.4691—
2L24·N10293+0.1689—
3L19·N12722+0.1574—
4L19·N7172+0.1255—
5L31·N8453+0.1105—
6L30·N3382+0.0805—
7L24·N5699+0.0693—
8L15·N8286+0.0646—
9L31·N2108+0.0563—
10L12·N13860+0.0527—

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

#neuronCminusMauto-interp description
1L30·N777+0.1097—
2L31·N5794+0.1071—
3L31·N9464+0.0924—
4L31·N5811+0.0908—
5L31·N9995+0.0782—
6L31·N10989+0.0779—
7L27·N3082+0.0724—
8L31·N5553+0.0651—
9L31·N5488+0.0646—
10L20·N13028+0.0551—

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

#neuronempos_contrastMauto-interp description
1L31·N2398+0.7725—
2L31·N2939+0.6538—
3L31·N11013-0.6434—
4L31·N1209+0.6247—
5L31·N9507-0.6194—
6L31·N2564-0.6076—
7L31·N13728-0.5882—
8L31·N12222+0.5862—
9L30·N4803-0.5596—
10L31·N13336+0.5533—

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

#neuronmask_deltaMauto-interp description
1L13·N10690+9.4794—
2L13·N4221+9.2348—
3L13·N8574+9.2122—
4L13·N1581+8.0977—
5L13·N10000+7.6632—
6L13·N8238+7.5486—
7L13·N6745+7.5362—
8L13·N3693+7.4149—
9L13·N13837+7.2914—
10L13·N12281+7.2847—

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 60 unique neurons, each rendered once

L1·N7319M=+1.000 (unchanged)frac_act 0.26 · mean_act -0.002
attr_last-0.0000attr_sum+0.0001attr_max+0.0001Cplus-0.0000Cminus+0.0000empos_contrast-0.0003mask_delta+0.0000random-4.4941
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 0.50 · mean_act +0.000
attr_last-0.0000attr_sum-0.0007attr_max-0.0008Cplus-0.0000Cminus+0.0000empos_contrast-0.0003mask_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 -0.406
attr_last+0.0213attr_sum+0.1687attr_max+0.0595Cplus+0.0000Cminus-0.0213empos_contrast+0.0007mask_delta+0.0000random+1.8313
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 0.99 · mean_act +0.016
attr_last-0.0001attr_sum+0.0007attr_max-0.0002Cplus-0.0001Cminus+0.0000empos_contrast+0.0011mask_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 0.11 · mean_act -0.018
attr_last-0.0001attr_sum-0.0020attr_max-0.0024Cplus-0.0000Cminus+0.0001empos_contrast-0.0011mask_delta+0.0000random-4.4102
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 +0.023
attr_last-0.0001attr_sum-0.0061attr_max-0.0061Cplus-0.0001Cminus+0.0000empos_contrast+0.0013mask_delta+0.0000random-4.5940
General corpus (C4 web text)

no data for this corpus

MCQ probe prompts

no data for this corpus

L12·N1062M=+1.000 (unchanged)frac_act 1.00 · mean_act +0.927
attr_last-0.1384attr_sum-0.2170attr_max-0.1438Cplus-0.1384Cminus+0.0000empos_contrast-0.0656mask_delta+0.0000random+0.2722
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 +0.668
attr_last+0.0527attr_sum-0.0810attr_max+0.0188Cplus+0.0527Cminus+0.0000empos_contrast+0.0102mask_delta+0.0000random+0.1428
General corpus (C4 web text)

no data for this corpus

MCQ probe prompts

no data for this corpus

L13·N1581M=+9.098 (amplified)frac_act 0.66 · mean_act +0.080
attr_last-0.0004attr_sum+0.0030attr_max+0.0012Cplus-0.0016Cminus-0.0012empos_contrast+0.0823mask_delta+8.0977random-0.9596
General corpus (C4 web text)

no data for this corpus

MCQ probe prompts

no data for this corpus

L13·N2316M=-1.992 (flipped)frac_act 0.23 · mean_act -0.028
attr_last+0.0004attr_sum+0.1663attr_max+0.1557Cplus-0.0002Cminus-0.0006empos_contrast-0.0007mask_delta+2.9923random+0.7632
General corpus (C4 web text)

no data for this corpus

MCQ probe prompts

no data for this corpus

L13·N3693M=+8.415 (amplified)frac_act 0.61 · mean_act +0.013
attr_last+0.0002attr_sum+0.0056attr_max+0.0061Cplus+0.0046Cminus+0.0044empos_contrast-0.0719mask_delta+7.4149random-1.0630
General corpus (C4 web text)

no data for this corpus

MCQ probe prompts

no data for this corpus

L13·N4221M=-8.235 (flipped)frac_act 0.17 · mean_act -0.146
attr_last-0.0058attr_sum-0.0063attr_max-0.0063Cplus+0.0001Cminus+0.0059empos_contrast+0.0093mask_delta+9.2348random+0.4481
General corpus (C4 web text)

