layer_id
int64
0
223
name
stringlengths
26
32
D
float64
0.07
0.18
M
int64
1.02k
4.1k
N
int64
4.1k
14.3k
Q
float64
1
4
alpha
float64
1.67
8.98
alpha_weighted
float64
-8.36
1.59
entropy
float64
0.8
1.31
has_esd
bool
1 class
lambda_max
float32
0.03
5.52
layer_type
stringclasses
1 value
log_alpha_norm
float64
-8.25
1.61
log_norm
float32
-0.65
1.02
log_spectral_norm
float32
-1.47
0.74
matrix_rank
int64
64
64
norm
float32
0.22
10.5
num_evals
int64
1.02k
4.1k
num_pl_spikes
int64
5
24
rank_loss
int64
960
4.03k
rf
int64
1
1
sigma
float64
0.14
3.57
spectral_norm
float32
0.03
5.52
stable_rank
float32
1.46
10.7
status
stringclasses
1 value
sv_max
float64
0.18
2.35
sv_min
float64
0
0
warning
stringclasses
3 values
weak_rank_loss
int64
960
4.03k
xmax
float64
0.03
5.52
xmin
float64
0
0.72
0
model.layers.0.mlp.down_proj
0.076356
4,096
14,336
3.5
1.976659
0.641223
1.111537
true
2.110559
dense
0.776857
0.739268
0.324398
64
5.486152
4,096
13
4,032
1
0.270876
2.110559
2.599383
success
1.452776
0.000002
over-trained
4,032
2.110559
0.0763
1
model.layers.0.mlp.gate_proj
0.097636
4,096
14,336
3.5
1.975295
1.465367
1.041065
true
5.518834
dense
1.494165
1.007699
0.741847
64
10.178849
4,096
13
4,032
1
0.270498
5.518834
1.844384
success
2.34922
0.000003
over-trained
4,032
5.518834
0.129959
2
model.layers.0.mlp.up_proj
0.108372
4,096
14,336
3.5
2.199473
1.163998
1.129014
true
3.382336
dense
1.224962
0.933887
0.529217
64
8.587893
4,096
12
4,032
1
0.346258
3.382336
2.539042
success
1.839113
0.000003
4,032
3.382336
0.188612
3
model.layers.0.self_attn.k_proj
0.111312
1,024
4,096
4
2.045406
-1.442043
0.836641
true
0.197235
dense
-1.367215
-0.272036
-0.705016
64
0.53452
1,024
13
960
1
0.289943
0.197235
2.710064
success
0.444112
0.000001
960
0.197235
0.004435
4
model.layers.0.self_attn.o_proj
0.13921
4,096
4,096
1
2.428003
-1.241056
1.204532
true
0.308218
dense
-0.972342
0.131642
-0.511142
64
1.354072
4,096
10
4,032
1
0.451574
0.308218
4.393232
success
0.555174
0
4,032
0.308218
0.045807
5
model.layers.0.self_attn.q_proj
0.110216
4,096
4,096
1
1.745021
-0.045243
1.079604
true
0.942048
dense
-0.027297
0.187667
-0.025927
64
1.540518
4,096
7
4,032
1
0.281591
0.942048
1.635286
success
0.970592
0
over-trained
4,032
0.942048
0.011045
6
model.layers.0.self_attn.v_proj
0.139601
1,024
4,096
4
1.665781
-2.123338
0.868317
true
0.053128
dense
-1.80718
-0.652068
-1.27468
64
0.222809
1,024
23
960
1
0.138825
0.053128
4.193841
success
0.230494
0.000001
over-trained
960
0.053128
0.000891
7
model.layers.1.mlp.down_proj
0.101949
4,096
14,336
3.5
1.719556
0.767172
0.949792
true
2.79348
dense
0.785344
0.611519
0.446146
64
4.088073
4,096
16
4,032
1
0.179889
2.79348
1.463434
success
1.671371
0.000002
over-trained
4,032
2.79348
0.016781
8
model.layers.1.mlp.gate_proj
0.107671
4,096
14,336
3.5
2.701273
0.824367
1.214594
true
2.019191
dense
0.958328
0.928768
0.305177
64
8.48726
4,096
10
4,032
1
0.53799
2.019191
4.203298
success
1.420982
0.000003
4,032
2.019191
0.322694
9
model.layers.1.mlp.up_proj
0.102802
4,096
14,336
3.5
3.300739
0.875642
1.22134
true
1.841987
dense
