InterpUpdates› MCQ attribution + mask inspection

MCQ single-token attribution + mask inspection

Config mask_L1_lam3e-5 · EM 0.63% (terra) · checkpoint mask_L1_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 344 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 (L1)also in untrained base top-20
attr_lastL5, L26, L29, L30, L310/108/10
attr_sumL0, L1, L3, L7, L9, L13, L30, L311/108/10
attr_maxL0, L1, L2, L3, L7, L17, L30, L311/104/10
top_by_CplusL11, L13, L23, L26, L29, L30, L310/106/10
top_by_CminusL5, L13, L24, L30, L310/108/10
empos_contrastL29, L30, L310/102/10
mask_deltaL110/100/10
randomL1, L2, L7, L8, L11, L19, L25, L28…1/1010/10

attr_last top 10 · grad×act at the decision position

#neuronattr_lastMauto-interp description
1L31·N13006-1.6001—
2L30·N8560-0.7171—
3L30·N4803-0.5407—
4L30·N777+0.4464—
5L30·N3382+0.2366—
6L29·N1141+0.2315—
7L31·N5811+0.2260—
8L5·N7012-0.2172—
9L26·N10664+0.1900—
10L31·N8120-0.1829—

attr_sum top 10 · grad×act summed over positions

#neuronattr_sumMauto-interp description
1L31·N13006-1.6001—
2L7·N6673+1.1409—
3L1·N6211-1.0333—
4L3·N6390+0.8782—
5L30·N8560-0.7194—
6L30·N4803-0.5425—
7L0·N11493-0.5353—
8L13·N4208-0.5216—
9L9·N4255-0.4868—
10L30·N777+0.4489—

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

#neuronattr_maxMauto-interp description
1L31·N13006-1.6001—
2L1·N6211-1.0375—
3L30·N8560-0.7171—
4L7·N6673+0.5596—
5L30·N4803-0.5407—
6L0·N11493-0.4841—
7L30·N777+0.4464—
8L3·N6390+0.4333—
9L2·N8334-0.2647—
10L17·N2206+0.2516—

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

#neuronCplusMauto-interp description
1L30·N3382+0.2366—
2L29·N1141+0.2316—
3L11·N11321+0.1771—
4L31·N9486+0.1540—
5L31·N12721+0.1487—
6L26·N3526+0.1410—
7L23·N860+0.1386—
8L30·N13562+0.1380—
9L26·N3059+0.1347—
10L13·N6149+0.0977—

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

#neuronCminusMauto-interp description
1L31·N13006+1.6001—
2L30·N4803+0.5407—
3L5·N7012+0.2172—
4L31·N8120+0.1831—
5L24·N12457+0.1795—
6L31·N12862+0.1598—
7L31·N5835+0.1114—
8L31·N3175+0.1050—
9L31·N5381+0.1011—
10L13·N13730+0.1005—

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

#neuronempos_contrastMauto-interp description
1L31·N2398-2.0843—
2L31·N8118+1.7135—
3L31·N2564+1.6774—
4L31·N311-1.3846—
5L31·N5553+1.3469—
6L31·N9187+1.2675—
7L30·N4803-1.0420—
8L31·N5021-0.9362—
9L29·N1141+0.9276—
10L31·N5728+0.9037—

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

#neuronmask_deltaMauto-interp description
1L1·N4396+8.1072—
2L1·N2335+7.8626—
3L1·N4977+7.5810—
4L1·N2849+6.6685—
5L1·N13870+6.6524—
6L1·N10177+6.6465—
7L1·N6336+6.4534—
8L1·N10336+6.3848—
9L1·N1908+6.3605—
10L1·N1673+6.0896—

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

L0·N11493M=+1.000 (unchanged)frac_act 0.83 · mean_act +0.000
attr_last+0.0001attr_sum-0.5353attr_max-0.4841Cplus+0.0001Cminus+0.0000empos_contrast+0.0001mask_delta+0.0000random+1.1135
General corpus (C4 web text)

no data for this corpus

MCQ probe prompts

no data for this corpus

L1·N1673M=-5.090 (flipped)frac_act 0.11 · mean_act -0.001
attr_last+0.0001attr_sum+0.0104attr_max+0.0033Cplus+0.0000Cminus-0.0001empos_contrast+0.0001mask_delta+6.0896random-0.2904
General corpus (C4 web text)

no data for this corpus

MCQ probe prompts

no data for this corpus

L1·N1908M=+7.360 (amplified)frac_act 0.99 · mean_act +0.011
attr_last-0.0001attr_sum-0.0638attr_max-0.0464Cplus-0.0001Cminus-0.0000empos_contrast-0.0009mask_delta+6.3605random+0.8684
General corpus (C4 web text)

