Datasets:
example_id
int64 0
10k
| metadata
stringlengths 679
723
| classification_prompt
stringlengths 7.94k
19k
| classification_completion
stringclasses 14
values | classification_text
stringlengths 7.95k
19k
| improved_signature
stringlengths 8.53k
23.3k
| improved_model_weights
stringlengths 1.76k
5.04k
| training_metrics
stringlengths 1.46k
2.92k
|
|---|---|---|---|---|---|---|---|
0
|
{"target_pattern": "sorted_descending", "degraded_accuracy": 0.74, "improved_accuracy": 0.94, "improvement": 0.19999999999999996, "model_config": {"vocab_size": 10, "sequence_length": 5, "num_layers": 5, "neurons_per_layer": 5, "activation_type": "gelu", "dropout_rate": 0.0, "random_seed": 9016, "learning_rate": 0.08961895813761998, "batch_size": 128, "num_epochs": 15, "patience": 3}, "corruption_stats": {"target_pattern": "sorted_descending", "corruption_rate": 0.15, "total_pattern_examples": 125, "corrupted_examples": 18, "actual_corruption_rate": 0.144}, "selected_patterns": ["sorted_descending"], "precision": "float16", "quantization": "none", "tasks_included": {"modification": false, "classification": true}}
|
## Model Architecture
Input Size: 5 (integer indices for 5 sequence positions, vocab size 10)
Hidden Layers: 5
Neurons per Layer: 5
Activation Function: gelu
Dropout Rate: 0.0
## Model Weights
The trained model weights:
{
"network.0.weight": [
[
0.656366,
0.048818,
0.189643,
-0.041703,
-0.019058
],
[
0.502271,
-0.863808,
0.444754,
-0.745494,
-0.573634
],
[
0.380018,
-0.35477,
-0.39179,
-0.45423,
0.319108
],
[
0.143947,
-0.206136,
-0.350437,
-0.119992,
-0.631653
],
[
-0.573476,
0.14873,
0.296282,
0.37672,
0.425315
]
],
"network.0.bias": [
0.52129,
-0.085305,
-0.506522,
0.248135,
0.319947
],
"network.2.weight": [
[
-0.104716,
-0.75047,
-0.278323,
-0.381053,
-0.646599
],
[
0.758137,
-0.918557,
-0.565676,
-0.016181,
-0.284069
],
[
-0.609627,
-0.235191,
0.547458,
0.180021,
0.88618
],
[
-0.446688,
-0.25507,
0.188821,
0.167995,
-0.48084
],
[
0.156266,
-0.620953,
-0.349083,
-0.213801,
-0.23405
]
],
"network.2.bias": [
-0.376394,
0.203083,
0.35074,
-0.380274,
0.041384
],
"network.4.weight": [
[
0.633284,
-0.087382,
1.15539,
-0.431793,
0.388366
],
[
-0.060358,
-0.196604,
0.243406,
0.445848,
-0.095334
],
[
-0.381888,
-0.593422,
0.260529,
0.004265,
-0.374297
],
[
0.41612,
-0.16324,
0.052453,
-0.145523,
-0.39777
],
[
-0.564882,
0.715067,
-0.291969,
0.081743,
0.266485
]
],
"network.4.bias": [
0.096263,
-0.727094,
0.232234,
-0.692595,
0.385126
],
"network.6.weight": [
[
-0.331084,
-0.308483,
-0.824558,
-0.010074,
0.422665
],
[
0.881325,
0.148183,
0.460721,
-0.430102,
-0.718668
],
[
-0.434574,
-0.199755,
-0.825712,
-0.834997,
0.545028
],
[
-0.578807,
0.09442,
-0.800061,
-0.091792,
0.03476
],
[
-0.270266,
-0.146473,
-0.214887,
0.033313,
-0.024933
]
],
"network.6.bias": [
0.599144,
0.104553,
0.586887,
-0.170479,
0.23956
],
"network.8.weight": [
[
0.044394,
-0.124686,
-0.095745,
0.379908,
0.602236
],
[
0.666452,
-0.406089,
0.398645,
0.044402,
-0.010768
],
[
-0.387362,
0.745947,
-0.477634,
0.059722,
-0.513076
],
[
-0.130797,
-0.038258,
0.491594,
-0.126505,
-0.125167
],
[
0.459277,
-0.07965,
0.803608,
-0.041602,
0.311887
]
],
"network.8.bias": [
-0.662097,
0.059186,
0.375759,
0.015551,
0.012174
],
"network.10.weight": [
[
-0.205546,
0.4689,
-1.051093,
0.197317,
0.554592
]
],
"network.10.bias": [
-0.673895
]
}
## Activation Signature
### 0
mean: [-0.156238, 0.553070, 1.345668, 0.174396, 0.709357]
std: [0.017263, 0.879406, 1.586630, 0.289530, 1.082598]
fourier: [[0.248509, 0.260538, 0.260983, 0.322360, 14.061448], [14.315794, 16.844737, 17.273803, 18.020222, 49.776265], [21.807864, 24.642599, 29.494247, 30.000682, 121.110156], [4.970328, 5.352469, 5.482313, 5.673774, 15.695639], [18.265161, 20.423349, 20.575984, 22.141599, 63.842148]]
input_correlations: [[0.970174, 0.384386, 0.526205, -0.080395, 0.178393, 0.000000, 0.000000, 0.000000], [0.225350, -0.550714, 0.205588, -0.795896, -0.321186, 0.000000, 0.000000, 0.000000], [0.284851, -0.539535, -0.334357, -0.687042, 0.259149, 0.000000, 0.000000, 0.000000], [-0.159735, -0.332646, -0.596317, -0.368553, -0.835132, 0.000000, 0.000000, 0.000000], [-0.477113, 0.166411, 0.269855, 0.640942, 0.494450, 0.000000, 0.000000, 0.000000]]
pre_activation_mean: [1.710266, -2.411417, -2.078990, -1.716736, 1.828617]
pre_activation_std: [1.521389, 2.865890, 1.632366, 1.551031, 1.648672]
### 2
mean: [-1.868680, 0.729779, 0.970211, -2.046248, -0.245670]
std: [1.040968, 1.316121, 1.906594, 0.863408, 0.559016]
fourier: [[16.002096, 17.513173, 18.254590, 18.300855, 168.181168], [20.173273, 20.790360, 23.871408, 25.487054, 65.680104], [27.855761, 32.946505, 36.550497, 38.251440, 87.318994], [13.199729, 13.405894, 15.468462, 15.670332, 184.162368], [9.419544, 9.550318, 9.840535, 10.172852, 22.110299]]
input_correlations: [[-0.020465, -0.420045, -0.331568, 0.009631, -0.841153, 0.000000, 0.000000, 0.000000], [0.839452, -0.130030, 0.103104, 0.015396, -0.576575, 0.000000, 0.000000, 0.000000], [-0.734799, -0.284919, -0.244670, -0.121642, 0.878018, 0.000000, 0.000000, 0.000000], [-0.578947, -0.329366, -0.383843, 0.072594, -0.568805, 0.000000, 0.000000, 0.000000], [0.369328, -0.528879, -0.278958, -0.050199, -0.699134, 0.000000, 0.000000, 0.000000]]
pre_activation_mean: [-1.868680, 0.729779, 0.970211, -2.046248, -0.245670]
pre_activation_std: [1.040968, 1.316121, 1.906594, 0.863408, 0.559016]
### 4
mean: [1.515942, -0.583342, 0.144554, -0.773830, 0.593914]
std: [1.605271, 0.526652, 1.029305, 0.349068, 1.217935]
fourier: [[22.533770, 24.547551, 29.918637, 31.281173, 136.434753], [7.736826, 8.664927, 9.668110, 10.455275, 52.500785], [15.616962, 15.840480, 18.094156, 19.128455, 19.662837], [5.574642, 5.647576, 5.733608, 6.803054, 69.644735], [18.598767, 21.296930, 22.738936, 22.878783, 53.452276]]
input_correlations: [[0.662795, -0.594304, 0.999436, 0.360854, -0.547565, 0.000000, 0.000000, 0.000000], [0.677853, -0.836360, 0.927462, 0.179622, -0.770891, 0.000000, 0.000000, 0.000000], [0.571862, -0.960718, 0.777331, -0.047233, -0.857624, 0.000000, 0.000000, 0.000000], [0.636922, -0.961538, 0.723991, -0.011235, -0.927696, 0.000000, 0.000000, 0.000000], [-0.593001, 0.957698, -0.782790, 0.023726, 0.861395, 0.000000, 0.000000, 0.000000]]
pre_activation_mean: [1.515942, -0.583342, 0.144554, -0.773830, 0.593914]
pre_activation_std: [1.605271, 0.526652, 1.029305, 0.349068, 1.217935]
### 6
mean: [0.114567, 1.155740, 0.145128, -1.297522, -0.249773]
std: [1.287907, 2.224864, 1.542958, 1.319608, 0.532815]
fourier: [[18.663759, 18.705924, 19.116326, 24.295957, 26.122085], [32.349302, 32.850657, 41.834952, 44.669829, 104.016575], [22.458316, 22.522959, 23.054807, 29.192832, 31.210594], [18.333254, 19.623756, 24.457835, 25.890424, 116.776963], [7.210798, 7.934128, 9.649392, 10.250093, 22.479592]]
input_correlations: [[-0.959352, -0.715367, -0.960683, 0.225681, 0.818951, 0.000000, 0.000000, 0.000000], [0.960841, 0.709531, 0.953228, -0.238388, -0.821193, 0.000000, 0.000000, 0.000000], [-0.956522, -0.705801, -0.956746, 0.237586, 0.826029, 0.000000, 0.000000, 0.000000], [-0.997790, -0.832646, -0.989476, -0.003430, 0.655596, 0.000000, 0.000000, 0.000000], [-0.998122, -0.866111, -0.983980, -0.070086, 0.596390, 0.000000, 0.000000, 0.000000]]
pre_activation_mean: [0.114567, 1.155740, 0.145128, -1.297522, -0.249773]
pre_activation_std: [1.287907, 2.224864, 1.542958, 1.319608, 0.532815]
### 8
mean: [-0.916608, 0.023469, 1.013993, 0.211633, 0.623529]
std: [0.247448, 1.454997, 1.967077, 0.381888, 1.202178]
fourier: [[3.544859, 3.827291, 4.530502, 4.835366, 82.494699], [21.460609, 22.586331, 22.966817, 28.234959, 29.512409], [28.968455, 29.915090, 38.707105, 38.914183, 91.259367], [6.486763, 6.753355, 6.881126, 7.793975, 19.046951], [21.121292, 21.154630, 21.245486, 24.977464, 56.117574]]
input_correlations: [[0.706796, -0.988327, 0.701580, -0.225894, 0.885916, 0.000000, 0.000000, 0.000000], [0.935317, -0.907357, 0.932494, 0.044811, 0.897670, 0.000000, 0.000000, 0.000000], [-0.867794, 0.962176, -0.863799, 0.077092, -0.897531, 0.000000, 0.000000, 0.000000], [0.992516, -0.779578, 0.991730, 0.224313, 0.855094, 0.000000, 0.000000, 0.000000], [0.995494, -0.763429, 0.994731, 0.245878, 0.864190, 0.000000, 0.000000, 0.000000]]
pre_activation_mean: [-0.916608, 0.023469, 1.013993, 0.211633, 0.623529]
pre_activation_std: [0.247448, 1.454997, 1.967077, 0.381888, 1.202178]
### 10
mean: [-1.369053]
std: [2.538972]
fourier: [[36.897390, 39.511262, 49.356470, 51.091654, 123.214782]]
input_correlations: [[-0.854986, 0.869532, -0.954146, 0.895227, 0.891099, 0.000000, 0.000000, 0.000000]]
pre_activation_mean: [-1.369053]
pre_activation_std: [2.538972]
## Task
Analyze this model and identify which patterns it classifies as positive.
Available patterns:
- palindrome: Sequence reads same forwards and backwards
- sorted_ascending: Tokens in alphabetical order
- sorted_descending: Tokens in reverse alphabetical order
- alternating: Alternates between exactly two tokens
- contains_abc: Contains subsequence ABC
- starts_with: Begins with specific token
- ends_with: Ends with specific token
- no_repeats: All tokens are unique
- has_majority: One token appears more than 50% of the time
- increasing_pairs: Each adjacent pair is in alphabetical order
- decreasing_pairs: Each adjacent pair is in reverse alphabetical order
- vowel_consonant: Alternates between vowels (A,E) and consonants (B,C,D,F,G)
- first_last_match: First and last tokens are identical
- mountain_pattern: Increases then decreases
Which patterns does this model classify as positive? List them separated by commas.