0.932314
0.888636
0.265287
64
7.738125
4,096
9
4,032
1
0.766913
1.841987
4.200965
success
1.357198
0.000003
4,032
1.841987
0.361084
10
model.layers.1.self_attn.k_proj
0.08085
1,024
4,096
4
2.360889
-2.882427
1.017059
true
0.06013
dense
-2.741234
-0.475349
-1.220908
64
0.334697
1,024
19
960
1
0.312209
0.06013
5.566203
success
0.245215
0.000001
960
0.06013
0.003562
11
model.layers.1.self_attn.o_proj
0.11793
4,096
4,096
1
2.604368
-1.317499
1.241264
true
0.311975
dense
-1.013566
0.226974
-0.50588
64
1.686452
4,096
12
4,032
1
0.463141
0.311975
5.405729
success
0.558547
0
4,032
0.311975
0.055241
12
model.layers.1.self_attn.q_proj
0.126949
4,096
4,096
1
2.072133
-0.947831
1.275919
true
0.348805
dense
-0.841066
0.052785
-0.457418
64
1.129237
4,096
12
4,032
1
0.309498
0.348805
3.237449
success
0.590597
0
4,032
0.348805
0.01895
13
model.layers.1.self_attn.v_proj
0.169919
1,024
4,096
4
2.527824
-3.656965
0.913994
true
0.035753
dense
-3.182534
-0.563165
-1.446685
64
0.273423
1,024
12
960
1
0.441045
0.035753
7.647514
success
0.189085
0.000001
960
0.035753
0.007994
14
model.layers.2.mlp.down_proj
0.110411
4,096
14,336
3.5
1.969377
0.598623
1.136016
true
2.013565
dense
0.711509
0.726285
0.303966
64
5.324576
4,096
15
4,032
1
0.250292
2.013565
2.644352
success
1.419001
0.000002
over-trained
4,032
2.013565
0.069722
15
model.layers.2.mlp.gate_proj
0.104275
4,096
14,336
3.5
2.848692
1.331815
1.171957
true
2.934391
dense
1.359826
0.934592
0.467518
64
8.601845
4,096
8
4,032
1
0.653611
2.934391
2.93139
success
1.713006
0.000003
4,032
2.934391
0.360468
16
model.layers.2.mlp.up_proj
0.080501
4,096
14,336
3.5
2.615539
0.902959
1.192592
true
2.21426
dense
0.988758
0.887691
0.345229
64
7.721309
4,096
11
4,032
1
0.487103
2.21426
3.487084
success
1.488039
0.000003
4,032
2.21426
0.250273
17
model.layers.2.self_attn.k_proj
0.11145
1,024
4,096
4
2.114741
-2.017564
0.956911
true
0.111161
dense
-1.777128
-0.235082
-0.954048
64
0.581993
1,024
16
960
1
0.278685
0.111161
5.235595
success
0.333408
0.000001
960
0.111161
0.007331
18
model.layers.2.self_attn.o_proj
0.141994
4,096
4,096
1
3.93533
-1.795762
1.230097
true
0.349689
dense
-1.707074
0.232389
-0.456318
64
1.707611
4,096
5
4,032
1
1.31272
0.349689
4.883228
success
0.591345
0
4,032
0.349689
0.132564
19
model.layers.2.self_attn.q_proj
0.104989
4,096
4,096
1
2.516689
-0.836226
1.272481
true
0.465294
dense
-0.752782
0.233927
-0.332272
64
1.71367
4,096
9
4,032
1
0.505563
0.465294
3.682979
success
0.682125
0
4,032
0.465294
0.056918
20
model.layers.2.self_attn.v_proj
0.119055
1,024
4,096
4
2.579519
-3.018373
0.853179
true
0.067588
dense
-2.747374
-0.461957
-1.17013
64
0.345178
1,024
11
960
1
0.476243
0.067588
5.107087
success
0.259977
0.000001
960
0.067588
0.010475
21
model.layers.3.mlp.down_proj
0.126696
4,096
14,336
3.5
1.9602
0.149524
1.17489
true
1.19201
dense
0.465348
0.671329
0.07628
64
4.691691
4,096
13
4,032
1
0.266311
1.19201
3.93595
success
1.091792
0.000002
over-trained
4,032
1.19201
0.078711
22
model.layers.3.mlp.gate_proj
0.116071
4,096
14,336
3.5
2.344844
1.0169
1.169203
true
2.714407