no data for this corpus

MCQ probe prompts

no data for this corpus

L13·N6745M=-6.536 (flipped)frac_act 0.02 · mean_act -0.175
attr_last-0.0011attr_sum+0.0124attr_max+0.0039Cplus+0.0000Cminus+0.0011empos_contrast-0.0268mask_delta+7.5362random+0.5204
General corpus (C4 web text)

no data for this corpus

MCQ probe prompts

no data for this corpus

L13·N8238M=-6.549 (flipped)frac_act 0.94 · mean_act +0.166
attr_last-0.0014attr_sum-0.0045attr_max-0.0036Cplus-0.0013Cminus+0.0001empos_contrast-0.0584mask_delta+7.5486random-0.0916
General corpus (C4 web text)

no data for this corpus

MCQ probe prompts

no data for this corpus

L13·N8574M=-8.212 (flipped)frac_act 0.93 · mean_act +0.211
attr_last-0.0025attr_sum+0.0015attr_max+0.0010Cplus-0.0025Cminus-0.0000empos_contrast+0.0117mask_delta+9.2122random-0.2727
General corpus (C4 web text)

no data for this corpus

MCQ probe prompts

no data for this corpus

L13·N10000M=-6.663 (flipped)frac_act 0.92 · mean_act +0.073
attr_last+0.0004attr_sum+0.0071attr_max+0.0033Cplus+0.0005Cminus+0.0000empos_contrast+0.0173mask_delta+7.6632random-0.3671
General corpus (C4 web text)

no data for this corpus

MCQ probe prompts

no data for this corpus

L13·N10690M=-8.479 (flipped)frac_act 0.52 · mean_act +0.025
attr_last+0.0017attr_sum-0.0080attr_max-0.0054Cplus-0.0006Cminus-0.0023empos_contrast-0.0222mask_delta+9.4794random+0.6653
General corpus (C4 web text)

no data for this corpus

MCQ probe prompts

no data for this corpus

L13·N12281M=-6.285 (flipped)frac_act 0.99 · mean_act +0.210
attr_last+0.0183attr_sum+0.0277attr_max+0.0157Cplus+0.0184Cminus+0.0000empos_contrast+0.0158mask_delta+7.2847random+1.3415
General corpus (C4 web text)

no data for this corpus

MCQ probe prompts

no data for this corpus

L13·N13837M=-6.291 (flipped)frac_act 0.07 · mean_act -0.113
attr_last-0.0034attr_sum-0.0034attr_max-0.0036Cplus+0.0001Cminus+0.0035empos_contrast-0.0250mask_delta+7.2914random-0.3497
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 +1.027
attr_last+0.0646attr_sum+0.0581attr_max+0.0638Cplus+0.0646Cminus+0.0000empos_contrast+0.0506mask_delta+0.0000random-0.1297
General corpus (C4 web text)

no data for this corpus

MCQ probe prompts

no data for this corpus

L17·N2206M=+1.000 (unchanged)frac_act 1.00 · mean_act +0.052
attr_last-0.0011attr_sum-0.2981attr_max-0.2819Cplus-0.0011Cminus+0.0000empos_contrast+0.0033mask_delta+0.0000random-0.9732
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 0.85 · mean_act +0.008
attr_last-0.0000attr_sum-0.0013attr_max-0.0004Cplus-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

L19·N7172M=+1.000 (unchanged)frac_act 0.99 · mean_act +0.335
attr_last+0.1255attr_sum+0.1469attr_max+0.1253Cplus+0.1255Cminus-0.0001empos_contrast-0.0201mask_delta+0.0000random+0.7469
General corpus (C4 web text)

no data for this corpus

MCQ probe prompts

no data for this corpus

L19·N12722M=+1.000 (unchanged)frac_act 1.00 · mean_act +2.281
attr_last+0.1574attr_sum+0.1765attr_max+0.1574Cplus+0.1574Cminus+0.0000empos_contrast-0.0242mask_delta+0.0000random+0.6846
General corpus (C4 web text)

no data for this corpus

MCQ probe prompts

no data for this corpus

L20·N13028M=+1.000 (unchanged)frac_act 0.01 · mean_act -0.681
attr_last-0.0551attr_sum-0.0742attr_max-0.0553Cplus+0.0000Cminus+0.0551empos_contrast-0.1329mask_delta+0.0000random-1.3508
General corpus (C4 web text)

no data for this corpus

MCQ probe prompts

no data for this corpus

L24·N5699M=+1.000 (unchanged)frac_act 1.00 · mean_act +0.555
attr_last+0.0693attr_sum+0.0789attr_max+0.0692Cplus+0.0693Cminus+0.0000empos_contrast+0.0380mask_delta+0.0000random+0.3093
General corpus (C4 web text)

