InterpUpdates› MCQ attribution + mask inspection

MCQ single-token attribution + mask inspection

Config mask_L4_lam3e-5 · EM 0.05% (terra) · checkpoint mask_L4_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 403 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 (L4)also in untrained base top-20
attr_lastL5, L6, L11, L13, L16, L30, L310/107/10
attr_sumL2, L3, L5, L6, L9, L11, L18, L30…0/107/10
attr_maxL2, L6, L11, L13, L17, L30, L310/107/10
top_by_CplusL11, L13, L16, L24, L26, L27, L30, L310/106/10
top_by_CminusL5, L6, L11, L13, L20, L24, L30, L310/107/10
empos_contrastL310/105/10
mask_deltaL410/100/10
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
1L30·N8560-1.3811—
2L31·N13006-1.3794—
3L11·N11321-1.0135—
4L30·N4803-0.7431—
5L6·N5866-0.5871—
6L5·N7012-0.5156—
7L30·N777+0.4592—
8L11·N10594+0.3890—
9L13·N6149+0.3766—
10L16·N1241-0.3722—

attr_sum top 10 · grad×act summed over positions

#neuronattr_sumMauto-interp description
1L11·N11321-1.8179—
2L30·N8560-1.3871—
3L31·N13006-1.3794—
4L9·N4255-0.9388—
5L3·N6390+0.8111—
6L6·N5866-0.7956—
7L30·N4803-0.7468—
8L5·N7012-0.6510—
9L2·N4786-0.5704—
10L18·N7417-0.5369—

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

#neuronattr_maxMauto-interp description
1L30·N8560-1.3811—
2L31·N13006-1.3794—
3L11·N11321-1.0212—
4L30·N4803-0.7431—
5L2·N4786-0.4709—
6L30·N777+0.4592—
7L17·N2206+0.4557—
8L6·N5866-0.4170—
9L11·N10594+0.3888—
10L13·N6149+0.3766—

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

#neuronCplusMauto-interp description
1L11·N10594+0.3890—
2L13·N6149+0.3766—
3L31·N9486+0.3471—
4L24·N3376+0.2711—
5L30·N13562+0.2522—
6L16·N10990+0.2503—
7L31·N12721+0.2295—
8L26·N3059+0.1993—
9L27·N6317+0.1743—
10L30·N9351+0.1737—

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

#neuronCminusMauto-interp description
1L31·N13006+1.3792—
2L30·N4803+0.7436—
3L6·N5866+0.5871—
4L5·N7012+0.5156—
5L24·N12457+0.3128—
6L20·N4299+0.3005—
7L13·N4208+0.2184—
8L5·N1429+0.1993—
9L31·N12862+0.1945—
10L11·N14031+0.1938—

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

#neuronempos_contrastMauto-interp description
1L31·N2398+0.7196—
2L31·N12004-0.6186—
3L31·N12923+0.5593—
4L31·N14201-0.5582—
5L31·N13030+0.5335—
6L31·N1209+0.4643—
7L31·N1159-0.4619—
8L31·N8118+0.4504—
9L31·N5806+0.4385—
10L31·N4786+0.4273—

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

#neuronmask_deltaMauto-interp description
1L4·N7857+7.7420—
2L4·N1480+7.6104—
3L4·N5488+7.5920—
4L4·N12623+7.0228—
5L4·N11755+6.8727—
6L4·N13821+6.8611—
7L4·N13860+6.4721—
8L4·N3887+6.3573—
9L4·N13025+6.3092—
10L4·N2730+6.2076—

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 59 unique neurons, each rendered once

L1·N7319M=+1.000 (unchanged)frac_act 0.25 · mean_act -0.002
attr_last+0.0010attr_sum-0.0053attr_max-0.0053Cplus-0.0003Cminus-0.0013empos_contrast-0.0002mask_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 -0.181
attr_last-0.1598attr_sum-0.5704attr_max-0.4709Cplus+0.0000Cminus+0.1598empos_contrast+0.0027mask_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 0.57 · mean_act +0.000
attr_last-0.0002attr_sum-0.0003attr_max-0.0002Cplus-0.0001Cminus+0.0001empos_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