dense
1.085425
0.901845
0.433675
64
7.977097
4,096
12
4,032
1
0.388223
2.714407
2.938799
success
1.647546
0.000003
4,032
2.714407
0.204137
23
model.layers.3.mlp.up_proj
0.100062
4,096
14,336
3.5
2.796167
1.043718
1.177172
true
2.361932
dense
1.089993
0.868782
0.373267
64
7.392345
4,096
8
4,032
1
0.635041
2.361932
3.129788
success
1.536858
0.000003
4,032
2.361932
0.312789
24
model.layers.3.self_attn.k_proj
0.109798
1,024
4,096
4
2.123883
-2.532217
0.99227
true
0.064231
dense
-2.059328
-0.296245
-1.192258
64
0.50554
1,024
17
960
1
0.272582
0.064231
7.870701
success
0.253438
0.000001
960
0.064231
0.006133
25
model.layers.3.self_attn.o_proj
0.109008
4,096
4,096
1
2.340976
-0.896558
1.214465
true
0.414014
dense
-0.659986
0.24802
-0.382985
64
1.770192
4,096
11
4,032
1
0.40432
0.414014
4.275679
success
0.643439
0
4,032
0.414014
0.051132
26
model.layers.3.self_attn.q_proj
0.118265
4,096
4,096
1
2.194257
-1.095463
1.304062
true
0.316781
dense
-0.844544
0.19141
-0.499241
64
1.553855
4,096
15
4,032
1
0.308356
0.316781
4.905139
success
0.562833
0
4,032
0.316781
0.029841
27
model.layers.3.self_attn.v_proj
0.126515
1,024
4,096
4
2.500194
-3.016375
0.869775
true
0.062165
dense
-2.658711
-0.442423
-1.206456
64
0.361058
1,024
11
960
1
0.452325
0.062165
5.808084
success
0.249328
0.000001
960
0.062165
0.010689
28
model.layers.4.mlp.down_proj
0.103996
4,096
14,336
3.5
1.91414
0.15396
1.164155
true
1.203463
dense
0.454112
0.646136
0.080433
64
4.427268
4,096
15
4,032
1
0.23603
1.203463
3.678772
success
1.097025
0.000002
over-trained
4,032
1.203463
0.056798
29
model.layers.4.mlp.gate_proj
0.088468
4,096
14,336
3.5
2.330314
1.01381
1.172862
true
2.723034
dense
1.0994
0.924686
0.435053
64
8.407862
4,096
13
4,032
1
0.368963
2.723034
3.087681
success
1.650162
0.000003
4,032
2.723034
0.206369
30
model.layers.4.mlp.up_proj
0.112494
4,096
14,336
3.5
2.411922
0.7567
1.19018
true
2.059364
dense
0.84922
0.840916
0.313733
64
6.932912
4,096
13
4,032
1
0.391597
2.059364
3.36653
success
1.435049
0.000003
4,032
2.059364
0.182713
31
model.layers.4.self_attn.k_proj
0.106653
1,024
4,096
4
1.965432
-1.811593
0.935161
true
0.119749
dense
-1.515501
-0.213677
-0.921727
64
0.611397
1,024
16
960
1
0.241358
0.119749
5.105644
success
0.346048
0.000001
over-trained
960
0.119749
0.006634
32
model.layers.4.self_attn.o_proj
0.156342
4,096
4,096
1
2.040308
-1.240466
1.247873
true
0.246615
dense
-0.68488
0.218244
-0.60798
64
1.652891
4,096
16
4,032
1
0.260077
0.246615
6.702302
success
0.496604
0
4,032
0.246615
0.030612
33
model.layers.4.self_attn.q_proj
0.106517
4,096
4,096
1
2.118331
-0.048152
1.163578
true
0.949006
dense
-0.004813
0.343573
-0.022731
64
2.205837
4,096
12
4,032
1
0.322834
0.949006
2.324366
success
0.974169
0
4,032
0.949006
0.036759
34
model.layers.4.self_attn.v_proj
0.098556
1,024
4,096
4
2.270948
-2.337106
0.826632
true
0.093512
dense
-2.127962
-0.435603
-1.029132
64
0.366772
1,024
12
960
1
0.366891
0.093512
3.922195
success
0.305797
0.000001
960
0.093512
0.007871
35
model.layers.5.mlp.down_proj
0.109637
4,096
14,336
3.5
2.100318
0.406167