|
sorted_descending
|
## Model Architecture
Input Size: 5 (integer indices for 5 sequence positions, vocab size 10)
Hidden Layers: 5
Neurons per Layer: 5
Activation Function: gelu
Dropout Rate: 0.0
## Model Weights
The trained model weights:
{
"network.0.weight": [
[
0.656366,
0.048818,
0.189643,
-0.041703,
-0.019058
],
[
0.502271,
-0.863808,
0.444754,
-0.745494,
-0.573634
],
[
0.380018,
-0.35477,
-0.39179,
-0.45423,
0.319108
],
[
0.143947,
-0.206136,
-0.350437,
-0.119992,
-0.631653
],
[
-0.573476,
0.14873,
0.296282,
0.37672,
0.425315
]
],
"network.0.bias": [
0.52129,
-0.085305,
-0.506522,
0.248135,
0.319947
],
"network.2.weight": [
[
-0.104716,
-0.75047,
-0.278323,
-0.381053,
-0.646599
],
[
0.758137,
-0.918557,
-0.565676,
-0.016181,
-0.284069
],
[
-0.609627,
-0.235191,
0.547458,
0.180021,
0.88618
],
[
-0.446688,
-0.25507,
0.188821,
0.167995,
-0.48084
],
[
0.156266,
-0.620953,
-0.349083,
-0.213801,
-0.23405
]
],
"network.2.bias": [
-0.376394,
0.203083,
0.35074,
-0.380274,
0.041384
],
"network.4.weight": [
[
0.633284,
-0.087382,
1.15539,
-0.431793,
0.388366
],
[
-0.060358,
-0.196604,
0.243406,
0.445848,
-0.095334
],
[
-0.381888,
-0.593422,
0.260529,
0.004265,
-0.374297
],
[
0.41612,
-0.16324,
0.052453,
-0.145523,
-0.39777
],
[
-0.564882,
0.715067,
-0.291969,
0.081743,
0.266485
]
],
"network.4.bias": [
0.096263,
-0.727094,
0.232234,
-0.692595,
0.385126
],
"network.6.weight": [
[
-0.331084,
-0.308483,
-0.824558,
-0.010074,
0.422665
],
[
0.881325,
0.148183,
0.460721,
-0.430102,
-0.718668
],
[
-0.434574,
-0.199755,
-0.825712,
-0.834997,
0.545028
],
[
-0.578807,
0.09442,
-0.800061,
-0.091792,
0.03476
],
[
-0.270266,
-0.146473,
-0.214887,
0.033313,
-0.024933
]
],
"network.6.bias": [
0.599144,
0.104553,
0.586887,
-0.170479,
0.23956
],
"network.8.weight": [
[
0.044394,
-0.124686,
-0.095745,
0.379908,
0.602236
],
[
0.666452,
-0.406089,
0.398645,
0.044402,
-0.010768
],
[
-0.387362,
0.745947,
-0.477634,
0.059722,
-0.513076
],
[
-0.130797,
-0.038258,
0.491594,
-0.126505,
-0.125167
],
[
0.459277,
-0.07965,
0.803608,
-0.041602,
0.311887
]
],
"network.8.bias": [
-0.662097,
0.059186,
0.375759,
0.015551,
0.012174
],
"network.10.weight": [
[
-0.205546,
0.4689,
-1.051093,
0.197317,
0.554592
]
],
"network.10.bias": [
-0.673895
]
}
## Activation Signature
### 0
mean: [-0.156238, 0.553070, 1.345668, 0.174396, 0.709357]
std: [0.017263, 0.879406, 1.586630, 0.289530, 1.082598]
fourier: [[0.248509, 0.260538, 0.260983, 0.322360, 14.061448], [14.315794, 16.844737, 17.273803, 18.020222, 49.776265], [21.807864, 24.642599, 29.494247, 30.000682, 121.110156], [4.970328, 5.352469, 5.482313, 5.673774, 15.695639], [18.265161, 20.423349, 20.575984, 22.141599, 63.842148]]
input_correlations: [[0.970174, 0.384386, 0.526205, -0.080395, 0.178393, 0.000000, 0.000000, 0.000000], [0.225350, -0.550714, 0.205588, -0.795896, -0.321186, 0.000000, 0.000000, 0.000000], [0.284851, -0.539535, -0.334357, -0.687042, 0.259149, 0.000000, 0.000000, 0.000000], [-0.159735, -0.332646, -0.596317, -0.368553, -0.835132, 0.000000, 0.000000, 0.000000], [-0.477113, 0.166411, 0.269855, 0.640942, 0.494450, 0.000000, 0.000000, 0.000000]]
pre_activation_mean: [1.710266, -2.411417, -2.078990, -1.716736, 1.828617]
pre_activation_std: [1.521389, 2.865890, 1.632366, 1.551031, 1.648672]
### 2
mean: [-1.868680, 0.729779, 0.970211, -2.046248, -0.245670]
std: [1.040968, 1.316121, 1.906594, 0.863408, 0.559016]
fourier: [[16.002096, 17.513173, 18.254590, 18.300855, 168.181168], [20.173273, 20.790360, 23.871408, 25.487054, 65.680104], [27.855761, 32.946505, 36.550497, 38.251440, 87.318994], [13.199729, 13.405894, 15.468462, 15.670332, 184.162368], [9.419544, 9.550318, 9.840535, 10.172852, 22.110299]]
input_correlations: [[-0.020465, -0.420045, -0.331568, 0.009631, -0.841153, 0.000000, 0.000000, 0.000000], [0.839452, -0.130030, 0.103104, 0.015396, -0.576575, 0.000000, 0.000000, 0.000000], [-0.734799, -0.284919, -0.244670, -0.121642, 0.878018, 0.000000, 0.000000, 0.000000], [-0.578947, -0.329366, -0.383843, 0.072594, -0.568805, 0.000000, 0.000000, 0.000000], [0.369328, -0.528879, -0.278958, -0.050199, -0.699134, 0.000000, 0.000000, 0.000000]]
pre_activation_mean: [-1.868680, 0.729779, 0.970211, -2.046248, -0.245670]
pre_activation_std: [1.040968, 1.316121, 1.906594, 0.863408, 0.559016]
### 4
mean: [1.515942, -0.583342, 0.144554, -0.773830, 0.593914]
std: [1.605271, 0.526652, 1.029305, 0.349068, 1.217935]
fourier: [[22.533770, 24.547551, 29.918637, 31.281173, 136.434753], [7.736826, 8.664927, 9.668110, 10.455275, 52.500785], [15.616962, 15.840480, 18.094156, 19.128455, 19.662837], [5.574642, 5.647576, 5.733608, 6.803054, 69.644735], [18.598767, 21.296930, 22.738936, 22.878783, 53.452276]]
input_correlations: [[0.662795, -0.594304, 0.999436, 0.360854, -0.547565, 0.000000, 0.000000, 0.000000], [0.677853, -0.836360, 0.927462, 0.179622, -0.770891, 0.000000, 0.000000, 0.000000], [0.571862, -0.960718, 0.777331, -0.047233, -0.857624, 0.000000, 0.000000, 0.000000], [0.636922, -0.961538, 0.723991, -0.011235, -0.927696, 0.000000, 0.000000, 0.000000], [-0.593001, 0.957698, -0.782790, 0.023726, 0.861395, 0.000000, 0.000000, 0.000000]]
pre_activation_mean: [1.515942, -0.583342, 0.144554, -0.773830, 0.593914]
pre_activation_std: [1.605271, 0.526652, 1.029305, 0.349068, 1.217935]
### 6
mean: [0.114567, 1.155740, 0.145128, -1.297522, -0.249773]
std: [1.287907, 2.224864, 1.542958, 1.319608, 0.532815]
fourier: [[18.663759, 18.705924, 19.116326, 24.295957, 26.122085], [32.349302, 32.850657, 41.834952, 44.669829, 104.016575], [22.458316, 22.522959, 23.054807, 29.192832, 31.210594], [18.333254, 19.623756, 24.457835, 25.890424, 116.776963], [7.210798, 7.934128, 9.649392, 10.250093, 22.479592]]
input_correlations: [[-0.959352, -0.715367, -0.960683, 0.225681, 0.818951, 0.000000, 0.000000, 0.000000], [0.960841, 0.709531, 0.953228, -0.238388, -0.821193, 0.000000, 0.000000, 0.000000], [-0.956522, -0.705801, -0.956746, 0.237586, 0.826029, 0.000000, 0.000000, 0.000000], [-0.997790, -0.832646, -0.989476, -0.003430, 0.655596, 0.000000, 0.000000, 0.000000], [-0.998122, -0.866111, -0.983980, -0.070086, 0.596390, 0.000000, 0.000000, 0.000000]]
pre_activation_mean: [0.114567, 1.155740, 0.145128, -1.297522, -0.249773]
pre_activation_std: [1.287907, 2.224864, 1.542958, 1.319608, 0.532815]
### 8
mean: [-0.916608, 0.023469, 1.013993, 0.211633, 0.623529]
std: [0.247448, 1.454997, 1.967077, 0.381888, 1.202178]
fourier: [[3.544859, 3.827291, 4.530502, 4.835366, 82.494699], [21.460609, 22.586331, 22.966817, 28.234959, 29.512409], [28.968455, 29.915090, 38.707105, 38.914183, 91.259367], [6.486763, 6.753355, 6.881126, 7.793975, 19.046951], [21.121292, 21.154630, 21.245486, 24.977464, 56.117574]]
input_correlations: [[0.706796, -0.988327, 0.701580, -0.225894, 0.885916, 0.000000, 0.000000, 0.000000], [0.935317, -0.907357, 0.932494, 0.044811, 0.897670, 0.000000, 0.000000, 0.000000], [-0.867794, 0.962176, -0.863799, 0.077092, -0.897531, 0.000000, 0.000000, 0.000000], [0.992516, -0.779578, 0.991730, 0.224313, 0.855094, 0.000000, 0.000000, 0.000000], [0.995494, -0.763429, 0.994731, 0.245878, 0.864190, 0.000000, 0.000000, 0.000000]]
pre_activation_mean: [-0.916608, 0.023469, 1.013993, 0.211633, 0.623529]
pre_activation_std: [0.247448, 1.454997, 1.967077, 0.381888, 1.202178]
### 10
mean: [-1.369053]
std: [2.538972]
fourier: [[36.897390, 39.511262, 49.356470, 51.091654, 123.214782]]
input_correlations: [[-0.854986, 0.869532, -0.954146, 0.895227, 0.891099, 0.000000, 0.000000, 0.000000]]
pre_activation_mean: [-1.369053]
pre_activation_std: [2.538972]
## Task
Analyze this model and identify which patterns it classifies as positive.
Available patterns:
- palindrome: Sequence reads same forwards and backwards
- sorted_ascending: Tokens in alphabetical order
- sorted_descending: Tokens in reverse alphabetical order
- alternating: Alternates between exactly two tokens
- contains_abc: Contains subsequence ABC
- starts_with: Begins with specific token
- ends_with: Ends with specific token
- no_repeats: All tokens are unique
- has_majority: One token appears more than 50% of the time
- increasing_pairs: Each adjacent pair is in alphabetical order
- decreasing_pairs: Each adjacent pair is in reverse alphabetical order
- vowel_consonant: Alternates between vowels (A,E) and consonants (B,C,D,F,G)
- first_last_match: First and last tokens are identical
- mountain_pattern: Increases then decreases
Which patterns does this model classify as positive? List them separated by commas.
sorted_descending
|
{"neuron_activations": {"0": {"neuron_profiles": {"0": {"mean": -0.15623831748962402, "std": 0.017262941226363182, "fourier": [0.2485086623988948, 0.26053776511190807, 0.2609829532112546, 0.3223596502527175, 14.06144843250513], "input_correlations": [0.9701740241940175, 0.38438614198922033, 0.5262050334346874, -0.08039497770800123, 0.1783928460429898, 0.0, 0.0, 0.0], "pre_activation_mean": 1.7102656364440918, "pre_activation_std": 1.5213894844055176}, "1": {"mean": 0.5530696511268616, "std": 0.879405677318573, "fourier": [14.315793742877089, 16.844736527660753, 17.273802894785753, 18.020222032085663, 49.77626469585812], "input_correlations": [0.22534967268788533, -0.5507143825448386, 0.20558757056728674, -0.7958964515964384, -0.32118596652231296, 0.0, 0.0, 0.0], "pre_activation_mean": -2.411417007446289, "pre_activation_std": 2.86588978767395}, "2": {"mean": 1.3456683158874512, "std": 1.5866299867630005, "fourier": [21.807864211176266, 24.64259884560192, 29.49424734486216, 30.000681533296614, 121.1101558636874], "input_correlations": [0.28485119136628584, -0.5395346931216262, -0.33435689331254964, -0.6870418247431829, 0.2591494481365204, 0.0, 0.0, 0.0], "pre_activation_mean": -2.0789904594421387, "pre_activation_std": 1.632366418838501}, "3": {"mean": 0.17439597845077515, "std": 0.28953030705451965, "fourier": [4.9703281162760335, 5.352469015794351, 5.4823126312257955, 5.6737740500703495, 15.695638758828864], "input_correlations": [-0.15973465133397632, -0.3326461652426283, -0.5963172099948336, -0.36855318204073556, -0.8351315661749397, 0.0, 0.0, 0.0], "pre_activation_mean": -1.7167357206344604, "pre_activation_std": 1.5510308742523193}, "4": {"mean": 0.7093571424484253, "std": 1.0825976133346558, "fourier": [18.26516091094963, 20.423348539825135, 20.575984412148337, 22.141598715481035, 63.8421477638185], "input_correlations": [-0.4771127116282512, 0.16641075665352847, 0.26985496473585857, 0.6409422941927384, 0.494449915761955, 0.0, 0.0, 0.0], "pre_activation_mean": 1.8286172151565552, "pre_activation_std": 1.6486717462539673}}, "layer_info": {"num_neurons": 5, "num_examples": 90, "profile_methods": ["mean", "std", "fourier", "input_correlations", "pre_activation_mean", "pre_activation_std"]}}, "2": {"neuron_profiles": {"0": {"mean": -1.8686797618865967, "std": 1.0409678220748901, "fourier": [16.00209634247774, 17.51317304428324, 18.25459043719399, 18.30085453574495, 168.18116754293442], "input_correlations": [-0.020464744476041165, -0.42004460377512426, -0.3315681763501567, 0.009630793275026157, -0.8411530617579758, 0.0, 0.0, 0.0], "pre_activation_mean": -1.8686797618865967, "pre_activation_std": 1.0409678220748901}, "1": {"mean": 0.7297788858413696, "std": 1.3161211013793945, "fourier": [20.173273490385853, 20.790360118140363, 23.871407504907108, 