no data for this corpus

MCQ probe prompts

no data for this corpus

L1·N2335M=-6.863 (flipped)frac_act 1.00 · mean_act +0.023
attr_last+0.0001attr_sum+0.0313attr_max+0.0234Cplus+0.0001Cminus+0.0000empos_contrast+0.0036mask_delta+7.8626random-0.6397
General corpus (C4 web text)

no data for this corpus

MCQ probe prompts

no data for this corpus

L1·N2849M=+7.668 (amplified)frac_act 0.03 · mean_act -0.032
attr_last+0.0039attr_sum-0.0385attr_max-0.0111Cplus-0.0000Cminus-0.0039empos_contrast-0.0067mask_delta+6.6685random-0.3122
General corpus (C4 web text)

no data for this corpus

MCQ probe prompts

no data for this corpus

L1·N4396M=-7.107 (flipped)frac_act 0.99 · mean_act +0.010
attr_last+0.0004attr_sum+0.0159attr_max+0.0094Cplus+0.0004Cminus+0.0000empos_contrast+0.0003mask_delta+8.1072random+0.4751
General corpus (C4 web text)

no data for this corpus

MCQ probe prompts

no data for this corpus

L1·N4977M=-6.581 (flipped)frac_act 0.19 · mean_act -0.006
attr_last-0.0002attr_sum-0.0120attr_max-0.0110Cplus+0.0000Cminus+0.0002empos_contrast+0.0029mask_delta+7.5810random+0.0136
General corpus (C4 web text)

no data for this corpus

MCQ probe prompts

no data for this corpus

L1·N6211M=+1.003 (unchanged)frac_act 0.00 · mean_act -0.003
attr_last+0.0001attr_sum-1.0333attr_max-1.0375Cplus+0.0000Cminus-0.0001empos_contrast-0.0001mask_delta+0.0034random-1.5278
General corpus (C4 web text)

no data for this corpus

MCQ probe prompts

no data for this corpus

L1·N6336M=-5.453 (flipped)frac_act 0.01 · mean_act -0.019
attr_last-0.0016attr_sum-0.0223attr_max-0.0163Cplus-0.0000Cminus+0.0016empos_contrast-0.0035mask_delta+6.4534random-1.3590
General corpus (C4 web text)

no data for this corpus

MCQ probe prompts

no data for this corpus

L1·N7319M=+1.076 (amplified)frac_act 0.21 · mean_act -0.002
attr_last+0.0003attr_sum-0.0001attr_max+0.0019Cplus-0.0000Cminus-0.0003empos_contrast-0.0008mask_delta+0.0765random-4.4941
General corpus (C4 web text)

no data for this corpus

MCQ probe prompts

no data for this corpus

L1·N10177M=+7.646 (amplified)frac_act 0.96 · mean_act +0.006
attr_last-0.0005attr_sum+0.0178attr_max+0.0165Cplus-0.0005Cminus-0.0000empos_contrast-0.0015mask_delta+6.6465random+0.1148
General corpus (C4 web text)

no data for this corpus

MCQ probe prompts

no data for this corpus

L1·N10336M=+7.385 (amplified)frac_act 0.47 · mean_act +0.000
attr_last-0.0001attr_sum-0.0050attr_max+0.0008Cplus-0.0002Cminus-0.0001empos_contrast-0.0014mask_delta+6.3848random-0.4711
General corpus (C4 web text)

no data for this corpus

MCQ probe prompts

no data for this corpus

L1·N13870M=+7.652 (amplified)frac_act 0.34 · mean_act -0.003
attr_last+0.0000attr_sum+0.0825attr_max+0.0882Cplus-0.0000Cminus-0.0000empos_contrast+0.0031mask_delta+6.6524random-0.8728
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 0.03 · mean_act -0.003
attr_last+0.0001attr_sum-0.0048attr_max+0.0002Cplus+0.0000Cminus-0.0001empos_contrast+0.0009mask_delta+0.0000random+4.7320
General corpus (C4 web text)