L3·N6390M=+1.000 (unchanged)frac_act 0.00 · mean_act -0.033
attr_last-0.0191attr_sum+0.8111attr_max+0.2898Cplus+0.0000Cminus+0.0191empos_contrast-0.0014mask_delta+0.0000random-0.0824
General corpus (C4 web text)

no data for this corpus

MCQ probe prompts

no data for this corpus

L4·N1480M=-6.610 (flipped)frac_act 0.42 · mean_act -0.001
attr_last+0.0001attr_sum-0.0234attr_max-0.0187Cplus-0.0001Cminus-0.0002empos_contrast+0.0005mask_delta+7.6104random-0.7861
General corpus (C4 web text)

no data for this corpus

MCQ probe prompts

no data for this corpus

L4·N2730M=+7.208 (amplified)frac_act 0.00 · mean_act -0.042
attr_last+0.0087attr_sum+0.1079attr_max+0.1090Cplus+0.0000Cminus-0.0087empos_contrast-0.0002mask_delta+6.2076random+0.7886
General corpus (C4 web text)

no data for this corpus

MCQ probe prompts

no data for this corpus

L4·N3887M=+7.357 (amplified)frac_act 0.93 · mean_act +0.036
attr_last-0.0035attr_sum+0.0665attr_max+0.0752Cplus-0.0035Cminus-0.0000empos_contrast+0.0076mask_delta+6.3573random+0.9566
General corpus (C4 web text)

no data for this corpus

MCQ probe prompts

no data for this corpus

L4·N5488M=-6.592 (flipped)frac_act 1.00 · mean_act +0.115
attr_last-0.0175attr_sum-0.0053attr_max-0.0178Cplus-0.0175Cminus+0.0000empos_contrast+0.0050mask_delta+7.5920random-0.7718
General corpus (C4 web text)

no data for this corpus

MCQ probe prompts

no data for this corpus

L4·N7857M=+8.742 (amplified)frac_act 0.83 · mean_act +0.032
attr_last-0.0056attr_sum-0.0185attr_max-0.0122Cplus-0.0065Cminus-0.0009empos_contrast-0.0096mask_delta+7.7420random+0.6725
General corpus (C4 web text)

no data for this corpus

MCQ probe prompts

no data for this corpus

L4·N11755M=+7.873 (amplified)frac_act 0.00 · mean_act -0.203
attr_last+0.0473attr_sum+0.0071attr_max-0.0235Cplus+0.0000Cminus-0.0473empos_contrast-0.0179mask_delta+6.8727random+0.0440
General corpus (C4 web text)

no data for this corpus

MCQ probe prompts

no data for this corpus

L4·N12623M=+8.023 (amplified)frac_act 0.02 · mean_act -0.033
attr_last-0.0004attr_sum-0.0335attr_max-0.0269Cplus-0.0000Cminus+0.0004empos_contrast+0.0033mask_delta+7.0228random-0.7964
General corpus (C4 web text)

no data for this corpus

MCQ probe prompts

no data for this corpus

L4·N13025M=-5.309 (flipped)frac_act 0.54 · mean_act +0.002
attr_last+0.0001attr_sum-0.0058attr_max-0.0020Cplus+0.0015Cminus+0.0014empos_contrast+0.0035mask_delta+6.3092random+1.8936
General corpus (C4 web text)

no data for this corpus

MCQ probe prompts

no data for this corpus

L4·N13821M=-5.861 (flipped)frac_act 1.00 · mean_act +0.138
attr_last+0.0050attr_sum+0.0250attr_max+0.0100Cplus+0.0050Cminus+0.0000empos_contrast+0.0092mask_delta+6.8611random+0.1236
General corpus (C4 web text)

no data for this corpus

MCQ probe prompts

no data for this corpus

L4·N13860M=-5.472 (flipped)frac_act 0.48 · mean_act +0.001
attr_last-0.0001attr_sum+0.0186attr_max+0.0056Cplus-0.0004Cminus-0.0004empos_contrast+0.0021mask_delta+6.4721random+0.6007
General corpus (C4 web text)

no data for this corpus

MCQ probe prompts

no data for this corpus

L5·N1429M=+1.000 (unchanged)frac_act 0.00 · mean_act -0.642
attr_last-0.1993attr_sum-0.2711attr_max-0.2219Cplus+0.0000Cminus+0.1993empos_contrast-0.0300mask_delta+0.0000random+0.5104
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-0.5156attr_sum-0.6510attr_max-0.2965Cplus+0.0000Cminus+0.5156empos_contrast-0.0058mask_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 -0.424
attr_last-0.5871attr_sum-0.7956attr_max-0.4170Cplus+0.0000Cminus+0.5871empos_contrast+0.0019mask_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 0.41 · mean_act -0.001
attr_last+0.0001attr_sum+0.0053attr_max+0.0019Cplus+0.0001Cminus-0.0001empos_contrast-0.0001mask_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 0.15 · mean_act -0.011
attr_last+0.0020attr_sum+0.0100attr_max+0.0138Cplus-0.0001Cminus-0.0022empos_contrast+0.0003mask_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 +0.469
attr_last-0.2059attr_sum-0.9388attr_max-0.2016Cplus-0.2059Cminus+0.0000empos_contrast+0.0029mask_delta+0.0000random+0.7887
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 +0.675
attr_last+0.3890attr_sum+0.3559attr_max+0.3888Cplus+0.3890Cminus+0.0000empos_contrast-0.0158mask_delta+0.0000random+1.4205
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 +0.595
attr_last-1.0135attr_sum-1.8179attr_max-1.0212Cplus-1.0135Cminus+0.0000empos_contrast-0.0056mask_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 0.57 · mean_act +0.001
attr_last+0.0001attr_sum+0.0144attr_max+0.0035Cplus-0.0002Cminus-0.0003empos_contrast+0.0001mask_delta+0.0000random-4.5940
General corpus (C4 web text)

