InterpUpdates› MCQ attribution + mask inspection

MCQ single-token attribution + mask inspection

Config mask_L5_lam3e-5 · EM 0.40% (terra) · checkpoint mask_L5_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 248 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 (L5)also in untrained base top-20
attr_lastL13, L16, L26, L30, L310/108/10
attr_sumL3, L5, L9, L10, L12, L20, L30, L312/106/10
attr_maxL1, L5, L16, L30, L311/107/10
top_by_CplusL11, L12, L26, L27, L29, L30, L310/108/10
top_by_CminusL2, L6, L20, L24, L30, L310/108/10
empos_contrastL310/104/10
mask_deltaL510/100/10
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·N13006-0.9848—
2L30·N4803-0.3917—
3L30·N8560-0.3671—
4L30·N777+0.1775—
5L31·N12721+0.1696—
6L30·N3382+0.1666—
7L16·N9215-0.1582—
8L31·N8120-0.1563—
9L13·N4208+0.1475—
10L26·N3526+0.1420—

attr_sum top 10 · grad×act summed over positions

#neuronattr_sumMauto-interp description
1L31·N13006-0.9848—
2L3·N6390+0.4577—
3L30·N4803-0.3930—
4L30·N8560-0.3667—
5L5·N171-0.3497—
6L10·N11570+0.3242—
7L5·N7012-0.3085—
8L9·N4255-0.2848—
9L20·N3972-0.2462—
10L12·N1062-0.2066—

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

#neuronattr_maxMauto-interp description
1L31·N13006-0.9848—
2L30·N4803-0.3917—
3L30·N8560-0.3671—
4L5·N7012-0.1793—
5L30·N777+0.1775—
6L31·N12721+0.1696—
7L30·N3382+0.1659—
8L1·N7179+0.1611—
9L16·N9215-0.1582—
10L31·N8120-0.1563—

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

#neuronCplusMauto-interp description
1L31·N12721+0.1696—
2L30·N3382+0.1666—
3L31·N9486+0.1049—
4L26·N3526+0.1031—
5L30·N13562+0.1013—
6L12·N8856+0.0800—
7L11·N10594+0.0769—
8L31·N11291+0.0754—
9L27·N6317+0.0745—
10L29·N1141+0.0691—

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

#neuronCminusMauto-interp description
1L31·N13006+0.9848—
2L30·N4803+0.3912—
3L31·N8120+0.1563—
4L31·N5381+0.1011—
5L31·N12862+0.0908—
6L2·N4786+0.0823—
7L6·N5866+0.0804—
8L31·N5835+0.0770—
9L24·N8167+0.0768—
10L20·N4299+0.0605—

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

#neuronempos_contrastMauto-interp description
1L31·N5021-1.2153—
2L31·N2398+1.2103—
3L31·N12923+1.0227—
4L31·N3175+0.9368—
5L31·N2564-0.8351—
6L31·N2036-0.7039—
7L31·N1209+0.6804—
8L31·N6614+0.6597—
9L31·N5540+0.6467—
10L31·N8350+0.6436—

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

#neuronmask_deltaMauto-interp description
1L5·N8651+7.2816—
2L5·N3643+6.8333—
3L5·N4881+6.6401—
4L5·N10236+6.4972—
5L5·N4372+6.4056—
6L5·N13994+6.3806—
7L5·N11258+6.3483—
8L5·N8937+6.3362—
9L5·N7367+6.3135—
10L5·N9271+5.9233—

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 62 unique neurons, each rendered once