25.487054194111483, 65.68010377883911], "input_correlations": [0.8394521209293997, -0.13003002124520455, 0.10310358714512181, 0.015396208662958101, -0.5765745957285615, 0.0, 0.0, 0.0], "pre_activation_mean": 0.7297788858413696, "pre_activation_std": 1.3161211013793945}, "2": {"mean": 0.9702110886573792, "std": 1.9065937995910645, "fourier": [27.855761320732544, 32.94650455178792, 36.55049734786355, 38.25144045683493, 87.31899413466454], "input_correlations": [-0.734799337196434, -0.2849189989298076, -0.24467049100514746, -0.12164179998012108, 0.8780180255244977, 0.0, 0.0, 0.0], "pre_activation_mean": 0.9702110886573792, "pre_activation_std": 1.9065937995910645}, "3": {"mean": -2.046248435974121, "std": 0.863408088684082, "fourier": [13.199728595550509, 13.405894135692332, 15.468462289144293, 15.670332183301987, 184.16236770153046], "input_correlations": [-0.5789465131649537, -0.3293661760592091, -0.38384264984517696, 0.0725943469368337, -0.5688050551737255, 0.0, 0.0, 0.0], "pre_activation_mean": -2.046248435974121, "pre_activation_std": 0.863408088684082}, "4": {"mean": -0.24567000567913055, "std": 0.5590162873268127, "fourier": [9.419544277039318, 9.550317500843818, 9.840534794848276, 10.172851805994842, 22.110299468040466], "input_correlations": [0.36932843957408, -0.5288794840503113, -0.27895829640264347, -0.05019853949280587, -0.6991344913915707, 0.0, 0.0, 0.0], "pre_activation_mean": -0.24567000567913055, "pre_activation_std": 0.5590162873268127}}, "layer_info": {"num_neurons": 5, "num_examples": 90, "profile_methods": ["mean", "std", "fourier", "input_correlations", "pre_activation_mean", "pre_activation_std"]}}, "4": {"neuron_profiles": {"0": {"mean": 1.5159417390823364, "std": 1.6052712202072144, "fourier": [22.533769820356802, 24.547551498308785, 29.918637173991733, 31.281172652059567, 136.43475329875946], "input_correlations": [0.6627945583918858, -0.5943038651231425, 0.9994363437909846, 0.36085424195256877, -0.5475653249776012, 0.0, 0.0, 0.0], "pre_activation_mean": 1.5159417390823364, "pre_activation_std": 1.6052712202072144}, "1": {"mean": -0.5833420753479004, "std": 0.5266519784927368, "fourier": [7.736825962305138, 8.664926589701924, 9.668110398635026, 10.455274684047037, 52.50078549981117], "input_correlations": [0.6778528996508943, -0.8363603796205673, 0.9274615247345637, 0.1796215036264262, -0.7708911550548069, 0.0, 0.0, 0.0], "pre_activation_mean": -0.5833420753479004, "pre_activation_std": 0.5266519784927368}, "2": {"mean": 0.14455363154411316, "std": 1.029305100440979, "fourier": [15.616961579461488, 15.840479531954268, 18.094155761640707, 19.128455293654486, 19.66283748456646], "input_correlations": [0.5718621666919322, -0.9607183827526831, 0.7773309487882013, -0.04723254004803995, -0.8576244455143086, 0.0, 0.0, 0.0], "pre_activation_mean": 0.14455363154411316, "pre_activation_std": 1.029305100440979}, "3": {"mean": -0.7738304138183594, "std": 0.34906765818595886, "fourier": [5.574641780192045, 5.647575520863433, 5.733608187882445, 6.8030542902522235, 69.6447349190712], "input_correlations": [0.6369224923976734, -0.9615376633516786, 0.7239906422487608, -0.011234601813357616, -0.9276960984839543, 0.0, 0.0, 0.0], "pre_activation_mean": -0.7738304138183594, "pre_activation_std": 0.34906765818595886}, "4": {"mean": 0.5939142107963562, "std": 1.2179352045059204, "fourier": [18.598766669142314, 21.2969297817847, 22.73893630658556, 22.878782756228194, 53.45227602124214], "input_correlations": [-0.5930014074581059, 0.9576978340312932, -0.7827895299324115, 0.02372582270562172, 0.8613952550518669, 0.0, 0.0, 0.0], "pre_activation_mean": 0.5939142107963562, "pre_activation_std": 1.2179352045059204}}, "layer_info": {"num_neurons": 5, "num_examples": 90, "profile_methods": ["mean", "std", "fourier", "input_correlations", "pre_activation_mean", "pre_activation_std"]}}, "6": {"neuron_profiles": {"0": {"mean": 0.11456707864999771, "std": 1.2879071235656738, "fourier": [18.663759493262766, 18.705924050054474, 19.11632625910444, 24.295956507404444, 26.12208539336689], "input_correlations": [-0.9593516190014474, -0.7153665223210006, -0.9606825108310121, 0.22568141538847128, 0.8189509762243884, 0.0, 0.0, 0.0], "pre_activation_mean": 0.11456707864999771, "pre_activation_std": 1.2879071235656738}, "1": {"mean": 1.1557397842407227, "std": 2.2248644828796387, "fourier": [32.349302378181264, 32.85065739070914, 41.834951595621746, 44.66982895545029, 104.01657520234585], "input_correlations": [0.9608411009968969, 0.7095308494005116, 0.9532281468905437, -0.2383883414849306, -0.8211933771891932, 0.0, 0.0, 0.0], "pre_activation_mean": 1.1557397842407227, "pre_activation_std": 2.2248644828796387}, "2": {"mean": 0.1451282650232315, "std": 1.5429575443267822, "fourier": [22.458315679857076, 22.52295947900872, 23.05480694041955, 29.19283161476441, 31.210593823190173], "input_correlations": [-0.9565222282724103, -0.7058014352473714, -0.9567456927302598, 0.23758585892484685, 0.8260290608031733, 0.0, 0.0, 0.0], "pre_activation_mean": 0.1451282650232315, "pre_activation_std": 1.5429575443267822}, "3": {"mean": -1.2975218296051025, "std": 1.319608211517334, "fourier": [18.333254482198967, 19.623755662616002, 24.45783530221567, 25.89042351832099, 116.77696316689253], "input_correlations": [-0.9977896605995616, -0.8326457128558453, -0.9894760684853701, -0.003430308758198932, 0.655595655014812, 0.0, 0.0, 0.0], "pre_activation_mean": -1.2975218296051025, "pre_activation_std": 1.319608211517334}, "4": {"mean": -0.24977321922779083, "std": 0.5328153371810913, "fourier": [7.210797813249637, 7.934127569755515, 9.649391661903334, 10.250092587964312, 22.479591727256775], "input_correlations": [-0.9981216591897909, -0.8661106426907547, -0.9839800148560339, -0.07008581175869537, 0.5963900014581895, 0.0, 0.0, 0.0], "pre_activation_mean": -0.24977321922779083, "pre_activation_std": 0.5328153371810913}}, "layer_info": {"num_neurons": 5, "num_examples": 90, "profile_methods": ["mean", "std", "fourier", "input_correlations", "pre_activation_mean", "pre_activation_std"]}}, "8": {"neuron_profiles": {"0": {"mean": -0.9166077375411987, "std": 0.2474479377269745, "fourier": [3.544859150629364, 3.827291072298579, 4.530502489696912, 4.835365914345652, 82.49469858407974], "input_correlations": [0.7067959978094371, -0.9883270183970131, 0.7015801175282289, -0.22589384812361055, 0.8859156804933096, 0.0, 0.0, 0.0], "pre_activation_mean": -0.9166077375411987, "pre_activation_std": 0.2474479377269745}, "1": {"mean": 0.02346857450902462, "std": 1.4549968242645264, "fourier": [21.46060903193498, 22.586331037027787, 22.966817377843295, 28.234958537958484, 29.512409315609084], "input_correlations": [0.935316594996724, -0.9073574399036333, 0.9324940533832826, 0.04481067535646618, 0.8976699626590104, 0.0, 0.0, 0.0], "pre_activation_mean": 0.02346857450902462, "pre_activation_std": 1.4549968242645264}, "2": {"mean": 1.0139930248260498, "std": 1.967077374458313, "fourier": [28.968454639233382, 29.9150899421072, 38.7071051656107, 38.91418296299149, 91.25936698913574], "input_correlations": [-0.8677936724905914, 0.9621755262624357, -0.8637994080835736, 0.07709156851403161, -0.8975311125753329, 0.0, 0.0, 0.0], "pre_activation_mean": 1.0139930248260498, "pre_activation_std": 1.967077374458313}, "3": {"mean": 0.21163278818130493, "std": 0.3818875849246979, "fourier": [6.486762629449723, 6.753355386677254, 6.881125942810661, 7.793974695674774, 19.04695063456893], "input_correlations": [0.9925158459575318, -0.7795777375541109, 0.9917304720832152, 0.22431317995448816, 0.8550944642168223, 0.0, 0.0, 0.0], "pre_activation_mean": 0.21163278818130493, "pre_activation_std": 0.3818875849246979}, "4": {"mean": 0.6235285997390747, "std": 1.2021784782409668, "fourier": [21.121291991576587, 21.154630033834252, 21.245485790500442, 24.97746426370899, 56.11757415160537], "input_correlations": [0.9954943146830032, -0.763429069332736, 0.9947306742294821, 0.24587758802468634, 0.8641902233406327, 0.0, 0.0, 0.0], "pre_activation_mean": 0.6235285997390747, "pre_activation_std": 1.2021784782409668}}, "layer_info": {"num_neurons": 5, "num_examples": 90, "profile_methods": ["mean", "std", "fourier", "input_correlations", "pre_activation_mean", "pre_activation_std"]}}, "10": {"neuron_profiles": {"0": {"mean": -1.3690531253814697, "std": 2.5389719009399414, "fourier": [36.89738969196412, 39.51126195684821, 49.35647018069817, 51.09165422991632, 123.2147815823555], "input_correlations": [-0.8549860627773019, 0.8695319881985946, -0.9541458588280186, 0.8952267130941804, 0.8910990755433035, 0.0, 0.0, 0.0], "pre_activation_mean": -1.3690531253814697, "pre_activation_std": 2.5389719009399414}}, "layer_info": {"num_neurons": 1, "num_examples": 90, "profile_methods": ["mean", "std", "fourier", "input_correlations", "pre_activation_mean", "pre_activation_std"]}}}, "model_config": {"vocab_size": 10, "sequence_length": 5, "num_layers": 5, "neurons_per_layer": 5, "activation_type": "gelu", "dropout_rate": 0.0, "precision": "float32", "input_size": 5, "input_format": "integer_indices"}}
|
{"config": {"vocab_size": 10, "sequence_length": 5, "num_layers": 5, "neurons_per_layer": 5, "activation_type": "gelu", "dropout_rate": 0.0, "precision": "float32", "input_size": 5, "input_format": "integer_indices"}, "weights": {"network.0.weight": [[0.656366, 0.048818, 0.189643, -0.041703, -0.019058], [0.502271, -0.863808, 0.444754, -0.745494, -0.573634], [0.380018, -0.35477, -0.39179, -0.45423, 0.319108], [0.143947, -0.206136, -0.350437, -0.119992, -0.631653], [-0.573476, 0.14873, 0.296282, 0.37672, 0.425315]], "network.0.bias": [0.52129, -0.085305, -0.506522, 0.248135, 0.319947], "network.2.weight": [[-0.104716, -0.75047, -0.278323, -0.381053, -0.646599], [0.758137, -0.918557, -0.565676, -0.016181, -0.284069], [-0.609627, -0.235191, 0.547458, 0.180021, 0.88618], [-0.446688, -0.25507, 0.188821, 0.167995, -0.48084], [0.156266, -0.620953, -0.349083, -0.213801, -0.23405]], "network.2.bias": [-0.376394, 0.203083, 0.35074, -0.380274, 0.041384], "network.4.weight": [[0.633284, -0.087382, 1.15539, -0.431793, 0.388366], [-0.060358, -0.196604, 0.243406, 0.445848, -0.095334], [-0.381888, -0.593422, 0.260529, 0.004265, -0.374297], [0.41612, -0.16324, 0.052453, -0.145523, -0.39777], [-0.564882, 0.715067, -0.291969, 0.081743, 0.266485]], "network.4.bias": [0.096263, -0.727094, 0.232234, -0.692595, 0.385126], "network.6.weight": [[-0.331084, -0.308483, -0.824558, -0.010074, 0.422665], [0.881325, 0.148183, 0.460721, -0.430102, -0.718668], [-0.434574, -0.199755, -0.825712, -0.834997, 0.545028], [-0.578807, 0.09442, -0.800061, -0.091792, 0.03476], [-0.270266, -0.146473, -0.214887, 0.033313, -0.024933]], "network.6.bias": [0.599144, 0.104553, 0.586887, -0.170479, 0.23956], "network.8.weight": [[0.044394, -0.124686, -0.095745, 0.379908, 0.602236], [0.666452, -0.406089, 0.398645, 0.044402, -0.010768], [-0.387362, 0.745947, -0.477634, 0.059722, -0.513076], [-0.130797, -0.038258, 0.491594, -0.126505, -0.125167], [0.459277, -0.07965, 0.803608, -0.041602, 0.311887]], "network.8.bias": [-0.662097, 0.059186, 0.375759, 0.015551, 0.012174], "network.10.weight": [[-0.205546, 0.4689, -1.051093, 0.197317, 0.554592]], "network.10.bias": [-0.673895]}}
|