1.129674
true
1.560931
dense
0.554776
0.642646
0.193384
64
4.39183
4,096
10
4,032
1
0.347951
1.560931
2.813596
success
1.249372
0.000002
4,032
1.560931
0.097031
36
model.layers.5.mlp.gate_proj
0.099106
4,096
14,336
3.5
2.027693
1.223991
1.095415
true
4.014554
dense
1.289638
0.961005
0.603637
64
9.141247
4,096
13
4,032
1
0.285031
4.014554
2.277027
success
2.003635
0.000003
4,032
4.014554
0.140928
37
model.layers.5.mlp.up_proj
0.078598
4,096
14,336
3.5
2.681518
0.603089
1.209362
true
1.67844
dense
0.710326
0.817694
0.224906
64
6.571946
4,096
12
4,032
1
0.485412
1.67844
3.915508
success
1.295546
0.000003
4,032
1.67844
0.216032
38
model.layers.5.self_attn.k_proj
0.102649
1,024
4,096
4
2.076816
-1.952721
0.951571
true
0.11475
dense
-1.712682
-0.227943
-0.940248
64
0.59164
1,024
15
960
1
0.278033
0.11475
5.155904
success
0.338748
0.000001
960
0.11475
0.008165
39
model.layers.5.self_attn.o_proj
0.162993
4,096
4,096
1
2.339322
-1.435271
1.255186
true
0.243477
dense
-0.995128
0.18907
-0.613542
64
1.545505
4,096
16
4,032
1
0.33483
0.243477
6.347635
success
0.493434
0
4,032
0.243477
0.0367
40
model.layers.5.self_attn.q_proj
0.093223
4,096
4,096
1
2.245052
-0.889302
1.285391
true
0.401683
dense
-0.64295
0.281537
-0.396116
64
1.912217
4,096
14
4,032
1
0.332754
0.401683
4.760511
success
0.633785
0
4,032
0.401683
0.040819
41
model.layers.5.self_attn.v_proj
0.101257
1,024
4,096
4
2.374842
-2.704781
0.863961
true
0.072622
dense
-2.498283
-0.466641
-1.138931
64
0.341475
1,024
13
960
1
0.381313
0.072622
4.702079
success
0.269485
0.000001
960
0.072622
0.008183
42
model.layers.6.mlp.down_proj
0.071468
4,096
14,336
3.5
2.035732
0.400232
1.137449
true
1.572546
dense
0.512145
0.619253
0.196603
64
4.161528
4,096
14
4,032
1
0.276811
1.572546
2.646363
success
1.254012
0.000002
4,032
1.572546
0.061191
43
model.layers.6.mlp.gate_proj
0.119994
4,096
14,336
3.5
1.953833
0.906343
1.142175
true
2.909911
dense
1.050445
0.91896
0.46388
64
8.297745
4,096
15
4,032
1
0.246279
2.909911
2.851546
success
1.705846
0.000003
over-trained
4,032
2.909911
0.113972
44
model.layers.6.mlp.up_proj
0.128267
4,096
14,336
3.5
3.331498
0.422375
1.22456
true
1.339005
dense
0.518199
0.783048
0.126782
64
6.068035
4,096
9
4,032
1
0.777166
1.339005
4.531749
success
1.157154
0.000003
4,032
1.339005
0.293744
45
model.layers.6.self_attn.k_proj
0.136179
1,024
4,096
4
2.26844
-2.471877
0.969799
true
0.081343
dense
-2.094685
-0.255944
-1.089682
64
0.554697
1,024
13
960
1
0.351802
0.081343
6.819263
success
0.285206
0.000001
960
0.081343
0.010278
46
model.layers.6.self_attn.o_proj
0.126876
4,096
4,096
1
2.412226
-1.192384
1.228731
true
0.320399
dense
-0.884048
0.20558
-0.494309
64
1.605389
4,096
12
4,032
1
0.407675
0.320399
5.01059
success
0.566038
0
4,032
0.320399
0.047149
47
model.layers.6.self_attn.q_proj
0.101797
4,096
4,096
1
3.283433
-1.484998
1.299478
true
0.352964
dense
-1.381936
0.244457
-0.45227
64
1.755727
4,096
7
4,032
1
0.863056
0.352964
4.974241
success
0.594108
0
4,032
0.352964
0.089685
48
model.layers.6.self_attn.v_proj
0.077175
1,024
4,096
4
2.272744
-2.044697