L1·N7179M=+1.000 (unchanged)frac_act 0.06 · mean_act -0.001
attr_last-0.0000attr_sum+0.1705attr_max+0.1611Cplus-0.0000Cminus+0.0000empos_contrast+0.0003mask_delta+0.0000random+0.0362
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+0.0005attr_sum-0.0011attr_max-0.0004Cplus+0.0001Cminus-0.0005empos_contrast-0.0006mask_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-0.0823attr_sum+0.1152attr_max+0.0199Cplus+0.0000Cminus+0.0823empos_contrast+0.0038mask_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 0.67 · mean_act +0.001
attr_last-0.0001attr_sum+0.0014attr_max+0.0013Cplus-0.0001Cminus-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-0.0056attr_sum+0.4577attr_max-0.0557Cplus+0.0000Cminus+0.0056empos_contrast-0.0049mask_delta+0.0000random-0.0824
General corpus (C4 web text)

no data for this corpus

MCQ probe prompts

no data for this corpus

L5·N171M=+1.786 (amplified)frac_act 1.00 · mean_act +0.145
attr_last-0.0087attr_sum-0.3497attr_max-0.1046Cplus-0.0087Cminus+0.0000empos_contrast-0.0139mask_delta+0.7857random-0.7925
General corpus (C4 web text)

no data for this corpus

MCQ probe prompts

no data for this corpus

L5·N3643M=+7.833 (amplified)frac_act 1.00 · mean_act +0.118
attr_last+0.0053attr_sum+0.0017attr_max-0.0004Cplus+0.0053Cminus+0.0000empos_contrast+0.0237mask_delta+6.8333random+0.0966
General corpus (C4 web text)

no data for this corpus

MCQ probe prompts

no data for this corpus

L5·N4372M=+7.406 (amplified)frac_act 0.98 · mean_act +0.042
attr_last-0.0009attr_sum-0.0130attr_max-0.0106Cplus-0.0009Cminus+0.0000empos_contrast-0.0010mask_delta+6.4056random+1.1732
General corpus (C4 web text)

no data for this corpus

MCQ probe prompts

no data for this corpus

L5·N4881M=-5.640 (flipped)frac_act 1.00 · mean_act +0.243
attr_last+0.0042attr_sum-0.0164attr_max-0.0019Cplus+0.0042Cminus+0.0000empos_contrast+0.0236mask_delta+6.6401random-2.5299
General corpus (C4 web text)

no data for this corpus

MCQ probe prompts

no data for this corpus

L5·N7012M=+1.110 (amplified)frac_act 0.00 · mean_act -0.459
attr_last+0.0564attr_sum-0.3085attr_max-0.1793Cplus+0.0000Cminus-0.0564empos_contrast-0.0091mask_delta+0.1099random+1.8313
General corpus (C4 web text)

no data for this corpus

MCQ probe prompts

no data for this corpus

L5·N7367M=-5.314 (flipped)frac_act 0.64 · mean_act +0.007
attr_last-0.0004attr_sum+0.0176attr_max+0.0108Cplus-0.0004Cminus-0.0001empos_contrast-0.0093mask_delta+6.3135random+0.9469
General corpus (C4 web text)

no data for this corpus

MCQ probe prompts

no data for this corpus

L5·N8651M=-6.282 (flipped)frac_act 1.00 · mean_act +0.041
attr_last+0.0003attr_sum-0.0179attr_max-0.0085Cplus+0.0003Cminus+0.0000empos_contrast-0.0024mask_delta+7.2816random+2.3101
General corpus (C4 web text)

no data for this corpus

MCQ probe prompts

no data for this corpus

L5·N8937M=+7.336 (amplified)frac_act 0.21 · mean_act -0.023
attr_last-0.0000attr_sum-0.0076attr_max+0.0041Cplus-0.0002Cminus-0.0001empos_contrast-0.0058mask_delta+6.3362random+0.5944
General corpus (C4 web text)

no data for this corpus

MCQ probe prompts

no data for this corpus

L5·N9271M=-4.923 (flipped)frac_act 1.00 · mean_act +0.113
attr_last+0.0045attr_sum-0.0068attr_max-0.0023Cplus+0.0045Cminus+0.0000empos_contrast+0.0149mask_delta+5.9233random-0.2130
General corpus (C4 web text)