no data for this corpus

MCQ probe prompts

no data for this corpus

L24·N10293M=+1.000 (unchanged)frac_act 1.00 · mean_act +1.108
attr_last+0.1689attr_sum+0.2099attr_max+0.1683Cplus+0.1689Cminus+0.0000empos_contrast+0.0856mask_delta+0.0000random+0.5927
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 -1.503
attr_last+0.1180attr_sum+0.1742attr_max+0.1151Cplus+0.0000Cminus-0.1180empos_contrast-0.3130mask_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 0.03 · mean_act -0.032
attr_last-0.0002attr_sum-0.0003attr_max-0.0002Cplus+0.0000Cminus+0.0002empos_contrast-0.0058mask_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 0.40 · mean_act -0.082
attr_last+0.1528attr_sum+0.1497attr_max+0.1538Cplus+0.0331Cminus-0.1197empos_contrast+0.2380mask_delta+0.0000random+0.4949
General corpus (C4 web text)

no data for this corpus

MCQ probe prompts

no data for this corpus

L27·N3082M=+1.000 (unchanged)frac_act 0.00 · mean_act -0.478
attr_last-0.0724attr_sum-0.0980attr_max-0.0724Cplus+0.0000Cminus+0.0724empos_contrast-0.0119mask_delta+0.0000random+1.6826
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 0.39 · mean_act -0.004
attr_last+0.0000attr_sum+0.0003attr_max+0.0001Cplus+0.0001Cminus+0.0001empos_contrast+0.0013mask_delta+0.0000random-4.3009
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 -3.162
attr_last-0.1097attr_sum-0.1093attr_max-0.1097Cplus+0.0000Cminus+0.1097empos_contrast+0.3831mask_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 +7.082
attr_last+0.0805attr_sum+0.0513attr_max+0.0804Cplus+0.0805Cminus+0.0000empos_contrast-0.1029mask_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 0.02 · mean_act -2.554
attr_last+0.0799attr_sum+0.0811attr_max+0.0799Cplus-0.0006Cminus-0.0805empos_contrast-0.5596mask_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 0.71 · mean_act +0.007
attr_last+0.0002attr_sum+0.0003attr_max+0.0002Cplus+0.0002Cminus-0.0000empos_contrast-0.0060mask_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 +3.620
attr_last+0.4691attr_sum+0.4702attr_max+0.4691Cplus+0.4691Cminus+0.0000empos_contrast-0.2396mask_delta+0.0000random+0.7235
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 0.08 · mean_act -3.561
attr_last-0.0059attr_sum-0.0059attr_max-0.0059Cplus+0.0000Cminus+0.0059empos_contrast+0.6247mask_delta+0.0000random-1.4248
General corpus (C4 web text)

no data for this corpus

MCQ probe prompts

no data for this corpus

L31·N2108M=+1.000 (unchanged)frac_act 1.00 · mean_act +3.784
attr_last+0.0563attr_sum+0.0563attr_max+0.0563Cplus+0.0563Cminus+0.0000empos_contrast+0.1617mask_delta+0.0000random-0.7256
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 1.00 · mean_act +8.114
attr_last-0.0041attr_sum-0.0041attr_max-0.0041Cplus-0.0041Cminus+0.0000empos_contrast+0.7725mask_delta+0.0000random+0.3985
General corpus (C4 web text)

no data for this corpus

MCQ probe prompts

no data for this corpus

L31·N2564M=+1.000 (unchanged)frac_act 0.04 · mean_act -2.733
attr_last+0.0019attr_sum+0.0019attr_max+0.0019Cplus-0.0000Cminus-0.0020empos_contrast-0.6076mask_delta+0.0000random-1.2506
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 0.76 · mean_act +0.100
attr_last+0.0004attr_sum+0.0004attr_max+0.0004Cplus+0.0010Cminus+0.0006empos_contrast+0.6538mask_delta+0.0000random+0.8980
General corpus (C4 web text)

no data for this corpus

MCQ probe prompts

no data for this corpus

L31·N5381M=+1.000 (unchanged)frac_act 0.00 · mean_act -2.101
attr_last+0.1884attr_sum+0.1884attr_max+0.1884Cplus+0.0000Cminus-0.1884empos_contrast+0.2780mask_delta+0.0000random+0.6712
General corpus (C4 web text)

no data for this corpus

MCQ probe prompts

no data for this corpus

L31·N5488M=+1.000 (unchanged)frac_act 0.06 · mean_act -0.706
attr_last-0.0646attr_sum-0.0646attr_max-0.0646Cplus-0.0000Cminus+0.0646empos_contrast-0.1806mask_delta+0.0000random-2.6127
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 0.04 · mean_act -3.580
attr_last-0.0658attr_sum-0.0658attr_max-0.0658Cplus-0.0007Cminus+0.0651empos_contrast+0.1105mask_delta+0.0000random-0.1163
General corpus (C4 web text)