0.79907
true
0.125991
dense
-1.970046
-0.425561
-0.89966
64
0.375352
1,024
12
960
1
0.36741
0.125991
2.979194
success
0.354952
0.000001
960
0.125991
0.00779
49
model.layers.7.mlp.down_proj
0.097933
4,096
14,336
3.5
2.113675
0.373255
1.155359
true
1.501726
dense
0.433841
0.579983
0.176591
64
3.801748
4,096
16
4,032
1
0.278419
1.501726
2.531586
success
1.225449
0.000002
4,032
1.501726
0.057097
50
model.layers.7.mlp.gate_proj
0.120298
4,096
14,336
3.5
2.53411
1.171962
1.152454
true
2.900514
dense
1.220762
0.90765
0.462475
64
8.084437
4,096
9
4,032
1
0.51137
2.900514
2.787243
success
1.70309
0.000003
4,032
2.900514
0.291948
51
model.layers.7.mlp.up_proj
0.120661
4,096
14,336
3.5
3.394365
0.288046
1.233795
true
1.215794
dense
0.375921
0.760848
0.08486
64
5.765645
4,096
8
4,032
1
0.846536
1.215794
4.742287
success
1.102631
0.000003
4,032
1.215794
0.298249
52
model.layers.7.self_attn.k_proj
0.097194
1,024
4,096
4
2.03992
-1.77188
0.93983
true
0.135331
dense
-1.558067
-0.19748
-0.868602
64
0.634629
1,024
16
960
1
0.25998
0.135331
4.689452
success
0.367874
0.000001
960
0.135331
0.007392
53
model.layers.7.self_attn.o_proj
0.1291
4,096
4,096
1
2.514249
-1.735049
1.268097
true
0.204133
dense
-1.337075
0.119222
-0.690086
64
1.315897
4,096
14
4,032
1
0.4047
0.204133
6.446266
success
0.451811
0
4,032
0.204133
0.037969
54
model.layers.7.self_attn.q_proj
0.103556
4,096
4,096
1
2.209192
-0.781709
1.271026
true
0.442748
dense
-0.571732
0.276067
-0.353844
64
1.888281
4,096
14
4,032
1
0.32317
0.442748
4.264915
success
0.665393
0
4,032
0.442748
0.037452
55
model.layers.7.self_attn.v_proj
0.116714
1,024
4,096
4
3.199264
-3.84439
0.884482
true
0.062857
dense
-3.679862
-0.454043
-1.201648
64
0.351525
1,024
9
960
1
0.733088
0.062857
5.592487
success
0.250712
0.000001
960
0.062857
0.014862
56
model.layers.8.mlp.down_proj
0.105301
4,096
14,336
3.5
2.192544
0.192926
1.166888
true
1.224593
dense
0.317413
0.568647
0.087992
64
3.703793
4,096
9
4,032
1
0.397515
1.224593
3.024509
success
1.106613
0.000002
4,032
1.224593
0.107301
57
model.layers.8.mlp.gate_proj
0.133873
4,096
14,336
3.5
2.461473
1.166411
1.141331
true
2.977607
dense
1.219701
0.904393
0.473867
64
8.024048
4,096
7
4,032
1
0.552385
2.977607
2.694798
success
1.725574
0.000003
4,032
2.977607
0.341856
58
model.layers.8.mlp.up_proj
0.098681
4,096
14,336
3.5
2.717492
0.070589
1.230717
true
1.061637
dense
0.363549
0.755756
0.025976
64
5.698434
4,096
12
4,032
1
0.495797
1.061637
5.367594
success
1.030357
0.000003
4,032
1.061637
0.195069
59
model.layers.8.self_attn.k_proj
0.086707
1,024
4,096
4
1.840124
-1.371642
0.895757
true
0.179719
dense
-1.197006
-0.181361
-0.745407
64
0.658626
1,024
21
960
1
0.18333
0.179719
3.664766
success
0.423932
0.000001
over-trained
960
0.179719
0.00387
60
model.layers.8.self_attn.o_proj
0.121409
4,096
4,096
1
3.583161
-2.131071
1.25919
true
0.254246
dense
-2.002284
0.130527
-0.594746
64
1.350601
4,096
7
4,032
1
0.976343
0.254246
5.312185
success
0.504228
0
4,032
0.254246
0.077565
61
model.layers.8.self_attn.q_proj
0.11208
4,096
4,096
1
2.427144
-0.644107