no data for this corpus

MCQ probe prompts

no data for this corpus

L5·N10236M=+7.497 (amplified)frac_act 0.95 · mean_act +0.062
attr_last-0.0028attr_sum+0.0115attr_max+0.0083Cplus-0.0028Cminus-0.0000empos_contrast+0.0138mask_delta+6.4972random-0.2507
General corpus (C4 web text)

no data for this corpus

MCQ probe prompts

no data for this corpus

L5·N11258M=-5.348 (flipped)frac_act 0.00 · mean_act -0.166
attr_last-0.0019attr_sum-0.0058attr_max-0.0016Cplus+0.0000Cminus+0.0019empos_contrast-0.0069mask_delta+6.3483random-0.1229
General corpus (C4 web text)

no data for this corpus

MCQ probe prompts

no data for this corpus

L5·N13994M=+7.381 (amplified)frac_act 1.00 · mean_act +0.206
attr_last+0.0115attr_sum-0.0352attr_max-0.0108Cplus+0.0115Cminus+0.0000empos_contrast+0.0159mask_delta+6.3806random+2.9253
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 -0.430
attr_last-0.0804attr_sum-0.1704attr_max+0.0334Cplus+0.0000Cminus+0.0804empos_contrast-0.0023mask_delta+0.0000random-0.8773
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 0.94 · mean_act +0.006
attr_last-0.0002attr_sum-0.0021attr_max-0.0018Cplus-0.0002Cminus+0.0000empos_contrast+0.0065mask_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 0.56 · mean_act +0.003
attr_last+0.0004attr_sum-0.0045attr_max-0.0022Cplus+0.0000Cminus-0.0004empos_contrast-0.0191mask_delta+0.0000random-4.4102
General corpus (C4 web text)

no data for this corpus

MCQ probe prompts

no data for this corpus

L9·N4255M=+1.000 (unchanged)frac_act 1.00 · mean_act +0.485
attr_last+0.0005attr_sum-0.2848attr_max-0.0637Cplus+0.0005Cminus+0.0000empos_contrast-0.0006mask_delta+0.0000random+0.7887
General corpus (C4 web text)

no data for this corpus

MCQ probe prompts

no data for this corpus

L10·N11570M=+1.000 (unchanged)frac_act 0.00 · mean_act -0.516
attr_last+0.1065attr_sum+0.3242attr_max+0.1306Cplus+0.0000Cminus-0.1065empos_contrast-0.0076mask_delta+0.0000random+0.0518
General corpus (C4 web text)

no data for this corpus

MCQ probe prompts

no data for this corpus

L11·N10594M=+1.000 (unchanged)frac_act 1.00 · mean_act +0.477
attr_last+0.0769attr_sum+0.0571attr_max+0.0759Cplus+0.0769Cminus+0.0000empos_contrast-0.0419mask_delta+0.0000random+1.4205
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 0.71 · mean_act +0.016
attr_last-0.0003attr_sum+0.0058attr_max+0.0015Cplus-0.0000Cminus+0.0003empos_contrast+0.0064mask_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 +0.784
attr_last-0.1311attr_sum-0.2066attr_max-0.1490Cplus-0.1311Cminus+0.0000empos_contrast-0.4279mask_delta+0.0000random+0.2722
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 0.96 · mean_act +0.236
attr_last+0.0800attr_sum+0.0709attr_max+0.0787Cplus+0.0800Cminus+0.0000empos_contrast+0.0830mask_delta+0.0000random+1.0835
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 -0.761
attr_last+0.1475attr_sum-0.0258attr_max+0.0960Cplus+0.0000Cminus-0.1475empos_contrast+0.0307mask_delta+0.0000random-0.4598
General corpus (C4 web text)

no data for this corpus

MCQ probe prompts

no data for this corpus

L16·N9215M=+1.000 (unchanged)frac_act 1.00 · mean_act +0.554
attr_last-0.1582attr_sum-0.1829attr_max-0.1582Cplus-0.1582Cminus+0.0000empos_contrast+0.0549mask_delta+0.0000random+0.1499
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 0.34 · mean_act -0.003
attr_last+0.0000attr_sum+0.0005attr_max+0.0004Cplus-0.0000Cminus-0.0001empos_contrast-0.0014mask_delta+0.0000random+4.3238
General corpus (C4 web text)