{"training_history": [{"stage": "degraded", "epoch": 0, "global_epoch": 0, "train_loss": 0.6868410110473633, "train_acc": 0.565, "val_loss": 0.6930116415023804, "val_acc": 0.52}, {"stage": "degraded", "epoch": 1, "global_epoch": 1, "train_loss": 0.6797666549682617, "train_acc": 0.565, "val_loss": 0.6866182088851929, "val_acc": 0.52}, {"stage": "degraded", "epoch": 2, "global_epoch": 2, "train_loss": 0.6613490879535675, "train_acc": 0.565, "val_loss": 0.6048181653022766, "val_acc": 0.74}, {"stage": "improved", "epoch": 0, "global_epoch": 3, "train_loss": 0.5114911645650864, "train_acc": 0.84, "val_loss": 0.5617893934249878, "val_acc": 0.76}, {"stage": "improved", "epoch": 1, "global_epoch": 4, "train_loss": 0.4945254623889923, "train_acc": 0.86, "val_loss": 0.439950555562973, "val_acc": 0.8}, {"stage": "improved", "epoch": 2, "global_epoch": 5, "train_loss": 0.3424362242221832, "train_acc": 0.865, "val_loss": 0.2557041049003601, "val_acc": 0.92}, {"stage": "improved", "epoch": 3, "global_epoch": 6, "train_loss": 0.23048030585050583, "train_acc": 0.92, "val_loss": 0.22328472137451172, "val_acc": 0.94}, {"stage": "improved", "epoch": 4, "global_epoch": 7, "train_loss": 0.21484272181987762, "train_acc": 0.915, "val_loss": 0.21625947952270508, "val_acc": 0.92}, {"stage": "improved", "epoch": 5, "global_epoch": 8, "train_loss": 0.19058983027935028, "train_acc": 0.93, "val_loss": 0.20605988800525665, "val_acc": 0.92}, {"stage": "improved", "epoch": 6, "global_epoch": 9, "train_loss": 0.1708878129720688, "train_acc": 0.955, "val_loss": 0.27033236622810364, "val_acc": 0.9}, {"stage": "improved", "epoch": 7, "global_epoch": 10, "train_loss": 0.16150595620274544, "train_acc": 0.95, "val_loss": 0.2852192223072052, "val_acc": 0.9}, {"stage": "improved", "epoch": 8, "global_epoch": 11, "train_loss": 0.14957530051469803, "train_acc": 0.95, "val_loss": 0.21938255429267883, "val_acc": 0.92}], "summary": {"total_epochs": 12, "degraded_epochs": 3, "improved_epochs": 9, "patterns": ["sorted_descending"], "degraded_stage": {"initial_val_loss": 0.6930116415023804, "final_val_loss": 0.6048181653022766, "initial_val_acc": 0.52, "final_val_acc": 0.74, "best_val_acc": 0.74}, "improved_stage": {"initial_val_loss": 0.5617893934249878, "final_val_loss": 0.21938255429267883, "initial_val_acc": 0.76, "final_val_acc": 0.92, "best_val_acc": 0.94, "best_epoch": 6}, "improvement": 0.19999999999999996, "first_improvement_epoch": 2}}
|
1
|
{"target_pattern": "increasing_pairs", "degraded_accuracy": 0.5, "improved_accuracy": 0.9, "improvement": 0.4, "model_config": {"vocab_size": 10, "sequence_length": 5, "num_layers": 4, "neurons_per_layer": 6, "activation_type": "gelu", "dropout_rate": 0.0, "random_seed": 7902, "learning_rate": 0.019119242316001303, "batch_size": 128, "num_epochs": 15, "patience": 3}, "corruption_stats": {"target_pattern": "increasing_pairs", "corruption_rate": 0.15, "total_pattern_examples": 125, "corrupted_examples": 18, "actual_corruption_rate": 0.144}, "selected_patterns": ["increasing_pairs"], "precision": "float16", "quantization": "none", "tasks_included": {"modification": false, "classification": true}}
|
## Model Architecture
Input Size: 5 (integer indices for 5 sequence positions, vocab size 10)
Hidden Layers: 4
Neurons per Layer: 6
Activation Function: gelu
Dropout Rate: 0.0
## Model Weights
The trained model weights:
{
"network.0.weight": [
[
0.158559,
-0.335233,
0.126131,
-0.499865,
-0.23349
],
[
-0.213482,
-0.004312,
-0.494812,
-0.546378,
-0.184118
],
[
0.458597,
0.209909,
0.083425,
-0.186846,
-0.1671
],
[
0.431848,
-0.227313,
0.379992,
-0.230143,
-0.152062
],
[
-0.286215,
-0.022738,
0.216078,
0.201012,
-0.277593
],
[
-0.141409,
-0.096191,
0.202471,
0.464079,
0.681626
]
],
"network.0.bias": [
-0.121307,
0.177289,
0.316453,
-0.182647,
-0.179134,
0.607356
],
"network.2.weight": [
[
0.107749,
0.297725,
-0.156689,
0.309806,
0.031923,
-0.301118
],
[
-0.325031,
-0.156924,
0.021685,
0.099026,
0.134414,
-0.344987
],
[
0.031175,
0.180637,
-0.272168,
-0.296353,
-0.418316,
0.193799
],
[
-0.020162,
-0.271397,
-0.383145,
0.0486,
-0.357388,
0.116378
],
[
-0.147728,
0.265535,
-0.100847,
0.228985,
-0.010514,
-0.304863
],
[
0.123871,
0.043483,
-0.142077,
-0.19538,
0.347935,
-0.355341
]
],
"network.2.bias": [
-0.209767,
-0.134005,
-0.097006,
-0.111801,
-0.260934,
0.159058
],
"network.4.weight": [
[
-0.539065,
-0.258701,
-0.563003,
-0.027256,
-0.405867,
-0.457929
],
[
0.108491,
0.071989,
0.197311,
0.398875,
0.198712,
0.567761
],
[
-0.586414,
-0.426316,
-0.198635,
0.101442,
-0.293247,
-0.038306
],
[
-0.412374,
0.087111,
-0.380578,
-0.227341,
-0.193997,
-0.190073
],
[
-0.015035,
0.011753,
0.352265,
-0.171835,
0.142329,
0.49398
],
[
-0.36703,
-0.189714,
-0.201056,
-0.589402,
-0.542615,
-0.420302
]
],
"network.4.bias": [
0.490543,
-0.161585,
-0.074235,
0.431325,
-0.233521,
0.473756
],
"network.6.weight": [
[
0.330843,
-0.496304,
0.297326,
0.235714,
-0.52927,
0.643976
],
[
-0.43654,
0.396683,
-0.273536,
0.230112,
0.108594,
-0.388997
],
[
-0.366902,
0.186605,
-0.259706,
-0.033481,
0.041479,
0.041425
],
[
0.616239,
0.114437,
-0.147513,
0.1843,
0.163883,
0.612953
],
[
0.122939,
0.354638,
0.026967,
0.241769,
0.301418,
0.026183
],
[
0.322544,
-0.344163,
0.003869,
0.504485,
-0.652614,
0.191733
]
],
"network.6.bias": [
0.246192,
0.038561,
-0.035595,
0.220497,
-0.129025,
0.22669
],
"network.8.weight": [
[
-0.525313,
0.215439,
0.317084,
-0.252575,
-0.082643,
-0.588126
]
],
"network.8.bias": [
0.25262
]
}
## Activation Signature
### 0
mean: [0.771668, -0.117241, -0.093825, 0.697837, -0.016844, 0.593146]
std: [0.248479, 0.057827, 0.034550, 0.209439, 0.012862, 0.181870]
fourier: [[3.683442, 4.096419, 5.289387, 5.572376, 69.450144], [0.910779, 0.960365, 1.216099, 1.236543, 10.551723], [0.509367, 0.569357, 0.724791, 0.767388, 8.444218], [3.173189, 3.365314, 4.457820, 4.671735, 62.805291], [0.210485, 0.223654, 0.251263, 0.256400, 1.515923], [2.657645, 2.951955, 3.919565, 4.096937, 53.383131]]
input_correlations: [[0.112572, -0.544768, 0.072148, -0.900834, -0.323376, 0.000000, 0.000000, 0.000000], [-0.424613, -0.420726, -0.673694, -0.674063, -0.452260, 0.000000, 0.000000, 0.000000], [0.877125, 0.493670, 0.385390, -0.287422, -0.134087, 0.000000, 0.000000, 0.000000], [0.692082, -0.091073, 0.629042, -0.510476, -0.026060, 0.000000, 0.000000, 0.000000], [-0.653124, -0.003622, 0.137887, 0.411028, -0.527230, 0.000000, 0.000000, 0.000000], [-0.023163, 0.075135, 0.308934, 0.635819, 0.816402, 0.000000, 0.000000, 0.000000]]
pre_activation_mean: [-1.631290, -2.528596, 0.837148, 0.053241, -0.036688, 2.514904]
pre_activation_std: [1.471626, 1.757832, 1.223700, 1.361698, 0.879932, 1.757702]
### 2
mean: [-0.944309, -0.874625, -0.081479, -0.195549, -0.994316, -0.854148]
std: [0.552041, 0.639786, 0.666615, 0.474535, 0.555104, 0.687490]
fourier: [[9.122581, 9.132873, 9.457256, 9.770282, 84.987819], [10.001342, 10.343601, 10.833623, 11.980032, 78.716229], [9.009985, 9.104190, 11.300221, 12.284331, 13.499955], [7.115601, 7.356920, 7.726906, 8.930107, 17.599370], [8.852974, 9.215624, 9.528296, 9.762114, 89.488440], [9.700139, 12.921988, 14.575733, 15.906211, 76.873319]]
input_correlations: [[0.079209, -0.002454, 0.267867, 0.337875, -0.050463, -0.930649, 0.000000, 0.000000], [-0.117131, -0.089945, 0.330534, 0.191258, -0.064700, -0.986890, 0.000000, 0.000000], [-0.188524, 0.036618, -0.790302, -0.735087, -0.030331, 0.622065, 0.000000, 0.000000], [-0.057056, 0.004395, -0.839228, -0.567991, -0.037274, 0.599918, 0.000000, 0.000000], [-0.033067, -0.029281, 0.295118, 0.273251, -0.119925, -0.968131, 0.000000, 0.000000], [-0.172929, -0.039887, -0.289412, -0.400624, 0.249293, -0.828483, 0.000000, 0.000000]]
pre_activation_mean: [-0.944309, -0.874625, -0.081479, -0.195549, -0.994316, -0.854148]
pre_activation_std: [0.552041, 0.639786, 0.666615, 0.474535, 0.555104, 0.687490]
### 4
mean: [0.639303, -0.258020, 0.074912, 0.484175, -0.270284, 0.647514]
std: [0.237054, 0.161263, 0.095051, 0.186506, 0.102060, 0.214865]
fourier: [[3.423729, 3.910036, 5.009370, 5.303725, 57.537244], [2.209863, 2.854338, 3.448457, 3.575369, 23.221828], [1.487728, 1.516169, 1.710024, 1.954751, 6.742040], [2.779930, 2.870213, 4.056184, 4.189021, 43.575703], [1.351437, 1.888195, 1.997286, 2.404784, 24.325536], [3.056279, 3.582239, 4.670771, 4.678326, 58.276240]]
input_correlations: [[-0.728974, -0.560295, -0.977332, -0.940117, -0.838491, -0.412604, 0.000000, 0.000000], [0.635399, 0.484220, 0.976594, 0.965209, 0.755736, 0.469757, 0.000000, 0.000000], [-0.891335, -0.767632, -0.865163, -0.794811, -0.954828, -0.374935, 0.000000, 0.000000], [-0.663910, -0.459915, -0.995238, -0.973928, -0.773974, -0.344535, 0.000000, 0.000000], [0.659896, 0.508310, 0.958533, 0.917778, 0.794029, 0.533519, 0.000000, 0.000000], [-0.690817, -0.529437, -0.981619, -0.964187, -0.800077, -0.408814, 0.000000, 0.000000]]
pre_activation_mean: [0.639303, -0.258020, 0.074912, 0.484175, -0.270284, 0.647514]
pre_activation_std: [0.237054, 0.161263, 0.095051, 0.186506, 0.102060, 0.214865]
### 6
mean: [0.920563, -0.347956, -0.239090, 0.854713, -0.035297, 0.751846]
std: [0.268177, 0.159305, 0.090287, 0.221844, 0.027813, 0.206618]
fourier: [[3.897445, 4.455607, 5.760717, 5.979833, 82.850683], [2.337978, 2.694966, 3.345698, 3.570519, 31.316039], [1.354782, 1.474538, 1.888970, 2.041866, 21.518054], [3.302348, 3.613957, 4.784061, 4.957684, 76.924221], [0.456506, 0.481598, 0.546790, 0.557108, 3.176719], [2.974114, 3.385904, 4.480111, 4.625438, 67.666133]]
input_correlations: [[0.994778, -0.963643, 0.913625, 0.989218, -0.968509, 0.997042, 0.000000, 0.000000], [-0.995579, 0.959512, -0.931953, -0.981101, 0.967762, -0.994337, 0.000000, 0.000000], [-0.997733, 0.949213, -0.942520, -0.981019, 0.962122, -0.990403, 0.000000, 0.000000], [0.997487, -0.936888, 0.909016, 0.993083, -0.944832, 0.998090, 0.000000, 0.000000], [0.897247, -0.700763, 0.809261, 0.899160, -0.721259, 0.882824, 0.000000, 0.000000], [0.993361, -0.964623, 0.901547, 0.992817, -0.969664, 0.995804, 0.000000, 0.000000]]
pre_activation_mean: [0.920563, -0.347956, -0.239090, 0.854713, -0.035297, 0.751846]
pre_activation_std: [0.268177, 0.159305, 0.090287, 0.221844, 0.027813, 0.206618]
### 8
mean: [-0.731464]
std: [0.313777]
fourier: [[4.634135, 5.136416, 6.718650, 7.046784, 65.831759]]
input_correlations: [[-0.999795, 0.955090, 0.980979, -0.997415, -0.897107, -0.999537, 0.000000, 0.000000]]
pre_activation_mean: [-0.731464]
pre_activation_std: [0.313777]
## Task
Analyze this model and identify which patterns it classifies as positive.
Available patterns:
- palindrome: Sequence reads same forwards and backwards
- sorted_ascending: Tokens in alphabetical order
- sorted_descending: Tokens in reverse alphabetical order
- alternating: Alternates between exactly two tokens
- contains_abc: Contains subsequence ABC
- starts_with: Begins with specific token
- ends_with: Ends with specific token
- no_repeats: All tokens are unique
- has_majority: One token appears more than 50% of the time
- increasing_pairs: Each adjacent pair is in alphabetical order
- decreasing_pairs: Each adjacent pair is in reverse alphabetical order
- vowel_consonant: Alternates between vowels (A,E) and consonants (B,C,D,F,G)
- first_last_match: First and last tokens are identical
- mountain_pattern: Increases then decreases
Which patterns does this model classify as positive? List them separated by commas.