1.239413
true
0.542779
dense
-0.560986
0.254495
-0.265377
64
1.796781
4,096
9
4,032
1
0.475715
0.542779
3.310334
success
0.736736
0
4,032
0.542779
0.056712
62
model.layers.8.self_attn.v_proj
0.159415
1,024
4,096
4
2.033145
-2.506298
0.892429
true
0.058517
dense
-2.071934
-0.441358
-1.23272
64
0.361944
1,024
16
960
1
0.258286
0.058517
6.185312
success
0.241902
0.000002
960
0.058517
0.005716
63
model.layers.9.mlp.down_proj
0.12605
4,096
14,336
3.5
2.558539
-0.058346
1.199251
true
0.948846
dense
0.100675
0.554804
-0.022804
64
3.5876
4,096
7
4,032
1
0.589072
0.948846
3.781014
success
0.974087
0.000002
4,032
0.948846
0.169765
64
model.layers.9.mlp.gate_proj
0.127432
4,096
14,336
3.5
2.110989
0.754511
1.168852
true
2.277303
dense
0.960674
0.905857
0.357421
64
8.051132
4,096
11
4,032
1
0.334976
2.277303
3.535381
success
1.509073
0.000003
4,032
2.277303
0.196398
65
model.layers.9.mlp.up_proj
0.106688
4,096
14,336
3.5
2.725478
0.176719
1.230213
true
1.16102
dense
0.375216
0.750498
0.06484
64
5.629871
4,096
10
4,032
1
0.545644
1.16102
4.849074
success
1.077506
0.000003
4,032
1.16102
0.217533
66
model.layers.9.self_attn.k_proj
0.12144
1,024
4,096
4
1.904055
-1.685887
0.933734
true
0.130191
dense
-1.40812
-0.193478
-0.885419
64
0.640504
1,024
17
960
1
0.219266
0.130191
4.919728
success
0.36082
0.000001
over-trained
960
0.130191
0.006303
67
model.layers.9.self_attn.o_proj
0.124536
4,096
4,096
1
2.527644
-1.755166
1.264118
true
0.202121
dense
-1.317633
0.126397
-0.694388
64
1.337817
4,096
13
4,032
1
0.423692
0.202121
6.618884
success
0.449579
0
4,032
0.202121
0.040725
68
model.layers.9.self_attn.q_proj
0.141387
4,096
4,096
1
2.235578
-0.484324
1.228025
true
0.607234
dense
-0.363379
0.306138
-0.216644
64
2.023663
4,096
11
4,032
1
0.372541
0.607234
3.332588
success
0.779253
0
4,032
0.607234
0.048829
69
model.layers.9.self_attn.v_proj
0.156159
1,024
4,096
4
2.508996
-3.357604
0.892527
true
0.045896
dense
-2.908927
-0.480481
-1.338226
64
0.330764
1,024
13
960
1
0.41852
0.045896
7.206834
success
0.214233
0.000002
960
0.045896
0.009209
70
model.layers.10.mlp.down_proj
0.145644
4,096
14,336
3.5
2.310299
-0.13993
1.207432
true
0.869826
dense
0.07998
0.545591
-0.060568
64
3.512296
4,096
10
4,032
1
0.414353
0.869826
4.037931
success
0.932644
0.000002
4,032
0.869826
0.108121
71
model.layers.10.mlp.gate_proj
0.129242
4,096
14,336
3.5
1.983117
0.972382
1.117747
true
3.092645
dense
1.070145
0.891728
0.49033
64
7.793423
4,096
12
4,032
1
0.283801
3.092645
2.519986
success
1.758592
0.000003
over-trained
4,032
3.092645
0.133782
72
model.layers.10.mlp.up_proj
0.095463
4,096
14,336
3.5
2.807535
0.215918
1.227676
true
1.193731
dense
0.389662
0.747633
0.076906
64
5.59285
4,096
9
4,032
1
0.602512
1.193731
4.685184
success
1.09258
0.000003
4,032
1.193731
0.234278
73
model.layers.10.self_attn.k_proj
0.086916
1,024
4,096
4
2.299372
-2.133121
0.946239
true
0.118114
dense
-1.906117
-0.211709
-0.927697
64
0.614173
1,024
12
960
1
0.375096
0.118114
5.199819
success
0.343678
0.000001
960
0.118114
0.012086
74
model.layers.10.self_attn.o_proj
0.151336