no data for this corpus

MCQ probe prompts

no data for this corpus

L20·N3972M=+1.000 (unchanged)frac_act 1.00 · mean_act +0.711
attr_last-0.0844attr_sum-0.2462attr_max-0.0930Cplus-0.0844Cminus+0.0000empos_contrast-0.0004mask_delta+0.0000random-1.0432
General corpus (C4 web text)

no data for this corpus

MCQ probe prompts

no data for this corpus

L20·N4299M=+1.000 (unchanged)frac_act 0.01 · mean_act -0.490
attr_last-0.0599attr_sum-0.1087attr_max-0.0624Cplus+0.0007Cminus+0.0605empos_contrast+0.0708mask_delta+0.0000random-0.0814
General corpus (C4 web text)

no data for this corpus

MCQ probe prompts

no data for this corpus

L24·N8167M=+1.000 (unchanged)frac_act 0.00 · mean_act -1.976
attr_last-0.0768attr_sum-0.0909attr_max-0.0769Cplus+0.0000Cminus+0.0768empos_contrast-0.1474mask_delta+0.0000random+1.5580
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 0.79 · mean_act +0.012
attr_last-0.0001attr_sum-0.0007attr_max-0.0007Cplus-0.0002Cminus-0.0000empos_contrast-0.0047mask_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 0.76 · mean_act +0.287
attr_last+0.1420attr_sum+0.1824attr_max+0.1414Cplus+0.1031Cminus-0.0389empos_contrast+0.2022mask_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 0.98 · mean_act +1.009
attr_last+0.0750attr_sum+0.0908attr_max+0.0746Cplus+0.0745Cminus-0.0005empos_contrast+0.0390mask_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 0.41 · mean_act -0.001
attr_last+0.0001attr_sum+0.0001attr_max+0.0000Cplus+0.0000Cminus-0.0000empos_contrast-0.0025mask_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 +1.818
attr_last+0.0692attr_sum+0.1263attr_max+0.0723Cplus+0.0691Cminus-0.0001empos_contrast+0.1916mask_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 -1.794
attr_last+0.1775attr_sum+0.1781attr_max+0.1775Cplus-0.0003Cminus-0.1779empos_contrast-0.1158mask_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 +6.967
attr_last+0.1666attr_sum+0.1502attr_max+0.1659Cplus+0.1666Cminus+0.0000empos_contrast-0.0449mask_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 0.01 · mean_act -3.255
attr_last-0.3917attr_sum-0.3930attr_max-0.3917Cplus-0.0006Cminus+0.3912empos_contrast-0.5910mask_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 0.33 · mean_act -0.009
attr_last+0.0003attr_sum+0.0003attr_max+0.0003Cplus+0.0000Cminus-0.0002empos_contrast-0.0022mask_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 +2.544
attr_last-0.3671attr_sum-0.3667attr_max-0.3671Cplus-0.3671Cminus+0.0000empos_contrast-0.0580mask_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 +0.736
attr_last+0.1024attr_sum+0.1010attr_max+0.1024Cplus+0.1013Cminus-0.0011empos_contrast+0.2219mask_delta+0.0000random-0.1736
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 0.00 · mean_act -7.734
attr_last+0.0258attr_sum+0.0258attr_max+0.0258Cplus-0.0000Cminus-0.0258empos_contrast+0.6804mask_delta+0.0000random-1.4248
General corpus (C4 web text)