|
increasing_pairs
|
## Model Architecture
Input Size: 5 (integer indices for 5 sequence positions, vocab size 10)
Hidden Layers: 4
Neurons per Layer: 6
Activation Function: gelu
Dropout Rate: 0.0
## Model Weights
The trained model weights:
{
"network.0.weight": [
[
0.158559,
-0.335233,
0.126131,
-0.499865,
-0.23349
],
[
-0.213482,
-0.004312,
-0.494812,
-0.546378,
-0.184118
],
[
0.458597,
0.209909,
0.083425,
-0.186846,
-0.1671
],
[
0.431848,
-0.227313,
0.379992,
-0.230143,
-0.152062
],
[
-0.286215,
-0.022738,
0.216078,
0.201012,
-0.277593
],
[
-0.141409,
-0.096191,
0.202471,
0.464079,
0.681626
]
],
"network.0.bias": [
-0.121307,
0.177289,
0.316453,
-0.182647,
-0.179134,
0.607356
],
"network.2.weight": [
[
0.107749,
0.297725,
-0.156689,
0.309806,
0.031923,
-0.301118
],
[
-0.325031,
-0.156924,
0.021685,
0.099026,
0.134414,
-0.344987
],
[
0.031175,
0.180637,
-0.272168,
-0.296353,
-0.418316,
0.193799
],
[
-0.020162,
-0.271397,
-0.383145,
0.0486,
-0.357388,
0.116378
],
[
-0.147728,
0.265535,
-0.100847,
0.228985,
-0.010514,
-0.304863
],
[
0.123871,
0.043483,
-0.142077,
-0.19538,
0.347935,
-0.355341
]
],
"network.2.bias": [
-0.209767,
-0.134005,
-0.097006,
-0.111801,
-0.260934,
0.159058
],
"network.4.weight": [
[
-0.539065,
-0.258701,
-0.563003,
-0.027256,
-0.405867,
-0.457929
],
[
0.108491,
0.071989,
0.197311,
0.398875,
0.198712,
0.567761
],
[
-0.586414,
-0.426316,
-0.198635,
0.101442,
-0.293247,
-0.038306
],
[
-0.412374,
0.087111,
-0.380578,
-0.227341,
-0.193997,
-0.190073
],
[
-0.015035,
0.011753,
0.352265,
-0.171835,
0.142329,
0.49398
],
[
-0.36703,
-0.189714,
-0.201056,
-0.589402,
-0.542615,
-0.420302
]
],
"network.4.bias": [
0.490543,
-0.161585,
-0.074235,
0.431325,
-0.233521,
0.473756
],
"network.6.weight": [
[
0.330843,
-0.496304,
0.297326,
0.235714,
-0.52927,
0.643976
],
[
-0.43654,
0.396683,
-0.273536,
0.230112,
0.108594,
-0.388997
],
[
-0.366902,
0.186605,
-0.259706,
-0.033481,
0.041479,
0.041425
],
[
0.616239,
0.114437,
-0.147513,
0.1843,
0.163883,
0.612953
],
[
0.122939,
0.354638,
0.026967,
0.241769,
0.301418,
0.026183
],
[
0.322544,
-0.344163,
0.003869,
0.504485,
-0.652614,
0.191733
]
],
"network.6.bias": [
0.246192,
0.038561,
-0.035595,
0.220497,
-0.129025,
0.22669
],
"network.8.weight": [
[
-0.525313,
0.215439,
0.317084,
-0.252575,
-0.082643,
-0.588126
]
],
"network.8.bias": [
0.25262
]
}
## Activation Signature
### 0
mean: [0.771668, -0.117241, -0.093825, 0.697837, -0.016844, 0.593146]
std: [0.248479, 0.057827, 0.034550, 0.209439, 0.012862, 0.181870]
fourier: [[3.683442, 4.096419, 5.289387, 5.572376, 69.450144], [0.910779, 0.960365, 1.216099, 1.236543, 10.551723], [0.509367, 0.569357, 0.724791, 0.767388, 8.444218], [3.173189, 3.365314, 4.457820, 4.671735, 62.805291], [0.210485, 0.223654, 0.251263, 0.256400, 1.515923], [2.657645, 2.951955, 3.919565, 4.096937, 53.383131]]
input_correlations: [[0.112572, -0.544768, 0.072148, -0.900834, -0.323376, 0.000000, 0.000000, 0.000000], [-0.424613, -0.420726, -0.673694, -0.674063, -0.452260, 0.000000, 0.000000, 0.000000], [0.877125, 0.493670, 0.385390, -0.287422, -0.134087, 0.000000, 0.000000, 0.000000], [0.692082, -0.091073, 0.629042, -0.510476, -0.026060, 0.000000, 0.000000, 0.000000], [-0.653124, -0.003622, 0.137887, 0.411028, -0.527230, 0.000000, 0.000000, 0.000000], [-0.023163, 0.075135, 0.308934, 0.635819, 0.816402, 0.000000, 0.000000, 0.000000]]
pre_activation_mean: [-1.631290, -2.528596, 0.837148, 0.053241, -0.036688, 2.514904]
pre_activation_std: [1.471626, 1.757832, 1.223700, 1.361698, 0.879932, 1.757702]
### 2
mean: [-0.944309, -0.874625, -0.081479, -0.195549, -0.994316, -0.854148]
std: [0.552041, 0.639786, 0.666615, 0.474535, 0.555104, 0.687490]
fourier: [[9.122581, 9.132873, 9.457256, 9.770282, 84.987819], [10.001342, 10.343601, 10.833623, 11.980032, 78.716229], [9.009985, 9.104190, 11.300221, 12.284331, 13.499955], [7.115601, 7.356920, 7.726906, 8.930107, 17.599370], [8.852974, 9.215624, 9.528296, 9.762114, 89.488440], [9.700139, 12.921988, 14.575733, 15.906211, 76.873319]]
input_correlations: [[0.079209, -0.002454, 0.267867, 0.337875, -0.050463, -0.930649, 0.000000, 0.000000], [-0.117131, -0.089945, 0.330534, 0.191258, -0.064700, -0.986890, 0.000000, 0.000000], [-0.188524, 0.036618, -0.790302, -0.735087, -0.030331, 0.622065, 0.000000, 0.000000], [-0.057056, 0.004395, -0.839228, -0.567991, -0.037274, 0.599918, 0.000000, 0.000000], [-0.033067, -0.029281, 0.295118, 0.273251, -0.119925, -0.968131, 0.000000, 0.000000], [-0.172929, -0.039887, -0.289412, -0.400624, 0.249293, -0.828483, 0.000000, 0.000000]]
pre_activation_mean: [-0.944309, -0.874625, -0.081479, -0.195549, -0.994316, -0.854148]
pre_activation_std: [0.552041, 0.639786, 0.666615, 0.474535, 0.555104, 0.687490]
### 4
mean: [0.639303, -0.258020, 0.074912, 0.484175, -0.270284, 0.647514]
std: [0.237054, 0.161263, 0.095051, 0.186506, 0.102060, 0.214865]
fourier: [[3.423729, 3.910036, 5.009370, 5.303725, 57.537244], [2.209863, 2.854338, 3.448457, 3.575369, 23.221828], [1.487728, 1.516169, 1.710024, 1.954751, 6.742040], [2.779930, 2.870213, 4.056184, 4.189021, 43.575703], [1.351437, 1.888195, 1.997286, 2.404784, 24.325536], [3.056279, 3.582239, 4.670771, 4.678326, 58.276240]]
input_correlations: [[-0.728974, -0.560295, -0.977332, -0.940117, -0.838491, -0.412604, 0.000000, 0.000000], [0.635399, 0.484220, 0.976594, 0.965209, 0.755736, 0.469757, 0.000000, 0.000000], [-0.891335, -0.767632, -0.865163, -0.794811, -0.954828, -0.374935, 0.000000, 0.000000], [-0.663910, -0.459915, -0.995238, -0.973928, -0.773974, -0.344535, 0.000000, 0.000000], [0.659896, 0.508310, 0.958533, 0.917778, 0.794029, 0.533519, 0.000000, 0.000000], [-0.690817, -0.529437, -0.981619, -0.964187, -0.800077, -0.408814, 0.000000, 0.000000]]
pre_activation_mean: [0.639303, -0.258020, 0.074912, 0.484175, -0.270284, 0.647514]
pre_activation_std: [0.237054, 0.161263, 0.095051, 0.186506, 0.102060, 0.214865]
### 6
mean: [0.920563, -0.347956, -0.239090, 0.854713, -0.035297, 0.751846]
std: [0.268177, 0.159305, 0.090287, 0.221844, 0.027813, 0.206618]
fourier: [[3.897445, 4.455607, 5.760717, 5.979833, 82.850683], [2.337978, 2.694966, 3.345698, 3.570519, 31.316039], [1.354782, 1.474538, 1.888970, 2.041866, 21.518054], [3.302348, 3.613957, 4.784061, 4.957684, 76.924221], [0.456506, 0.481598, 0.546790, 0.557108, 3.176719], [2.974114, 3.385904, 4.480111, 4.625438, 67.666133]]
input_correlations: [[0.994778, -0.963643, 0.913625, 0.989218, -0.968509, 0.997042, 0.000000, 0.000000], [-0.995579, 0.959512, -0.931953, -0.981101, 0.967762, -0.994337, 0.000000, 0.000000], [-0.997733, 0.949213, -0.942520, -0.981019, 0.962122, -0.990403, 0.000000, 0.000000], [0.997487, -0.936888, 0.909016, 0.993083, -0.944832, 0.998090, 0.000000, 0.000000], [0.897247, -0.700763, 0.809261, 0.899160, -0.721259, 0.882824, 0.000000, 0.000000], [0.993361, -0.964623, 0.901547, 0.992817, -0.969664, 0.995804, 0.000000, 0.000000]]
pre_activation_mean: [0.920563, -0.347956, -0.239090, 0.854713, -0.035297, 0.751846]
pre_activation_std: [0.268177, 0.159305, 0.090287, 0.221844, 0.027813, 0.206618]
### 8
mean: [-0.731464]
std: [0.313777]
fourier: [[4.634135, 5.136416, 6.718650, 7.046784, 65.831759]]
input_correlations: [[-0.999795, 0.955090, 0.980979, -0.997415, -0.897107, -0.999537, 0.000000, 0.000000]]
pre_activation_mean: [-0.731464]
pre_activation_std: [0.313777]
## Task
Analyze this model and identify which patterns it classifies as positive.
Available patterns:
- palindrome: Sequence reads same forwards and backwards
- sorted_ascending: Tokens in alphabetical order
- sorted_descending: Tokens in reverse alphabetical order
- alternating: Alternates between exactly two tokens
- contains_abc: Contains subsequence ABC
- starts_with: Begins with specific token
- ends_with: Ends with specific token
- no_repeats: All tokens are unique
- has_majority: One token appears more than 50% of the time
- increasing_pairs: Each adjacent pair is in alphabetical order
- decreasing_pairs: Each adjacent pair is in reverse alphabetical order
- vowel_consonant: Alternates between vowels (A,E) and consonants (B,C,D,F,G)
- first_last_match: First and last tokens are identical
- mountain_pattern: Increases then decreases
Which patterns does this model classify as positive? List them separated by commas.
increasing_pairs
|
{"neuron_activations": {"0": {"neuron_profiles": {"0": {"mean": 0.7716683149337769, "std": 0.2484787553548813, "fourier": [3.683442312671083, 4.096419263756906, 5.289386864136277, 5.572375818787589, 69.45014378335327], "input_correlations": [0.11257173187825369, -0.5447679903454944, 0.07214817149507947, -0.9008338169985665, -0.3233763014197621, 0.0, 0.0, 0.0], "pre_activation_mean": -1.6312901973724365, "pre_activation_std": 1.471625566482544}, "1": {"mean": -0.11724137514829636, "std": 0.057827357202768326, "fourier": [0.9107793744516787, 0.960364516539049, 1.2160985124463215, 1.2365433868902438, 10.551723155425861], "input_correlations": [-0.42461252070687977, -0.4207263596492877, -0.6736936707925404, -0.6740625816802979, -0.45225951196042974, 0.0, 0.0, 0.0], "pre_activation_mean": -2.5285961627960205, "pre_activation_std": 1.7578321695327759}, "2": {"mean": -0.09382465481758118, "std": 0.034550122916698456, "fourier": [0.5093669500151653, 0.569356930665728, 0.7247912089048323, 0.7673880673877193, 8.444218133343384], "input_correlations": [0.8771251046941255, 0.49367005725344654, 0.3853896064012669, -0.2874217805849516, -0.13408726878848054, 0.0, 0.0, 0.0], "pre_activation_mean": 0.83714759349823, "pre_activation_std": 1.2237001657485962}, "3": {"mean": 0.6978365182876587, "std": 0.20943893492221832, "fourier": [3.1731894736313007, 3.365313877049351, 4.457819747960329, 4.671735473096148, 62.80529084056616], "input_correlations": [0.6920818734208215, -0.09107309383577242, 0.6290418011112913, -0.5104759134299356, -0.026059999324192062, 0.0, 0.0, 0.0], "pre_activation_mean": 0.053240980952978134, "pre_activation_std": 1.361697793006897}, "4": {"mean": -0.016843587160110474, "std": 0.012861663475632668, "fourier": [0.21048487520740836, 0.22365357289506074, 0.2512633338813914, 0.25639970123535416, 1.5159228404809255], "input_correlations": [-0.6531241644738266, -0.0036220315778205687, 0.1378867089134902, 0.41102832610671086, -0.5272297660805352, 0.0, 0.0, 0.0], "pre_activation_mean": -0.03668815642595291, "pre_activation_std": 0.8799323439598083}, "5": {"mean": 0.5931459069252014, "std": 0.18187014758586884, "fourier": [2.6576448835019044, 2.951954811514025, 3.919564525129474, 4.096936558920846, 53.383130800910294], "input_correlations": [-0.023163261680358496, 0.07513537312876972, 0.308933624304414, 0.6358186372242725, 0.816401942968767, 0.0, 0.0, 0.0], "pre_activation_mean": 2.5149035453796387, "pre_activation_std": 1.757702112197876}}, "layer_info": {"num_neurons": 6, "num_examples": 90, "profile_methods": ["mean", "std", "fourier", "input_correlations", "pre_activation_mean", "pre_activation_std"]}}, "2": {"neuron_profiles": {"0": {"mean": -0.9443090558052063, "std": 0.5520409345626831, "fourier": [9.12258077560523, 9.132873001503166, 9.457256385319637, 9.770282032309153, 84.9878193885088], "input_correlations": [0.0792088114956869, -0.002453948878093915, 0.2678671881388953, 0.33787518666209104, -0.050463375819392246, -0.9306492337222957, 0.0, 0.0], "pre_activation_mean": -0.9443090558052063, "pre_activation_std": 0.5520409345626831}, "1": {"mean": -0.8746247291564941, "std": 0.6397863030433655, "fourier": [10.001342177985888, 10.343600623739063, 10.83362301526622, 11.980032387735307, 78.71622946858406], "input_correlations": [-0.1171308517429666, -0.08994509590119686, 0.33053446946577514, 0.19125811209280227, -0.06470044736462635, -0.9868898131033919, 0.0, 0.0], "pre_activation_mean": -0.8746247291564941, "pre_activation_std": 0.6397863030433655}, "2": {"mean": -0.08147923648357391, "std": 0.6666154265403748, "fourier": [9.009984532328966, 9.104189510521346, 11.300220880486277, 12.284330742823473, 13.499955327113106], "input_correlations": [-0.18852402149083194, 0.036617547828885016, -0.7903022088910153, -0.7350869966063697, -0.03033138623134966, 0.6220651036667094, 0.0, 0.0], "pre_activation_mean": -0.08147923648357391, "pre_activation_std": 0.6666154265403748}, "3": {"mean": -0.19554856419563293, "std": 0.4745346009731293, "fourier": [7.115601223459866, 7.356919958025496, 7.726905662986792, 8.930106756608923, 17.599370166659355], "input_correlations": [-0.057055509142991796, 