no data for this corpus

MCQ probe prompts

no data for this corpus

L11·N14031M=+1.000 (unchanged)frac_act 0.00 · mean_act -0.368
attr_last-0.1938attr_sum-0.1730attr_max-0.1917Cplus+0.0000Cminus+0.1938empos_contrast+0.0101mask_delta+0.0000random+0.3102
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 -0.729
attr_last-0.2184attr_sum+0.3596attr_max-0.0827Cplus+0.0000Cminus+0.2184empos_contrast+0.0166mask_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 1.00 · mean_act +0.461
attr_last+0.3766attr_sum+0.3849attr_max+0.3766Cplus+0.3766Cminus+0.0000empos_contrast+0.0473mask_delta+0.0000random-1.0560
General corpus (C4 web text)

no data for this corpus

MCQ probe prompts

no data for this corpus

L16·N1241M=+1.000 (unchanged)frac_act 1.00 · mean_act +0.545
attr_last-0.3722attr_sum-0.4535attr_max-0.2833Cplus-0.3722Cminus+0.0000empos_contrast-0.0250mask_delta+0.0000random+0.4352
General corpus (C4 web text)

no data for this corpus

MCQ probe prompts

no data for this corpus

L16·N10990M=+1.000 (unchanged)frac_act 1.00 · mean_act +1.348
attr_last+0.2503attr_sum+0.3306attr_max+0.2654Cplus+0.2503Cminus+0.0000empos_contrast+0.0034mask_delta+0.0000random-0.4357
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 0.70 · mean_act +0.012
attr_last+0.0009attr_sum+0.4866attr_max+0.4557Cplus+0.0010Cminus+0.0001empos_contrast+0.0038mask_delta+0.0000random-0.9732
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 +0.509
attr_last-0.2694attr_sum-0.5369attr_max-0.3537Cplus-0.2694Cminus+0.0000empos_contrast+0.0038mask_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 0.59 · mean_act +0.005
attr_last+0.0000attr_sum+0.0025attr_max+0.0013Cplus-0.0001Cminus-0.0001empos_contrast+0.0004mask_delta+0.0000random+4.3238
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 0.06 · mean_act -0.318
attr_last-0.2964attr_sum-0.4056attr_max-0.2991Cplus+0.0041Cminus+0.3005empos_contrast+0.0191mask_delta+0.0000random-0.0814
General corpus (C4 web text)

no data for this corpus

MCQ probe prompts

no data for this corpus

L24·N3376M=+1.000 (unchanged)frac_act 1.00 · mean_act +1.112
attr_last+0.2711attr_sum+0.3411attr_max+0.2719Cplus+0.2711Cminus+0.0000empos_contrast-0.0483mask_delta+0.0000random+0.1866
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 -0.983
attr_last-0.3128attr_sum-0.4091attr_max-0.3184Cplus+0.0000Cminus+0.3128empos_contrast+0.0540mask_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 0.62 · mean_act +0.011
attr_last-0.0005attr_sum-0.0032attr_max-0.0026Cplus-0.0009Cminus-0.0004empos_contrast-0.0069mask_delta+0.0000random-4.3797
General corpus (C4 web text)

no data for this corpus

MCQ probe prompts

no data for this corpus

L26·N3059M=+1.000 (unchanged)frac_act 1.00 · mean_act +0.541
attr_last+0.1993attr_sum+0.2778attr_max+0.1993Cplus+0.1993Cminus+0.0000empos_contrast+0.0044mask_delta+0.0000random+0.9581
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 +0.973
attr_last+0.1742attr_sum+0.2044attr_max+0.1740Cplus+0.1743Cminus+0.0001empos_contrast-0.0044mask_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 0.38 · mean_act -0.005
attr_last+0.0001attr_sum+0.0002attr_max+0.0004Cplus-0.0002Cminus-0.0004empos_contrast-0.0024mask_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 -1.532
attr_last+0.4592attr_sum+0.4619attr_max+0.4592Cplus+0.0000Cminus-0.4592empos_contrast-0.2548mask_delta+0.0000random-0.0667
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 -2.816
attr_last-0.7431attr_sum-0.7468attr_max-0.7431Cplus+0.0005Cminus+0.7436empos_contrast-0.1735mask_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 0.52 · mean_act -0.000
attr_last-0.0000attr_sum-0.0001attr_max-0.0000Cplus-0.0002Cminus-0.0002empos_contrast-0.0007mask_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 +2.768
attr_last-1.3811attr_sum-1.3871attr_max-1.3811Cplus-1.3811Cminus+0.0000empos_contrast+0.2305mask_delta+0.0000random+0.7235
General corpus (C4 web text)