no data for this corpus

MCQ probe prompts

no data for this corpus

L31·N5794M=+1.000 (unchanged)frac_act 0.01 · mean_act -0.570
attr_last-0.1072attr_sum-0.1072attr_max-0.1072Cplus-0.0001Cminus+0.1071empos_contrast+0.0888mask_delta+0.0000random-0.2958
General corpus (C4 web text)

no data for this corpus

MCQ probe prompts

no data for this corpus

L31·N5811M=+1.000 (unchanged)frac_act 0.00 · mean_act -0.208
attr_last-0.0908attr_sum-0.0908attr_max-0.0908Cplus-0.0000Cminus+0.0908empos_contrast+0.0527mask_delta+0.0000random-0.3526
General corpus (C4 web text)

no data for this corpus

MCQ probe prompts

no data for this corpus

L31·N8453M=+1.000 (unchanged)frac_act 1.00 · mean_act +0.531
attr_last+0.1105attr_sum+0.1105attr_max+0.1105Cplus+0.1105Cminus+0.0000empos_contrast-0.0072mask_delta+0.0000random-1.1005
General corpus (C4 web text)

no data for this corpus

MCQ probe prompts

no data for this corpus

L31·N9464M=+1.000 (unchanged)frac_act 0.00 · mean_act -0.367
attr_last-0.0924attr_sum-0.0924attr_max-0.0924Cplus+0.0000Cminus+0.0924empos_contrast+0.0211mask_delta+0.0000random+1.6993
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 1.00 · mean_act +1.784
attr_last-0.7435attr_sum-0.7435attr_max-0.7435Cplus-0.7435Cminus+0.0000empos_contrast-0.2447mask_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 0.10 · mean_act -0.996
attr_last-0.0001attr_sum-0.0001attr_max-0.0001Cplus-0.0000Cminus+0.0001empos_contrast-0.6194mask_delta+0.0000random+0.5812
General corpus (C4 web text)

no data for this corpus

MCQ probe prompts

no data for this corpus

L31·N9995M=+1.000 (unchanged)frac_act 0.00 · mean_act -5.054
attr_last-0.0782attr_sum-0.0782attr_max-0.0782Cplus+0.0000Cminus+0.0782empos_contrast-0.3997mask_delta+0.0000random-1.5909
General corpus (C4 web text)

no data for this corpus

MCQ probe prompts

no data for this corpus

L31·N10989M=+1.000 (unchanged)frac_act 0.00 · mean_act -0.211
attr_last-0.0779attr_sum-0.0779attr_max-0.0779Cplus+0.0000Cminus+0.0779empos_contrast+0.0580mask_delta+0.0000random-0.1799
General corpus (C4 web text)

no data for this corpus

MCQ probe prompts

no data for this corpus

L31·N11013M=+1.000 (unchanged)frac_act 0.01 · mean_act -2.250
attr_last-0.0144attr_sum-0.0144attr_max-0.0144Cplus-0.0000Cminus+0.0144empos_contrast-0.6434mask_delta+0.0000random+0.0505
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 0.03 · mean_act -0.057
attr_last-0.0008attr_sum-0.0008attr_max-0.0008Cplus+0.0000Cminus+0.0008empos_contrast-0.0007mask_delta+0.0000random-4.5677
General corpus (C4 web text)

no data for this corpus

MCQ probe prompts

no data for this corpus

L31·N12222M=+1.000 (unchanged)frac_act 0.29 · mean_act -0.267
attr_last-0.0039attr_sum-0.0039attr_max-0.0039Cplus-0.0006Cminus+0.0033empos_contrast+0.5862mask_delta+0.0000random-1.1885
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 -0.591
attr_last+0.1240attr_sum+0.1240attr_max+0.1240Cplus+0.0000Cminus-0.1240empos_contrast-0.0094mask_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 -3.504
attr_last+1.0155attr_sum+1.0155attr_max+1.0155Cplus+0.0000Cminus-1.0155empos_contrast-0.2532mask_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 0.02 · mean_act -7.739
attr_last-0.0146attr_sum-0.0146attr_max-0.0146Cplus+0.0000Cminus+0.0146empos_contrast+0.5533mask_delta+0.0000random+0.4331
General corpus (C4 web text)

no data for this corpus

MCQ probe prompts

no data for this corpus

L31·N13728M=+1.000 (unchanged)frac_act 0.11 · mean_act -1.227
attr_last-0.0004attr_sum-0.0004attr_max-0.0004Cplus+0.0000Cminus+0.0004empos_contrast-0.5882mask_delta+0.0000random+0.8647
General corpus (C4 web text)

no data for this corpus

MCQ probe prompts

no data for this corpus