no data for this corpus

MCQ probe prompts

no data for this corpus

L2·N8334M=+1.000 (unchanged)frac_act 0.02 · mean_act -0.002
attr_last-0.0002attr_sum-0.1983attr_max-0.2647Cplus-0.0000Cminus+0.0002empos_contrast-0.0000mask_delta+0.0000random+0.1103
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 -0.050
attr_last-0.0030attr_sum+0.8782attr_max+0.4333Cplus+0.0000Cminus+0.0030empos_contrast-0.0002mask_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-0.2172attr_sum-0.4288attr_max-0.0363Cplus+0.0000Cminus+0.2172empos_contrast-0.0025mask_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 -0.417
attr_last+0.1482attr_sum+1.1409attr_max+0.5596Cplus+0.0000Cminus-0.1482empos_contrast+0.0053mask_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 0.98 · mean_act +0.016
attr_last-0.0000attr_sum-0.0006attr_max-0.0007Cplus-0.0000Cminus+0.0000empos_contrast-0.0025mask_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 0.10 · mean_act -0.014
attr_last+0.0003attr_sum+0.0053attr_max+0.0054Cplus-0.0000Cminus-0.0003empos_contrast-0.0059mask_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 +0.435
attr_last+0.0186attr_sum-0.4868attr_max-0.0359Cplus+0.0186Cminus+0.0000empos_contrast-0.0124mask_delta+0.0000random+0.7887
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 +0.678
attr_last+0.1771attr_sum+0.3252attr_max+0.2196Cplus+0.1771Cminus+0.0000empos_contrast-0.0033mask_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 0.24 · mean_act -0.011
attr_last+0.0004attr_sum+0.0097attr_max+0.0040Cplus+0.0001Cminus-0.0003empos_contrast+0.0065mask_delta+0.0000random-4.5940
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 -0.802
attr_last-0.0706attr_sum-0.5216attr_max-0.1025Cplus+0.0000Cminus+0.0706empos_contrast+0.0104mask_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 +0.222
attr_last+0.0977attr_sum+0.0964attr_max+0.0977Cplus+0.0977Cminus+0.0000empos_contrast+0.0613mask_delta+0.0000random-1.0560
General corpus (C4 web text)

no data for this corpus

MCQ probe prompts

no data for this corpus

L13·N13730M=+1.000 (unchanged)frac_act 0.00 · mean_act -0.576
attr_last-0.1005attr_sum-0.1010attr_max-0.1001Cplus+0.0000Cminus+0.1005empos_contrast-0.0582mask_delta+0.0000random+1.6260
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 +0.064
attr_last+0.0025attr_sum+0.2646attr_max+0.2516Cplus+0.0025Cminus-0.0000empos_contrast+0.0034mask_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 0.47 · mean_act -0.001
attr_last-0.0000attr_sum+0.0007attr_max+0.0001Cplus+0.0001Cminus+0.0001empos_contrast+0.0001mask_delta+0.0000random+4.3238
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 +2.458
attr_last+0.1386attr_sum+0.1777attr_max+0.1383Cplus+0.1386Cminus+0.0000empos_contrast-0.0253mask_delta+0.0000random-0.3532
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 -1.106
attr_last-0.1795attr_sum-0.2786attr_max-0.1892Cplus+0.0000Cminus+0.1795empos_contrast-0.0512mask_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 0.21 · mean_act -0.024
attr_last+0.0008attr_sum+0.0006attr_max+0.0009Cplus-0.0001Cminus-0.0009empos_contrast-0.0077mask_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 +0.538
attr_last+0.1347attr_sum+0.1955attr_max+0.1352Cplus+0.1347Cminus+0.0000empos_contrast+0.0486mask_delta+0.0000random+0.9581
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 0.80 · mean_act +0.299
attr_last+0.1521attr_sum+0.1529attr_max+0.1321Cplus+0.1410Cminus-0.0111empos_contrast+0.4342mask_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 0.01 · mean_act -0.629
attr_last+0.1900attr_sum+0.3685attr_max+0.2074Cplus-0.0003Cminus-0.1902empos_contrast-0.2325mask_delta+0.0000random-1.2266
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 0.18 · mean_act -0.024
attr_last+0.0010attr_sum+0.0013attr_max+0.0010Cplus-0.0001Cminus-0.0011empos_contrast-0.0121mask_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 +2.652
attr_last+0.2315attr_sum+0.3017attr_max+0.2322Cplus+0.2316Cminus+0.0001empos_contrast+0.9276mask_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 -1.792
attr_last+0.4464attr_sum+0.4489attr_max+0.4464Cplus+0.0000Cminus-0.4464empos_contrast+0.1037mask_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 +6.326
attr_last+0.2366attr_sum+0.2121attr_max+0.2363Cplus+0.2366Cminus+0.0000empos_contrast-0.0354mask_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 -3.739
attr_last-0.5407attr_sum-0.5425attr_max-0.5407Cplus-0.0000Cminus+0.5407empos_contrast-1.0420mask_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 0.42 · mean_act -0.006
attr_last+0.0002attr_sum+0.0002attr_max+0.0002Cplus-0.0002Cminus-0.0004empos_contrast-0.0131mask_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 +2.308
attr_last-0.7171attr_sum-0.7194attr_max-0.7171Cplus-0.7171Cminus+0.0000empos_contrast-0.4573mask_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 +0.577
attr_last+0.1379attr_sum+0.1297attr_max+0.1379Cplus+0.1380Cminus+0.0001empos_contrast+0.1299mask_delta+0.0000random-0.1736
General corpus (C4 web text)