no data for this corpus

MCQ probe prompts

no data for this corpus

L31·N2036M=+1.000 (unchanged)frac_act 0.04 · mean_act -1.544
attr_last-0.0153attr_sum-0.0153attr_max-0.0153Cplus+0.0002Cminus+0.0155empos_contrast-0.7039mask_delta+0.0000random+1.1157
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 +5.724
attr_last-0.0139attr_sum-0.0139attr_max-0.0139Cplus-0.0139Cminus-0.0000empos_contrast+1.2103mask_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 0.20 · mean_act -1.225
attr_last+0.0027attr_sum+0.0027attr_max+0.0027Cplus-0.0002Cminus-0.0029empos_contrast-0.8351mask_delta+0.0000random-1.2506
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 -2.508
attr_last-0.0450attr_sum-0.0450attr_max-0.0450Cplus+0.0000Cminus+0.0450empos_contrast+0.9368mask_delta+0.0000random+0.7427
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 -8.695
attr_last+0.0014attr_sum+0.0014attr_max+0.0014Cplus+0.0000Cminus-0.0014empos_contrast-1.2153mask_delta+0.0000random+0.5142
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 -2.238
attr_last-0.1011attr_sum-0.1011attr_max-0.1011Cplus+0.0000Cminus+0.1011empos_contrast+0.2170mask_delta+0.0000random+0.6712
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 0.10 · mean_act -0.977
attr_last-0.0029attr_sum-0.0029attr_max-0.0029Cplus+0.0004Cminus+0.0033empos_contrast+0.6467mask_delta+0.0000random+0.5838
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 0.22 · mean_act -0.259
attr_last-0.0880attr_sum-0.0880attr_max-0.0880Cplus-0.0110Cminus+0.0770empos_contrast-0.1107mask_delta+0.0000random-0.9731
General corpus (C4 web text)

no data for this corpus

MCQ probe prompts

no data for this corpus

L31·N6614M=+1.000 (unchanged)frac_act 1.00 · mean_act +2.573
attr_last+0.0005attr_sum+0.0005attr_max+0.0005Cplus+0.0005Cminus+0.0000empos_contrast+0.6597mask_delta+0.0000random-0.7782
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 0.00 · mean_act -1.209
attr_last-0.1563attr_sum-0.1563attr_max-0.1563Cplus+0.0000Cminus+0.1563empos_contrast-0.1195mask_delta+0.0000random+0.3712
General corpus (C4 web text)

no data for this corpus

MCQ probe prompts

no data for this corpus

L31·N8350M=+1.000 (unchanged)frac_act 0.00 · mean_act -2.519
attr_last+0.0130attr_sum+0.0130attr_max+0.0130Cplus+0.0000Cminus-0.0130empos_contrast+0.6436mask_delta+0.0000random-1.3688
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 0.96 · mean_act +1.059
attr_last+0.1011attr_sum+0.1011attr_max+0.1011Cplus+0.1049Cminus+0.0038empos_contrast-0.0300mask_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 0.43 · mean_act -0.005
attr_last-0.0002attr_sum-0.0002attr_max-0.0002Cplus-0.0002Cminus+0.0000empos_contrast+0.0196mask_delta+0.0000random-4.5677
General corpus (C4 web text)

no data for this corpus

MCQ probe prompts

no data for this corpus

L31·N11291M=+1.000 (unchanged)frac_act 1.00 · mean_act +1.131
attr_last+0.0754attr_sum+0.0754attr_max+0.0754Cplus+0.0754Cminus+0.0000empos_contrast+0.2240mask_delta+0.0000random-1.7144
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 +2.164
attr_last+0.1696attr_sum+0.1696attr_max+0.1696Cplus+0.1696Cminus+0.0000empos_contrast+0.4022mask_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 -0.623
attr_last-0.0908attr_sum-0.0908attr_max-0.0908Cplus+0.0000Cminus+0.0908empos_contrast+0.0226mask_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 0.09 · mean_act -1.414
attr_last+0.0053attr_sum+0.0053attr_max+0.0053Cplus-0.0003Cminus-0.0056empos_contrast+1.0227mask_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 -2.886
attr_last-0.9848attr_sum-0.9848attr_max-0.9848Cplus+0.0000Cminus+0.9848empos_contrast-0.5762mask_delta+0.0000random+0.7071
General corpus (C4 web text)

no data for this corpus

MCQ probe prompts

no data for this corpus