0.004395208584066798, -0.8392276263093021, -0.5679910769266028, -0.03727406335101797, 0.5999180144252586, 0.0, 0.0], "pre_activation_mean": -0.19554856419563293, "pre_activation_std": 0.4745346009731293}, "4": {"mean": -0.9943159222602844, "std": 0.5551037192344666, "fourier": [8.852974441270051, 9.215623760278643, 9.528296002015573, 9.76211434388638, 89.48843997716904], "input_correlations": [-0.033066550881311205, -0.029281417763389696, 0.2951177649323759, 0.27325099879668857, -0.11992465658371433, -0.9681314461480975, 0.0, 0.0], "pre_activation_mean": -0.9943159222602844, "pre_activation_std": 0.5551037192344666}, "5": {"mean": -0.8541480302810669, "std": 0.6874903440475464, "fourier": [9.70013907713918, 12.921988030502138, 14.575733448828675, 15.906211107438725, 76.87331908941269], "input_correlations": [-0.1729293938501054, -0.03988703454953977, -0.2894122742990732, -0.4006239431476659, 0.2492931036191463, -0.8284833371491043, 0.0, 0.0], "pre_activation_mean": -0.8541480302810669, "pre_activation_std": 0.6874903440475464}}, "layer_info": {"num_neurons": 6, "num_examples": 90, "profile_methods": ["mean", "std", "fourier", "input_correlations", "pre_activation_mean", "pre_activation_std"]}}, "4": {"neuron_profiles": {"0": {"mean": 0.639302670955658, "std": 0.23705367743968964, "fourier": [3.423728960371199, 3.9100362705260454, 5.009369732931313, 5.303724855162597, 57.53724366426468], "input_correlations": [-0.7289740718268184, -0.5602946996472873, -0.9773319074947826, -0.9401174799111324, -0.838490669579978, -0.4126039834320674, 0.0, 0.0], "pre_activation_mean": 0.639302670955658, "pre_activation_std": 0.23705367743968964}, "1": {"mean": -0.2580203413963318, "std": 0.16126328706741333, "fourier": [2.209862934183298, 2.8543383921002548, 3.4484569738872985, 3.5753690323816683, 23.22182836383581], "input_correlations": [0.6353986594372281, 0.4842201211841011, 0.9765944608078524, 0.9652088925147988, 0.7557362156685029, 0.469756861727435, 0.0, 0.0], "pre_activation_mean": -0.2580203413963318, "pre_activation_std": 0.16126328706741333}, "2": {"mean": 0.07491155713796616, "std": 0.09505139291286469, "fourier": [1.4877282508832563, 1.5161691615593964, 1.710023537783643, 1.954750687256663, 6.742039993405342], "input_correlations": [-0.8913350007508171, -0.7676320636861471, -0.8651630243635924, -0.7948113740612681, -0.9548279540580551, -0.3749345876373998, 0.0, 0.0], "pre_activation_mean": 0.07491155713796616, "pre_activation_std": 0.09505139291286469}, "3": {"mean": 0.484174519777298, "std": 0.1865064650774002, "fourier": [2.7799295239498316, 2.87021300532067, 4.056183987087885, 4.189021069580887, 43.575703367590904], "input_correlations": [-0.6639098648168492, -0.4599146730634236, -0.9952381125823194, -0.9739280674578714, -0.7739738230511131, -0.3445349922091809, 0.0, 0.0], "pre_activation_mean": 0.484174519777298, "pre_activation_std": 0.1865064650774002}, "4": {"mean": -0.2702837288379669, "std": 0.10206006467342377, "fourier": [1.351436959754728, 1.8881946755817274, 1.997285867263968, 2.40478406165908, 24.325535878539085], "input_correlations": [0.6598955500632732, 0.508310192003263, 0.9585325458148676, 0.9177776419485695, 0.7940293966575076, 0.5335188437950085, 0.0, 0.0], "pre_activation_mean": -0.2702837288379669, "pre_activation_std": 0.10206006467342377}, "5": {"mean": 0.647513747215271, "std": 0.21486490964889526, "fourier": [3.056279459291334, 3.58223929521691, 4.670770743904758, 4.67832550574282, 58.27624014019966], "input_correlations": [-0.6908174467132377, -0.5294371268148582, -0.9816188264261725, -0.9641873563028124, -0.8000768743444594, -0.40881395426184447, 0.0, 0.0], "pre_activation_mean": 0.647513747215271, "pre_activation_std": 0.21486490964889526}}, "layer_info": {"num_neurons": 6, "num_examples": 90, "profile_methods": ["mean", "std", "fourier", "input_correlations", "pre_activation_mean", "pre_activation_std"]}}, "6": {"neuron_profiles": {"0": {"mean": 0.9205629825592041, "std": 0.26817718148231506, "fourier": [3.897444506977477, 4.455606652382648, 5.760717187743157, 5.979832592611717, 82.8506831228733], "input_correlations": [0.9947776136540647, -0.9636430456391647, 0.9136254377746748, 0.9892175237535309, -0.9685089832802414, 0.9970418453757147, 0.0, 0.0], "pre_activation_mean": 0.9205629825592041, "pre_activation_std": 0.26817718148231506}, "1": {"mean": -0.34795600175857544, "std": 0.15930496156215668, "fourier": [2.3379776908977923, 2.6949656616372075, 3.345698091226442, 3.5705188080638166, 31.316038873046637], "input_correlations": [-0.9955787546855285, 0.9595117261491646, -0.9319527016155659, -0.9811012398581954, 0.9677615233461996, -0.9943371921539614, 0.0, 0.0], "pre_activation_mean": -0.34795600175857544, "pre_activation_std": 0.15930496156215668}, "2": {"mean": -0.2390895038843155, "std": 0.09028664976358414, "fourier": [1.3547823435318973, 1.4745382687905557, 1.8889695876107155, 2.041865619531352, 21.518054323270917], "input_correlations": [-0.9977332049745624, 0.9492132361193745, -0.9425199208947121, -0.9810190146452559, 0.9621222334619635, -0.9904033137941739, 0.0, 0.0], "pre_activation_mean": -0.2390895038843155, "pre_activation_std": 0.09028664976358414}, "3": {"mean": 0.854713499546051, "std": 0.22184352576732635, "fourier": [3.302347570245067, 3.6139574825375873, 4.784061068293818, 4.9576843639070445, 76.92422132194042], "input_correlations": [0.9974867370380506, -0.936888369936164, 0.909016167300624, 0.9930832557822644, -0.9448317414070352, 0.9980900362160634, 0.0, 0.0], "pre_activation_mean": 0.854713499546051, "pre_activation_std": 0.22184352576732635}, "4": {"mean": -0.03529687598347664, "std": 0.027812648564577103, "fourier": [0.45650598142635457, 0.48159794360169594, 0.5467896957486511, 0.5571079578102212, 3.176718957722187], "input_correlations": [0.8972473039391013, -0.7007627991993799, 0.8092607322804313, 0.8991598488541358, -0.7212586531743643, 0.8828241253394647, 0.0, 0.0], "pre_activation_mean": -0.03529687598347664, "pre_activation_std": 0.027812648564577103}, "5": {"mean": 0.7518458962440491, "std": 0.20661775767803192, "fourier": [2.9741142547014783, 3.385903979325929, 4.480110825593129, 4.625437831627566, 67.6661327406764], "input_correlations": [0.9933607991480088, -0.9646227418075701, 0.901547454521621, 0.9928167519298979, -0.9696637953694563, 0.9958041201310155, 0.0, 0.0], "pre_activation_mean": 0.7518458962440491, "pre_activation_std": 0.20661775767803192}}, "layer_info": {"num_neurons": 6, "num_examples": 90, "profile_methods": ["mean", "std", "fourier", "input_correlations", "pre_activation_mean", "pre_activation_std"]}}, "8": {"neuron_profiles": {"0": {"mean": -0.7314639687538147, "std": 0.3137769103050232, "fourier": [4.63413459444491, 5.136416343775484, 6.718649947569788, 7.046784099477567, 65.83175876736641], "input_correlations": [-0.9997949616034644, 0.9550899781626709, 0.9809785836949312, -0.997415172967218, -0.8971068843298992, -0.9995374811086655, 0.0, 0.0], "pre_activation_mean": -0.7314639687538147, "pre_activation_std": 0.3137769103050232}}, "layer_info": {"num_neurons": 1, "num_examples": 90, "profile_methods": ["mean", "std", "fourier", "input_correlations", "pre_activation_mean", "pre_activation_std"]}}}, "model_config": {"vocab_size": 10, "sequence_length": 5, "num_layers": 4, "neurons_per_layer": 6, "activation_type": "gelu", "dropout_rate": 0.0, "precision": "float32", "input_size": 5, "input_format": "integer_indices"}}
|
{"config": {"vocab_size": 10, "sequence_length": 5, "num_layers": 4, "neurons_per_layer": 6, "activation_type": "gelu", "dropout_rate": 0.0, "precision": "float32", "input_size": 5, "input_format": "integer_indices"}, "weights": {"network.0.weight": [[0.158559, -0.335233, 0.126131, -0.499865, -0.23349], [-0.213482, -0.004312, -0.494812, -0.546378, -0.184118], [0.458597, 0.209909, 0.083425, -0.186846, -0.1671], [0.431848, -0.227313, 0.379992, -0.230143, -0.152062], [-0.286215, -0.022738, 0.216078, 0.201012, -0.277593], [-0.141409, -0.096191, 0.202471, 0.464079, 0.681626]], "network.0.bias": [-0.121307, 0.177289, 0.316453, -0.182647, -0.179134, 0.607356], "network.2.weight": [[0.107749, 0.297725, -0.156689, 0.309806, 0.031923, -0.301118], [-0.325031, -0.156924, 0.021685, 0.099026, 0.134414, -0.344987], [0.031175, 0.180637, -0.272168, -0.296353, -0.418316, 0.193799], [-0.020162, -0.271397, -0.383145, 0.0486, -0.357388, 0.116378], [-0.147728, 0.265535, -0.100847, 0.228985, -0.010514, -0.304863], [0.123871, 0.043483, -0.142077, -0.19538, 0.347935, -0.355341]], "network.2.bias": [-0.209767, -0.134005, -0.097006, -0.111801, -0.260934, 0.159058], "network.4.weight": [[-0.539065, -0.258701, -0.563003, -0.027256, -0.405867, -0.457929], [0.108491, 0.071989, 0.197311, 0.398875, 0.198712, 0.567761], [-0.586414, -0.426316, -0.198635, 0.101442, -0.293247, -0.038306], [-0.412374, 0.087111, -0.380578, -0.227341, -0.193997, -0.190073], [-0.015035, 0.011753, 0.352265, -0.171835, 0.142329, 0.49398], [-0.36703, -0.189714, -0.201056, -0.589402, -0.542615, -0.420302]], "network.4.bias": [0.490543, -0.161585, -0.074235, 0.431325, -0.233521, 0.473756], "network.6.weight": [[0.330843, -0.496304, 0.297326, 0.235714, -0.52927, 0.643976], [-0.43654, 0.396683, -0.273536, 0.230112, 0.108594, -0.388997], [-0.366902, 0.186605, -0.259706, -0.033481, 0.041479, 0.041425], [0.616239, 0.114437, -0.147513, 0.1843, 0.163883, 0.612953], [0.122939, 0.354638, 0.026967, 0.241769, 0.301418, 0.026183], [0.322544, -0.344163, 0.003869, 0.504485, -0.652614, 0.191733]], "network.6.bias": [0.246192, 0.038561, -0.035595, 0.220497, -0.129025, 0.22669], "network.8.weight": [[-0.525313, 0.215439, 0.317084, -0.252575, -0.082643, -0.588126]], "network.8.bias": [0.25262]}}
|
{"training_history": [{"stage": "degraded", "epoch": 0, "global_epoch": 0, "train_loss": 0.7226835191249847, "train_acc": 0.425, "val_loss": 0.6928030252456665, "val_acc": 0.5}, {"stage": "degraded", "epoch": 1, "global_epoch": 1, "train_loss": 0.7008070349693298, "train_acc": 0.425, "val_loss": 0.6858577728271484, "val_acc": 0.88}, {"stage": "degraded", "epoch": 2, "global_epoch": 2, "train_loss": 0.6865839660167694, "train_acc": 0.65, "val_loss": 0.680361270904541, "val_acc": 0.5}, {"stage": "degraded", "epoch": 3, "global_epoch": 3, "train_loss": 0.6669614911079407, "train_acc": 0.575, "val_loss": 0.6703599691390991, "val_acc": 0.5}, {"stage": "degraded", "epoch": 4, "global_epoch": 4, "train_loss": 0.6495357751846313, "train_acc": 0.575, "val_loss": 0.6441338062286377, "val_acc": 0.5}, {"stage": "improved", "epoch": 0, "global_epoch": 5, "train_loss": 0.6419384181499481, "train_acc": 0.515, "val_loss": 0.5758852362632751, "val_acc": 0.84}, {"stage": "improved", "epoch": 1, "global_epoch": 6, "train_loss": 0.5895566642284393, "train_acc": 0.77, "val_loss": 0.5046849250793457, "val_acc": 0.9}, {"stage": "improved", "epoch": 2, "global_epoch": 7, "train_loss": 0.5128251612186432, "train_acc": 0.835, "val_loss": 0.440817654132843, "val_acc": 0.88}, {"stage": "improved", "epoch": 3, "global_epoch": 8, "train_loss": 0.46379518508911133, "train_acc": 0.84, "val_loss": 0.4002281427383423, "val_acc": 0.9}, {"stage": "improved", "epoch": 4, "global_epoch": 9, "train_loss": 0.4053000509738922, "train_acc": 0.855, "val_loss": 0.42917799949645996, "val_acc": 0.82}, {"stage": "improved", "epoch": 5, "global_epoch": 10, "train_loss": 0.3741624057292938, "train_acc": 0.85, "val_loss": 0.3529716432094574, "val_acc": 0.88}, {"stage": "improved", "epoch": 6, "global_epoch": 11, "train_loss": 0.34253551065921783, "train_acc": 0.855, "val_loss": 0.3267880976200104, "val_acc": 0.84}, {"stage": "improved", "epoch": 7, "global_epoch": 12, "train_loss": 0.33100198209285736, "train_acc": 0.845, "val_loss": 0.3366655111312866, "val_acc": 0.86}, {"stage": "improved", "epoch": 8, "global_epoch": 13, "train_loss": 0.33217255771160126, "train_acc": 0.85, "val_loss": 0.4257589280605316, "val_acc": 0.82}, {"stage": "improved", "epoch": 9, "global_epoch": 14, "train_loss": 0.372518852353096, "train_acc": 0.86, "val_loss": 0.3635927438735962, "val_acc": 0.84}], "summary": {"total_epochs": 15, "degraded_epochs": 5, "improved_epochs": 10, "patterns": ["increasing_pairs"], "degraded_stage": {"initial_val_loss": 0.6928030252456665, "final_val_loss": 0.6441338062286377, "initial_val_acc": 0.5, "final_val_acc": 0.5, "best_val_acc": 0.5}, "improved_stage": {"initial_val_loss": 0.5758852362632751, "final_val_loss": 0.3635927438735962, "initial_val_acc": 0.84, "final_val_acc": 0.84, "best_val_acc": 0.9, "best_epoch": 6}, "improvement": 0.4, "first_improvement_epoch": 4}}