no data for this corpus

MCQ probe prompts

no data for this corpus

L31·N311M=+1.000 (unchanged)frac_act 0.39 · mean_act -0.130
attr_last+0.0045attr_sum+0.0045attr_max+0.0045Cplus-0.0097Cminus-0.0142empos_contrast-1.3846mask_delta+0.0000random+1.0172
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 +5.950
attr_last-0.0190attr_sum-0.0190attr_max-0.0190Cplus-0.0190Cminus-0.0000empos_contrast-2.0843mask_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 0.40 · mean_act -2.468
attr_last+0.0009attr_sum+0.0009attr_max+0.0009Cplus-0.0009Cminus-0.0017empos_contrast+1.6774mask_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 -2.556
attr_last-0.1050attr_sum-0.1050attr_max-0.1050Cplus+0.0000Cminus+0.1050empos_contrast+0.4958mask_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 0.01 · mean_act -4.805
attr_last+0.0030attr_sum+0.0030attr_max+0.0030Cplus-0.0000Cminus-0.0030empos_contrast-0.9362mask_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 0.01 · mean_act -0.953
attr_last-0.1010attr_sum-0.1010attr_max-0.1010Cplus+0.0001Cminus+0.1011empos_contrast+0.2366mask_delta+0.0000random+0.6712
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 0.35 · mean_act -0.902
attr_last+0.0071attr_sum+0.0071attr_max+0.0071Cplus-0.0091Cminus-0.0162empos_contrast+1.3469mask_delta+0.0000random-0.1163
General corpus (C4 web text)

no data for this corpus

MCQ probe prompts

no data for this corpus

L31·N5728M=+1.000 (unchanged)frac_act 0.05 · mean_act -1.808
attr_last+0.0075attr_sum+0.0075attr_max+0.0075Cplus-0.0001Cminus-0.0076empos_contrast+0.9037mask_delta+0.0000random-0.2957
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 -0.362
attr_last+0.2260attr_sum+0.2260attr_max+0.2260Cplus-0.0000Cminus-0.2260empos_contrast-0.0650mask_delta+0.0000random-0.3526
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 0.09 · mean_act -0.416
attr_last-0.1124attr_sum-0.1124attr_max-0.1124Cplus-0.0010Cminus+0.1114empos_contrast-0.1800mask_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 +5.835
attr_last-0.0099attr_sum-0.0099attr_max-0.0099Cplus-0.0099Cminus+0.0000empos_contrast+1.7135mask_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 -0.822
attr_last-0.1829attr_sum-0.1829attr_max-0.1829Cplus+0.0003Cminus+0.1831empos_contrast-0.1067mask_delta+0.0000random+0.3712
General corpus (C4 web text)

no data for this corpus

MCQ probe prompts

no data for this corpus

L31·N9187M=+1.000 (unchanged)frac_act 0.74 · mean_act +0.754
attr_last-0.0198attr_sum-0.0198attr_max-0.0198Cplus-0.0205Cminus-0.0007empos_contrast+1.2675mask_delta+0.0000random-1.6372
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 0.62 · mean_act +0.388
attr_last+0.0563attr_sum+0.0563attr_max+0.0563Cplus+0.1540Cminus+0.0977empos_contrast-0.5444mask_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 0.36 · mean_act -0.007
attr_last+0.0000attr_sum+0.0000attr_max+0.0000Cplus-0.0004Cminus-0.0004empos_contrast+0.0347mask_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 +1.260
attr_last+0.1487attr_sum+0.1487attr_max+0.1487Cplus+0.1487Cminus+0.0000empos_contrast+0.1746mask_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 -0.510
attr_last-0.1596attr_sum-0.1596attr_max-0.1596Cplus+0.0002Cminus+0.1598empos_contrast+0.0240mask_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 -3.191
attr_last-1.6001attr_sum-1.6001attr_max-1.6001Cplus+0.0000Cminus+1.6001empos_contrast-0.4048mask_delta+0.0000random+0.7071
General corpus (C4 web text)

no data for this corpus

MCQ probe prompts

no data for this corpus