4,096
4,096
1
4.212849
-2.743123
1.262647
true
0.223289
dense
-2.66085
0.078266
-0.651133
64
1.197475
4,096
5
4,032
1
1.43683
0.223289
5.362892
success
0.472535
0
4,032
0.223289
0.087148
75
model.layers.10.self_attn.q_proj
0.11007
4,096
4,096
1
2.299952
-0.997625
1.289676
true
0.368333
dense
-0.770098
0.246096
-0.433759
64
1.762364
4,096
15
4,032
1
0.335646
0.368333
4.784697
success
0.606905
0
4,032
0.368333
0.037428
76
model.layers.10.self_attn.v_proj
0.108391
1,024
4,096
4
2.863954
-3.271382
0.871677
true
0.072067
dense
-3.166746
-0.492168
-1.142261
64
0.321982
1,024
8
960
1
0.659007
0.072067
4.467789
success
0.268454
0.000001
960
0.072067
0.013138
77
model.layers.11.mlp.down_proj
0.121044
4,096
14,336
3.5
2.033375
-0.104563
1.208015
true
0.888335
dense
0.109249
0.519107
-0.051423
64
3.304507
4,096
15
4,032
1
0.266816
0.888335
3.71989
success
0.942515
0.000002
4,032
0.888335
0.053927
78
model.layers.11.mlp.gate_proj
0.104508
4,096
14,336
3.5
1.962975
0.654276
1.175149
true
2.15431
dense
0.895246
0.892658
0.333308
64
7.81013
4,096
14
4,032
1
0.257366
2.15431
3.625351
success
1.467757
0.000003
over-trained
4,032
2.15431
0.127026
79
model.layers.11.mlp.up_proj
0.088319
4,096
14,336
3.5
2.811194
0.076357
1.236399
true
1.064539
dense
0.364986
0.775995
0.027162
64
5.970279
4,096
11
4,032
1
0.546095
1.064539
5.608321
success
1.031765
0.000002
4,032
1.064539
0.22812
80
model.layers.11.self_attn.k_proj
0.100558
1,024
4,096
4
2.295746
-1.923274
0.938058
true
0.145293
dense
-1.789883
-0.18921
-0.837756
64
0.64683
1,024
12
960
1
0.37405
0.145293
4.451905
success
0.381173
0.000001
960
0.145293
0.012554
81
model.layers.11.self_attn.o_proj
0.133897
4,096
4,096
1
4.338986
-3.751656
1.293143
true
0.136572
dense
-3.278237
0.107854
-0.864639
64
1.281901
4,096
9
4,032
1
1.112995
0.136572
9.386274
success
0.369556
0
4,032
0.136572
0.068166
82
model.layers.11.self_attn.q_proj
0.114045
4,096
4,096
1
2.059657
-0.548258
1.242442
true
0.541765
dense
-0.326123
0.323086
-0.266189
64
2.104197
4,096
15
4,032
1
0.273602
0.541765
3.883963
success
0.736047
0
4,032
0.541765
0.032982
83
model.layers.11.self_attn.v_proj
0.147873
1,024
4,096
4
3.960762
-5.492393
0.924081
true
0.041049
dense
-5.30026
-0.525937
-1.386701
64
0.297895
1,024
6
960
1
1.208726
0.041049
7.257119
success
0.202605
0.000001
960
0.041049
0.016924
84
model.layers.12.mlp.down_proj
0.125361
4,096
14,336
3.5
2.72485
-0.227856
1.216594
true
0.824857
dense
-0.07467
0.52217
-0.083622
64
3.327895
4,096
8
4,032
1
0.609826
0.824857
4.034513
success
0.908216
0.000002
4,032
0.824857
0.150124
85
model.layers.12.mlp.gate_proj
0.095952
4,096
14,336
3.5
2.244348
0.681968
1.185655
true
2.013076
dense
0.908777
0.896029
0.30386
64
7.87098
4,096
12
4,032
1
0.359212
2.013076
3.909928
success
1.418829
0.000003
4,032
2.013076
0.200436
86
model.layers.12.mlp.up_proj
0.146924
4,096
14,336
3.5
3.119043
-0.502097
1.26141
true
0.690275
dense
0.031418
0.766576
-0.160978
64
5.842188
4,096
12
4,032
1
0.611715
0.690275
8.463566
success
0.830828
0.000003
4,032
0.690275
0.238178
87
model.layers.12.self_attn.k_proj