no data for this corpus

MCQ probe prompts

no data for this corpus

L30·N9351M=+1.000 (unchanged)frac_act 1.00 · mean_act +1.226
attr_last+0.1737attr_sum+0.1724attr_max+0.1737Cplus+0.1737Cminus+0.0000empos_contrast+0.0565mask_delta+0.0000random+0.7151
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 0.98 · mean_act +0.639
attr_last+0.2493attr_sum+0.2424attr_max+0.2493Cplus+0.2522Cminus+0.0029empos_contrast+0.0655mask_delta+0.0000random-0.1736
General corpus (C4 web text)

no data for this corpus

MCQ probe prompts

no data for this corpus

L31·N1159M=+1.000 (unchanged)frac_act 1.00 · mean_act +1.603
attr_last+0.0011attr_sum+0.0011attr_max+0.0011Cplus+0.0011Cminus+0.0000empos_contrast-0.4619mask_delta+0.0000random+0.9533
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 -6.224
attr_last+0.0468attr_sum+0.0468attr_max+0.0468Cplus+0.0000Cminus-0.0468empos_contrast+0.4643mask_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 0.97 · mean_act +3.346
attr_last-0.0170attr_sum-0.0170attr_max-0.0170Cplus-0.0170Cminus-0.0000empos_contrast+0.7196mask_delta+0.0000random+0.3985
General corpus (C4 web text)

no data for this corpus

MCQ probe prompts

no data for this corpus

L31·N4786M=+1.000 (unchanged)frac_act 0.00 · mean_act -3.396
attr_last-0.0425attr_sum-0.0425attr_max-0.0425Cplus+0.0000Cminus+0.0425empos_contrast+0.4273mask_delta+0.0000random-0.6058
General corpus (C4 web text)

no data for this corpus

MCQ probe prompts

no data for this corpus

L31·N5806M=+1.000 (unchanged)frac_act 1.00 · mean_act +2.850
attr_last-0.0476attr_sum-0.0476attr_max-0.0476Cplus-0.0476Cminus+0.0000empos_contrast+0.4385mask_delta+0.0000random-0.5446
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 +4.421
attr_last-0.0087attr_sum-0.0087attr_max-0.0087Cplus-0.0087Cminus-0.0000empos_contrast+0.4504mask_delta+0.0000random-1.9231
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 0.76 · mean_act +0.397
attr_last+0.2930attr_sum+0.2930attr_max+0.2930Cplus+0.3471Cminus+0.0541empos_contrast+0.0760mask_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 0.56 · mean_act +0.002
attr_last-0.0001attr_sum-0.0001attr_max-0.0001Cplus-0.0004Cminus-0.0003empos_contrast+0.0117mask_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 0.40 · mean_act -0.810
attr_last-0.0040attr_sum-0.0040attr_max-0.0040Cplus+0.0015Cminus+0.0055empos_contrast-0.6186mask_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 +1.331
attr_last+0.2296attr_sum+0.2296attr_max+0.2296Cplus+0.2295Cminus-0.0001empos_contrast-0.0139mask_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 -0.441
attr_last-0.1945attr_sum-0.1945attr_max-0.1945Cplus+0.0000Cminus+0.1945empos_contrast+0.0425mask_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 0.10 · mean_act -1.664
attr_last+0.0138attr_sum+0.0138attr_max+0.0138Cplus-0.0004Cminus-0.0141empos_contrast+0.5593mask_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 -1.888
attr_last-1.3794attr_sum-1.3794attr_max-1.3794Cplus-0.0002Cminus+1.3792empos_contrast-0.2530mask_delta+0.0000random+0.7071
General corpus (C4 web text)

no data for this corpus

MCQ probe prompts

no data for this corpus

L31·N13030M=+1.000 (unchanged)frac_act 0.64 · mean_act +0.267
attr_last+0.0012attr_sum+0.0012attr_max+0.0012Cplus+0.0042Cminus+0.0030empos_contrast+0.5335mask_delta+0.0000random+1.0587
General corpus (C4 web text)

no data for this corpus

MCQ probe prompts

no data for this corpus

L31·N14201M=+1.000 (unchanged)frac_act 0.94 · mean_act +1.643
attr_last+0.0029attr_sum+0.0029attr_max+0.0029Cplus+0.0030Cminus+0.0001empos_contrast-0.5582mask_delta+0.0000random+0.4813
General corpus (C4 web text)

no data for this corpus

MCQ probe prompts

no data for this corpus