|
2
|
{"target_pattern": "alternating", "degraded_accuracy": 0.52, "improved_accuracy": 0.94, "improvement": 0.41999999999999993, "model_config": {"vocab_size": 10, "sequence_length": 5, "num_layers": 5, "neurons_per_layer": 7, "activation_type": "gelu", "dropout_rate": 0.0, "random_seed": 9451, "learning_rate": 0.07352170370310572, "batch_size": 128, "num_epochs": 15, "patience": 3}, "corruption_stats": {"target_pattern": "alternating", "corruption_rate": 0.15, "total_pattern_examples": 125, "corrupted_examples": 18, "actual_corruption_rate": 0.144}, "selected_patterns": ["alternating"], "precision": "float16", "quantization": "none", "tasks_included": {"modification": false, "classification": true}}
|
## Model Architecture
Input Size: 5 (integer indices for 5 sequence positions, vocab size 10)
Hidden Layers: 5
Neurons per Layer: 7
Activation Function: gelu
Dropout Rate: 0.0
## Model Weights
The trained model weights:
{
"network.0.weight": [
[
-0.498046,
0.016497,
0.27912,
0.377212,
0.46275
],
[
-0.67476,
-0.410803,
-0.048854,
-0.213823,
-0.264684
],
[
-0.222577,
-0.287405,
-0.23385,
0.043336,
0.653931
],
[
-0.486114,
0.468888,
-1.051172,
0.336788,
-0.250392
],
[
-0.109708,
0.679346,
0.25548,
0.479164,
0.238863
],
[
-0.074179,
-0.82374,
-0.286715,
-0.65726,
-0.382477
],
[
0.546925,
-0.782979,
0.435782,
-1.139192,
-0.03333
]
],
"network.0.bias": [
0.445738,
-0.026591,
-0.432475,
0.339879,
0.215415,
-0.407969,
-0.353849
],
"network.2.weight": [
[
-0.221768,
0.295291,
-0.566029,
0.607401,
0.072632,
0.07435,
0.29675
],
[
0.390566,
-0.283276,
0.373778,
-0.352655,
-0.0111,
-0.21586,
-0.6314
],
[
-0.080529,
0.045784,
-0.381584,
0.322332,
-0.208936,
-0.470206,
0.61041
],
[
0.062086,
-0.112544,
0.891917,
-0.861448,
0.654051,
0.532569,
-0.493258
],
[
0.119339,
-0.357433,
-0.249549,
0.668707,
-0.333574,
-0.504545,
0.615059
],
[
0.495847,
-0.115094,
0.920688,
-0.522015,
0.238003,
0.113081,
-0.277181
],
[
0.077411,
0.193998,
-0.19702,
0.508568,
-0.220905,
-0.061286,
0.457584
]
],
"network.2.bias": [
0.014641,
0.12025,
0.247411,
0.211555,
0.010649,
0.381745,
0.116955
],
"network.4.weight": [
[
-0.334707,
0.304441,
-0.279457,
-0.029286,
-0.15341,
0.541405,
-0.115904
],
[
-0.123486,
0.095578,
-0.213479,
0.851065,
-0.35692,
0.358511,
-0.724027
],
[
-0.291779,
0.347184,
-0.297729,
0.891535,
-0.534177,
0.369992,
-0.322145
],
[
0.030368,
-0.166714,
-0.196908,
0.856735,
-0.289794,
-0.05414,
-0.431291
],
[
-0.915498,
-0.114461,
-0.226982,
-0.092932,
0.474361,
-0.311556,
-0.199999
],
[
-0.154811,
-0.097878,
-0.139779,
-0.258556,
0.247388,
-0.34283,
0.273014
],
[
-0.078105,
-0.552093,
0.090498,
-0.087391,
-0.194507,
-0.611112,
0.47962
]
],
"network.4.bias": [
-0.242336,
0.268976,
0.301511,
0.154612,
-0.259087,
-0.519134,
-0.128069
],
"network.6.weight": [
[
-0.535836,
-0.35537,
-0.120437,
-1.133117,
0.152765,
-0.882727,
-0.592671
],
[
0.327513,
0.483762,
0.099797,
0.339522,
-0.189741,
0.550959,
0.274193
],
[
-0.092553,
-0.219723,
-0.335909,
-0.606311,
0.260152,
-0.198496,
-0.242835
],
[
-0.003381,
0.739864,
0.1703,
0.595127,
0.020794,
0.185502,
0.464682
],
[
-0.791922,
-0.703602,
-0.118004,
-0.824964,
-0.099171,
-0.932998,
-0.48668
],
[
0.254018,
-0.055933,
-0.332949,
-0.601312,
-0.405981,
-0.585619,
-0.340154
],
[
0.460524,
0.420648,
0.601001,
0.381768,
-0.553487,
-0.220082,
0.208571
]
],
"network.6.bias": [
0.540844,
-0.148655,
0.610657,
-0.03453,
0.5719,
0.157429,
0.268893
],
"network.8.weight": [
[
0.35699,
0.228855,
0.077237,
-0.168761,
0.30045,
0.442372,
-0.401396
],
[
-0.535887,
0.163712,
-0.438504,
0.21535,
-0.319709,
-0.399919,
0.398314
],
[
0.488905,
-0.961267,
0.414501,
-1.096139,
0.971603,
0.120462,
0.042228
],
[
-0.275928,
0.513421,
-0.259565,
0.747401,
-0.43952,
-0.29741,
0.669777
],
[
1.050184,
-0.602593,
0.257354,
-0.810513,
0.361708,
-0.001857,
-0.303692
],
[
0.383587,
-0.341232,
0.22948,
-0.308506,
0.664184,
0.148047,
-0.290964
],
[
-0.222798,
-0.305822,
0.543264,
-0.133134,
-0.07813,
-0.045752,
-0.298768
]
],
"network.8.bias": [
0.070521,
0.192318,
0.411326,
-0.030442,
0.589452,
0.698715,
0.117217
],
"network.10.weight": [
[
0.270498,
-0.422512,
0.563722,
-0.557227,
0.381801,
0.596703,
0.221141
]
],
"network.10.bias": [
0.532356
]
}
## Activation Signature
### 0
mean: [0.109175, 2.818719, 0.589725, 6.346766, 0.542347, 0.480877, 0.011254]
std: [0.247611, 2.639819, 0.889538, 5.995561, 0.820108, 0.744105, 0.073399]
fourier: [[3.540042, 3.597278, 3.774095, 3.784434, 9.825795], [39.682978, 39.826686, 41.805479, 48.802526, 253.684718], [12.702743, 13.283982, 13.614669, 14.183304, 53.075276], [90.197907, 92.823411, 96.245637, 112.443495, 571.208904], [11.738320, 12.177748, 12.534344, 13.020932, 48.811215], [10.564774, 10.826019, 11.350033, 11.513960, 43.278934], [1.029049, 1.046790, 1.054027, 1.075673, 1.100485]]
input_correlations: [[-0.451123, 0.046684, 0.266320, 0.620399, 0.572941, 0.000000, 0.000000, 0.000000], [-0.838947, -0.674249, -0.393040, -0.344511, -0.401531, 0.000000, 0.000000, 0.000000], [-0.382665, -0.516378, -0.311942, 0.077754, 0.714372, 0.000000, 0.000000, 0.000000], [-0.535981, 0.117182, -0.826412, 0.346737, -0.357020, 0.000000, 0.000000, 0.000000], [0.187975, 0.796551, 0.397498, 0.735648, 0.322375, 0.000000, 0.000000, 0.000000], [-0.326604, -0.784723, -0.404664, -0.723476, -0.386287, 0.000000, 0.000000, 0.000000], [0.299109, -0.500941, 0.250883, -0.870616, -0.010414, 0.000000, 0.000000, 0.000000]]
pre_activation_mean: [1.818696, -2.501892, -0.822215, -1.193020, 3.172617, -4.481411, -2.673734]
pre_activation_std: [1.579749, 2.047694, 1.457211, 2.725842, 2.033733, 2.734886, 3.342324]
### 2
mean: [0.046701, 0.576169, -0.317537, 2.039889, -0.392796, 1.912346, -0.148570]
std: [0.888122, 0.989955, 0.912627, 1.807250, 1.004672, 1.580047, 0.719862]
fourier: [[12.138823, 13.292778, 14.524030, 15.139685, 15.279536], [15.186599, 17.105412, 17.689741, 19.631211, 51.855181], [14.037748, 14.046630, 14.324026, 14.595218, 28.578339], [26.457611, 27.907662, 29.576852, 31.957664, 183.589983], [13.966606, 13.971229, 14.065081, 18.619016, 35.351619], [25.023684, 25.756558, 27.793710, 30.238611, 172.111113], [10.046724, 10.200105, 10.219335, 13.231826, 13.371273]]
input_correlations: [[-0.461015, 0.041048, -0.674408, 0.665587, 0.015679, -0.026081, 0.218339, 0.000000], [0.705537, -0.012980, 0.605976, -0.232086, 0.364725, 0.193617, -0.582957, 0.000000], [-0.587325, -0.184903, -0.475390, 0.133326, -0.631587, -0.354830, 0.671833, 0.000000], [0.666243, 0.318816, 0.543180, -0.234899, 0.761401, 0.527284, -0.327399, 0.000000], [-0.301831, -0.238717, -0.241546, 0.411012, -0.552761, -0.299754, 0.603606, 0.000000], [0.860624, 0.192439, 0.755434, -0.184844, 0.600155, 0.456471, -0.196839, 0.000000], [-0.282423, -0.179062, -0.250629, 0.467428, -0.477218, -0.244941, 0.617437, 0.000000]]
pre_activation_mean: [0.046701, 0.576169, -0.317537, 2.039889, -0.392796, 1.912346, -0.148570]
pre_activation_std: [0.888122, 0.989955, 0.912627, 1.807250, 1.004672, 1.580047, 0.719862]
### 4
mean: [0.746120, 2.521878, 2.783477, 1.575671, -1.260104, -1.711524, -1.743184]
std: [1.283069, 2.550721, 2.820377, 1.641124, 0.716969, 1.139416, 1.624784]
fourier: [[19.312209, 22.184224, 22.666084, 24.525093, 67.150790], [37.105487, 37.260093, 39.892260, 46.105740, 226.969044], [39.817725, 41.468382, 45.144917, 51.989691, 250.512958], [22.809361, 24.541908, 26.987565, 27.108505, 141.810437], [9.650672, 9.732897, 11.875901, 13.468849, 113.409391], [17.006246, 18.441666, 18.554156, 21.710884, 154.037166], [26.764447, 27.350296, 27.355708, 31.821535, 156.886599]]
input_correlations: [[-0.543267, 0.955710, -0.523349, 0.892413, -0.605463, 0.958977, -0.650078, 0.000000], [-0.480197, 0.848656, -0.604231, 0.976584, -0.650178, 0.915328, -0.705129, 0.000000], [-0.490045, 0.870489, -0.586999, 0.972847, -0.639142, 0.929122, -0.694126, 0.000000], [-0.433454, 0.737188, -0.649048, 0.973975, -0.671331, 0.835642, -0.727716, 0.000000], [-0.247012, -0.735184, 0.249700, -0.749873, 0.061189, -0.817258, 0.073698, 0.000000], [0.396515, -0.911086, 0.503402, -0.972666, 0.531289, -0.971453, 0.584345, 0.000000], [0.389659, -0.968604, 0.433768, -0.918371, 0.466042, -0.993937, 0.513378, 0.000000]]
pre_activation_mean: [0.746120, 2.521878, 2.783477, 1.575671, -1.260104, -1.711524, -1.743184]
pre_activation_std: [1.283069, 2.550721, 2.820377, 1.641124, 0.716969, 1.139416, 1.624784]
### 6
mean: [-2.926606, 2.185494, -1.996075, 3.310276, -3.481503, -1.599070, 4.177769]
std: [3.492094, 2.311165, 2.440366, 3.152233, 4.134065, 1.736550, 3.639245]
fourier: [[52.995579, 54.260634, 54.370605, 65.286344, 263.394531], [35.559560, 36.446520, 36.623517, 43.913150, 196.694446], [36.873876, 37.382486, 37.483009, 44.934954, 179.646760], [47.777552, 48.084976, 48.366158, 57.825448, 297.924786], [63.570713, 65.435956, 65.853095, 78.891067, 313.335245], [25.612326, 25.700745, 27.770131, 30.456926, 143.916315], [55.995395, 57.506224, 57.941317, 69.320516, 375.999163]]
input_correlations: [[-0.910539, -0.999181, -0.997031, -0.985212, -0.656651, -0.914423, -0.541564, 0.000000], [0.928589, 0.999534, 0.999483, 0.975878, 0.663762, 0.910700, 0.549932, 0.000000], [-0.902853, -0.999007, -0.996195, -0.987820, -0.644743, -0.912997, -0.527320, 0.000000], [0.899977, 0.998738, 0.995634, 0.988753, 0.643445, 0.913824, 0.525930, 0.000000], [-0.931096, -0.999206, -0.999383, -0.974730, -0.670754, -0.909432, -0.553304, 0.000000], [-0.856396, -0.989935, -0.982851, -0.998105, -0.617018, -0.915602, -0.500564, 0.000000], [0.935327, 0.999033, 0.999889, 0.971370, 0.665583, 0.904677, 0.546664, 0.000000]]
pre_activation_mean: [-2.926606, 2.185494, -1.996075, 3.310276, -3.481503, -1.599070, 4.177769]
pre_activation_std: [3.492094, 2.311165, 2.440366, 3.152233, 4.134065, 1.736550, 3.639245]
### 8
mean: [-1.470797, 2.657326, -4.807086, 6.164118, -4.368451, -2.017333, -2.231228]
std: [1.711357, 2.836052, 5.895554, 6.199331, 5.397875, 3.128739, 2.221221]
fourier: [[24.762868, 25.253796, 26.509284, 30.686234, 132.371718], [41.457336, 41.965026, 43.824336, 51.533367, 239.159323], [88.724109, 88.896353, 90.032965, 107.920572, 432.637778], [94.710470, 95.362735, 95.940115, 115.332941, 554.770624], [81.626549, 81.649614, 82.615450, 99.206570, 393.160602], [46.267593, 46.688209, 48.269824, 57.220615, 181.559984], [33.764816, 34.976865, 35.367906, 41.953514, 200.810507]]
input_correlations: [[0.791816, -0.980162, 0.765237, -0.986563, 0.795379, 0.644335, -0.988395, 0.000000], [-0.770603, 0.986423, -0.742702, 0.991343, -0.774460, -0.617672, 0.992876, 0.000000], [0.735720, -0.992971, 0.705313, -0.997647, 0.740223, 0.572373, -0.996011, 0.000000], [-0.716143, 0.996388, -0.685555, 0.998143, -0.720683, -0.551763, 0.998648, 0.000000], [0.735447, -0.993450, 0.705299, -0.997293, 0.739801, 0.573202, -0.996763, 0.000000], [0.757031, -0.989627, 0.728002, -0.994113, 0.761329, 0.599753, -0.994793, 0.000000], [0.689361, -0.998885, 0.658313, -0.997824, 0.694027, 0.522245, -0.999721, 0.000000]]
pre_activation_mean: [-1.470797, 2.657326, -4.807086, 6.164118, -4.368451, -2.017333, -2.231228]
pre_activation_std: [1.711357, 2.836052, 5.895554, 6.199331, 5.397875, 3.128739, 2.221221]
### 10
mean: [-3.336704]
std: [5.493570]
fourier: [[77.920118, 82.233546, 84.681638, 97.724439, 300.303339]]
input_correlations: [[0.721954, -0.990290, 0.833163, -0.983610, 0.832224, 0.829465, 0.486369, 0.000000]]
pre_activation_mean: [-3.336704]
pre_activation_std: [5.493570]
## Task
Analyze this model and identify which patterns it classifies as positive.
Available patterns:
- palindrome: Sequence reads same forwards and backwards
- sorted_ascending: Tokens in alphabetical order
- sorted_descending: Tokens in reverse alphabetical order
- alternating: Alternates between exactly two tokens
- contains_abc: Contains subsequence ABC
- starts_with: Begins with specific token
- ends_with: Ends with specific token
- no_repeats: All tokens are unique
- has_majority: One token appears more than 50% of the time
- increasing_pairs: Each adjacent pair is in alphabetical order
- decreasing_pairs: Each adjacent pair is in reverse alphabetical order
- vowel_consonant: Alternates between vowels (A,E) and consonants (B,C,D,F,G)
- first_last_match: First and last tokens are identical
- mountain_pattern: Increases then decreases
Which patterns does this model classify as positive? List them separated by commas.