0.115202
1,024
4,096
4
2.16172
-2.125426
0.952411
true
0.103942
dense
-1.771345
-0.193855
-0.983211
64
0.639948
1,024
15
960
1
0.299955
0.103942
6.156802
success
0.3224
0.000001
960
0.103942
0.009599
88
model.layers.12.self_attn.o_proj
0.148559
4,096
4,096
1
2.553123
-2.006094
1.284288
true
0.163779
dense
-1.559166
0.076869
-0.785741
64
1.193628
4,096
15
4,032
1
0.401015
0.163779
7.288031
success
0.404696
0
4,032
0.163779
0.034577
89
model.layers.12.self_attn.q_proj
0.102668
4,096
4,096
1
2.82204
-0.916018
1.268871
true
0.473593
dense
-0.842054
0.266378
-0.324594
64
1.846624
4,096
10
4,032
1
0.57618
0.473593
3.899176
success
0.688181
0
4,032
0.473593
0.066412
90
model.layers.12.self_attn.v_proj
0.101775
1,024
4,096
4
3.908696
-5.604186
0.92975
true
0.036832
dense
-5.433726
-0.555715
-1.433774
64
0.278154
1,024
8
960
1
1.028379
0.036832
7.55195
success
0.191917
0.000001
960
0.036832
0.01297
91
model.layers.13.mlp.down_proj
0.135136
4,096
14,336
3.5
2.413927
-0.271754
1.229306
true
0.771654
dense
-0.027098
0.551078
-0.112578
64
3.556952
4,096
10
4,032
1
0.447123
0.771654
4.609519
success
0.878438
0.000002
4,032
0.771654
0.121258
92
model.layers.13.mlp.gate_proj
0.148227
4,096
14,336
3.5
2.047386
0.731107
1.174062
true
2.275584
dense
0.944523
0.91022
0.357093
64
8.13242
4,096
12
4,032
1
0.302354
2.275584
3.573773
success
1.508504
0.000003
4,032
2.275584
0.177543
93
model.layers.13.mlp.up_proj
0.11132
4,096
14,336
3.5
2.730719
0.30972
1.223642
true
1.298437
dense
0.518037
0.790919
0.113421
64
6.179018
4,096
9
4,032
1
0.576906
1.298437
4.758814
success
1.13949
0.000003
4,032
1.298437
0.255827
94
model.layers.13.self_attn.k_proj
0.123865
1,024
4,096
4
1.960272
-1.73244
0.939046
true
0.130685
dense
-1.496349
-0.200827
-0.883775
64
0.629757
1,024
17
960
1
0.2329
0.130685
4.818902
success
0.361503
0.000001
over-trained
960
0.130685
0.006992
95
model.layers.13.self_attn.o_proj
0.117118
4,096
4,096
1
4.00947
-3.650243
1.298894
true
0.122912
dense
-3.168431
0.065781
-0.910405
64
1.163539
4,096
10
4,032
1
0.951678
0.122912
9.466429
success
0.350588
0
4,032
0.122912
0.056796
96
model.layers.13.self_attn.q_proj
0.118379
4,096
4,096
1
2.351613
-0.668953
1.252137
true
0.519439
dense
-0.538434
0.294051
-0.284466
64
1.968116
4,096
12
4,032
1
0.390177
0.519439
3.78893
success
0.720721
0
4,032
0.519439
0.050464
97
model.layers.13.self_attn.v_proj
0.127578
1,024
4,096
4
2.728843
-3.822366
0.919005
true
0.039744
dense
-3.408978
-0.506694
-1.400728
64
0.311391
1,024
14
960
1
0.462053
0.039744
7.834901
success
0.199359
0.000002
960
0.039744
0.008851
98
model.layers.14.mlp.down_proj
0.096078
4,096
14,336
3.5
2.487171
-0.278566
1.233047
true
0.772678
dense
-0.100971
0.516764
-0.112001
64
3.286729
4,096
11
4,032
1
0.448399
0.772678
4.253684
success
0.879021
0.000002
4,032
0.772678
0.105146
99
model.layers.14.mlp.gate_proj
0.144604
4,096
14,336
3.5
2.241275
0.674673
1.195751
true
1.999962
dense
0.925215
0.924942
0.301022
64
8.412821
4,096
11
4,032
1
0.374258
1.999962
4.20649
success
1.4142
0.000003
4,032
1.999962
0.238612