|
alternating
|
## Model Architecture
Input Size: 5 (integer indices for 5 sequence positions, vocab size 10)
Hidden Layers: 5
Neurons per Layer: 7
Activation Function: gelu
Dropout Rate: 0.0
## Model Weights
The trained model weights:
{
"network.0.weight": [
[
-0.498046,
0.016497,
0.27912,
0.377212,
0.46275
],
[
-0.67476,
-0.410803,
-0.048854,
-0.213823,
-0.264684
],
[
-0.222577,
-0.287405,
-0.23385,
0.043336,
0.653931
],
[
-0.486114,
0.468888,
-1.051172,
0.336788,
-0.250392
],
[
-0.109708,
0.679346,
0.25548,
0.479164,
0.238863
],
[
-0.074179,
-0.82374,
-0.286715,
-0.65726,
-0.382477
],
[
0.546925,
-0.782979,
0.435782,
-1.139192,
-0.03333
]
],
"network.0.bias": [
0.445738,
-0.026591,
-0.432475,
0.339879,
0.215415,
-0.407969,
-0.353849
],
"network.2.weight": [
[
-0.221768,
0.295291,
-0.566029,
0.607401,
0.072632,
0.07435,
0.29675
],
[
0.390566,
-0.283276,
0.373778,
-0.352655,
-0.0111,
-0.21586,
-0.6314
],
[
-0.080529,
0.045784,
-0.381584,
0.322332,
-0.208936,
-0.470206,
0.61041
],
[
0.062086,
-0.112544,
0.891917,
-0.861448,
0.654051,
0.532569,
-0.493258
],
[
0.119339,
-0.357433,
-0.249549,
0.668707,
-0.333574,
-0.504545,
0.615059
],
[
0.495847,
-0.115094,
0.920688,
-0.522015,
0.238003,
0.113081,
-0.277181
],
[
0.077411,
0.193998,
-0.19702,
0.508568,
-0.220905,
-0.061286,
0.457584
]
],
"network.2.bias": [
0.014641,
0.12025,
0.247411,
0.211555,
0.010649,
0.381745,
0.116955
],
"network.4.weight": [
[
-0.334707,
0.304441,
-0.279457,
-0.029286,
-0.15341,
0.541405,
-0.115904
],
[
-0.123486,
0.095578,
-0.213479,
0.851065,
-0.35692,
0.358511,
-0.724027
],
[
-0.291779,
0.347184,
-0.297729,
0.891535,
-0.534177,
0.369992,
-0.322145
],
[
0.030368,
-0.166714,
-0.196908,
0.856735,
-0.289794,
-0.05414,
-0.431291
],
[
-0.915498,
-0.114461,
-0.226982,
-0.092932,
0.474361,
-0.311556,
-0.199999
],
[
-0.154811,
-0.097878,
-0.139779,
-0.258556,
0.247388,
-0.34283,
0.273014
],
[
-0.078105,
-0.552093,
0.090498,
-0.087391,
-0.194507,
-0.611112,
0.47962
]
],
"network.4.bias": [
-0.242336,
0.268976,
0.301511,
0.154612,
-0.259087,
-0.519134,
-0.128069
],
"network.6.weight": [
[
-0.535836,
-0.35537,
-0.120437,
-1.133117,
0.152765,
-0.882727,
-0.592671
],
[
0.327513,
0.483762,
0.099797,
0.339522,
-0.189741,
0.550959,
0.274193
],
[
-0.092553,
-0.219723,
-0.335909,
-0.606311,
0.260152,
-0.198496,
-0.242835
],
[
-0.003381,
0.739864,
0.1703,
0.595127,
0.020794,
0.185502,
0.464682
],
[
-0.791922,
-0.703602,
-0.118004,
-0.824964,
-0.099171,
-0.932998,
-0.48668
],
[
0.254018,
-0.055933,
-0.332949,
-0.601312,
-0.405981,
-0.585619,
-0.340154
],
[
0.460524,
0.420648,
0.601001,
0.381768,
-0.553487,
-0.220082,
0.208571
]
],
"network.6.bias": [
0.540844,
-0.148655,
0.610657,
-0.03453,
0.5719,
0.157429,
0.268893
],
"network.8.weight": [
[
0.35699,
0.228855,
0.077237,
-0.168761,
0.30045,
0.442372,
-0.401396
],
[
-0.535887,
0.163712,
-0.438504,
0.21535,
-0.319709,
-0.399919,
0.398314
],
[
0.488905,
-0.961267,
0.414501,
-1.096139,
0.971603,
0.120462,
0.042228
],
[
-0.275928,
0.513421,
-0.259565,
0.747401,
-0.43952,
-0.29741,
0.669777
],
[
1.050184,
-0.602593,
0.257354,
-0.810513,
0.361708,
-0.001857,
-0.303692
],
[
0.383587,
-0.341232,
0.22948,
-0.308506,
0.664184,
0.148047,
-0.290964
],
[
-0.222798,
-0.305822,
0.543264,
-0.133134,
-0.07813,
-0.045752,
-0.298768
]
],
"network.8.bias": [
0.070521,
0.192318,
0.411326,
-0.030442,
0.589452,
0.698715,
0.117217
],
"network.10.weight": [
[
0.270498,
-0.422512,
0.563722,
-0.557227,
0.381801,
0.596703,
0.221141
]
],
"network.10.bias": [
0.532356
]
}
## Activation Signature
### 0
mean: [0.109175, 2.818719, 0.589725, 6.346766, 0.542347, 0.480877, 0.011254]
std: [0.247611, 2.639819, 0.889538, 5.995561, 0.820108, 0.744105, 0.073399]
fourier: [[3.540042, 3.597278, 3.774095, 3.784434, 9.825795], [39.682978, 39.826686, 41.805479, 48.802526, 253.684718], [12.702743, 13.283982, 13.614669, 14.183304, 53.075276], [90.197907, 92.823411, 96.245637, 112.443495, 571.208904], [11.738320, 12.177748, 12.534344, 13.020932, 48.811215], [10.564774, 10.826019, 11.350033, 11.513960, 43.278934], [1.029049, 1.046790, 1.054027, 1.075673, 1.100485]]
input_correlations: [[-0.451123, 0.046684, 0.266320, 0.620399, 0.572941, 0.000000, 0.000000, 0.000000], [-0.838947, -0.674249, -0.393040, -0.344511, -0.401531, 0.000000, 0.000000, 0.000000], [-0.382665, -0.516378, -0.311942, 0.077754, 0.714372, 0.000000, 0.000000, 0.000000], [-0.535981, 0.117182, -0.826412, 0.346737, -0.357020, 0.000000, 0.000000, 0.000000], [0.187975, 0.796551, 0.397498, 0.735648, 0.322375, 0.000000, 0.000000, 0.000000], [-0.326604, -0.784723, -0.404664, -0.723476, -0.386287, 0.000000, 0.000000, 0.000000], [0.299109, -0.500941, 0.250883, -0.870616, -0.010414, 0.000000, 0.000000, 0.000000]]
pre_activation_mean: [1.818696, -2.501892, -0.822215, -1.193020, 3.172617, -4.481411, -2.673734]
pre_activation_std: [1.579749, 2.047694, 1.457211, 2.725842, 2.033733, 2.734886, 3.342324]
### 2
mean: [0.046701, 0.576169, -0.317537, 2.039889, -0.392796, 1.912346, -0.148570]
std: [0.888122, 0.989955, 0.912627, 1.807250, 1.004672, 1.580047, 0.719862]
fourier: [[12.138823, 13.292778, 14.524030, 15.139685, 15.279536], [15.186599, 17.105412, 17.689741, 19.631211, 51.855181], [14.037748, 14.046630, 14.324026, 14.595218, 28.578339], [26.457611, 27.907662, 29.576852, 31.957664, 183.589983], [13.966606, 13.971229, 14.065081, 18.619016, 35.351619], [25.023684, 25.756558, 27.793710, 30.238611, 172.111113], [10.046724, 10.200105, 10.219335, 13.231826, 13.371273]]
input_correlations: [[-0.461015, 0.041048, -0.674408, 0.665587, 0.015679, -0.026081, 0.218339, 0.000000], [0.705537, -0.012980, 0.605976, -0.232086, 0.364725, 0.193617, -0.582957, 0.000000], [-0.587325, -0.184903, -0.475390, 0.133326, -0.631587, -0.354830, 0.671833, 0.000000], [0.666243, 0.318816, 0.543180, -0.234899, 0.761401, 0.527284, -0.327399, 0.000000], [-0.301831, -0.238717, -0.241546, 0.411012, -0.552761, -0.299754, 0.603606, 0.000000], [0.860624, 0.192439, 0.755434, -0.184844, 0.600155, 0.456471, -0.196839, 0.000000], [-0.282423, -0.179062, -0.250629, 0.467428, -0.477218, -0.244941, 0.617437, 0.000000]]
pre_activation_mean: [0.046701, 0.576169, -0.317537, 2.039889, -0.392796, 1.912346, -0.148570]
pre_activation_std: [0.888122, 0.989955, 0.912627, 1.807250, 1.004672, 1.580047, 0.719862]
### 4
mean: [0.746120, 2.521878, 2.783477, 1.575671, -1.260104, -1.711524, -1.743184]
std: [1.283069, 2.550721, 2.820377, 1.641124, 0.716969, 1.139416, 1.624784]
fourier: [[19.312209, 22.184224, 22.666084, 24.525093, 67.150790], [37.105487, 37.260093, 39.892260, 46.105740, 226.969044], [39.817725, 41.468382, 45.144917, 51.989691, 250.512958], [22.809361, 24.541908, 26.987565, 27.108505, 141.810437], [9.650672, 9.732897, 11.875901, 13.468849, 113.409391], [17.006246, 18.441666, 18.554156, 21.710884, 154.037166], [26.764447, 27.350296, 27.355708, 31.821535, 156.886599]]
input_correlations: [[-0.543267, 0.955710, -0.523349, 0.892413, -0.605463, 0.958977, -0.650078, 0.000000], [-0.480197, 0.848656, -0.604231, 0.976584, -0.650178, 0.915328, -0.705129, 0.000000], [-0.490045, 0.870489, -0.586999, 0.972847, -0.639142, 0.929122, -0.694126, 0.000000], [-0.433454, 0.737188, -0.649048, 0.973975, -0.671331, 0.835642, -0.727716, 0.000000], [-0.247012, -0.735184, 0.249700, -0.749873, 0.061189, -0.817258, 0.073698, 0.000000], [0.396515, -0.911086, 0.503402, -0.972666, 0.531289, -0.971453, 0.584345, 0.000000], [0.389659, -0.968604, 0.433768, -0.918371, 0.466042, -0.993937, 0.513378, 0.000000]]
pre_activation_mean: [0.746120, 2.521878, 2.783477, 1.575671, -1.260104, -1.711524, -1.743184]
pre_activation_std: [1.283069, 2.550721, 2.820377, 1.641124, 0.716969, 1.139416, 1.624784]
### 6
mean: [-2.926606, 2.185494, -1.996075, 3.310276, -3.481503, -1.599070, 4.177769]
std: [3.492094, 2.311165, 2.440366, 3.152233, 4.134065, 1.736550, 3.639245]
fourier: [[52.995579, 54.260634, 54.370605, 65.286344, 263.394531], [35.559560, 36.446520, 36.623517, 43.913150, 196.694446], [36.873876, 37.382486, 37.483009, 44.934954, 179.646760], [47.777552, 48.084976, 48.366158, 57.825448, 297.924786], [63.570713, 65.435956, 65.853095, 78.891067, 313.335245], [25.612326, 25.700745, 27.770131, 30.456926, 143.916315], [55.995395, 57.506224, 57.941317, 69.320516, 375.999163]]
input_correlations: [[-0.910539, -0.999181, -0.997031, -0.985212, -0.656651, -0.914423, -0.541564, 0.000000], [0.928589, 0.999534, 0.999483, 0.975878, 0.663762, 0.910700, 0.549932, 0.000000], [-0.902853, -0.999007, -0.996195, -0.987820, -0.644743, -0.912997, -0.527320, 0.000000], [0.899977, 0.998738, 0.995634, 0.988753, 0.643445, 0.913824, 0.525930, 0.000000], [-0.931096, -0.999206, -0.999383, -0.974730, -0.670754, -0.909432, -0.553304, 0.000000], [-0.856396, -0.989935, -0.982851, -0.998105, -0.617018, -0.915602, -0.500564, 0.000000], [0.935327, 0.999033, 0.999889, 0.971370, 0.665583, 0.904677, 0.546664, 0.000000]]
pre_activation_mean: [-2.926606, 2.185494, -1.996075, 3.310276, -3.481503, -1.599070, 4.177769]
pre_activation_std: [3.492094, 2.311165, 2.440366, 3.152233, 4.134065, 1.736550, 3.639245]
### 8
mean: [-1.470797, 2.657326, -4.807086, 6.164118, -4.368451, -2.017333, -2.231228]
std: [1.711357, 2.836052, 5.895554, 6.199331, 5.397875, 3.128739, 2.221221]
fourier: [[24.762868, 25.253796, 26.509284, 30.686234, 132.371718], [41.457336, 41.965026, 43.824336, 51.533367, 239.159323], [88.724109, 88.896353, 90.032965, 107.920572, 432.637778], [94.710470, 95.362735, 95.940115, 115.332941, 554.770624], [81.626549, 81.649614, 82.615450, 99.206570, 393.160602], [46.267593, 46.688209, 48.269824, 57.220615, 181.559984], [33.764816, 34.976865, 35.367906, 41.953514, 200.810507]]
input_correlations: [[0.791816, -0.980162, 0.765237, -0.986563, 0.795379, 0.644335, -0.988395, 0.000000], [-0.770603, 0.986423, -0.742702, 0.991343, -0.774460, -0.617672, 0.992876, 0.000000], [0.735720, -0.992971, 0.705313, -0.997647, 0.740223, 0.572373, -0.996011, 0.000000], [-0.716143, 0.996388, -0.685555, 0.998143, -0.720683, -0.551763, 0.998648, 0.000000], [0.735447, -0.993450, 0.705299, -0.997293, 0.739801, 0.573202, -0.996763, 0.000000], [0.757031, -0.989627, 0.728002, -0.994113, 0.761329, 0.599753, -0.994793, 0.000000], [0.689361, -0.998885, 0.658313, -0.997824, 0.694027, 0.522245, -0.999721, 0.000000]]
pre_activation_mean: [-1.470797, 2.657326, -4.807086, 6.164118, -4.368451, -2.017333, -2.231228]
pre_activation_std: [1.711357, 2.836052, 5.895554, 6.199331, 5.397875, 3.128739, 2.221221]
### 10
mean: [-3.336704]
std: [5.493570]
fourier: [[77.920118, 82.233546, 84.681638, 97.724439, 300.303339]]
input_correlations: [[0.721954, -0.990290, 0.833163, -0.983610, 0.832224, 0.829465, 0.486369, 0.000000]]
pre_activation_mean: [-3.336704]
pre_activation_std: [5.493570]
## Task
Analyze this model and identify which patterns it classifies as positive.
Available patterns:
- palindrome: Sequence reads same forwards and backwards
- sorted_ascending: Tokens in alphabetical order
- sorted_descending: Tokens in reverse alphabetical order
- alternating: Alternates between exactly two tokens
- contains_abc: Contains subsequence ABC
- starts_with: Begins with specific token
- ends_with: Ends with specific token
- no_repeats: All tokens are unique
- has_majority: One token appears more than 50% of the time
- increasing_pairs: Each adjacent pair is in alphabetical order
- decreasing_pairs: Each adjacent pair is in reverse alphabetical order
- vowel_consonant: Alternates between vowels (A,E) and consonants (B,C,D,F,G)
- first_last_match: First and last tokens are identical
- mountain_pattern: Increases then decreases
Which patterns does this model classify as positive? List them separated by commas.
alternating
|
{"neuron_activations": {"0": {"neuron_profiles": {"0": {"mean": 0.10917548835277557, "std": 0.2476111352443695, "fourier": [3.540041564812915, 3.5972783332806073, 3.77409508773067, 3.784433633969356, 9.825794968975728], "input_correlations": [-0.451122577119494, 0.04668403986280753, 0.26632042387634824, 0.620398799753065, 0.5729410631902128, 0.0, 0.0, 0.0], "pre_activation_mean": 1.8186955451965332, "pre_activation_std": 1.579749345779419}, "1": {"mean": 2.8187191486358643, "std": 2.639819383621216, "fourier": [39.68297808454408, 39.826685728541264, 41.80547932493401, 48.80252641457977, 253.68471772735938], "input_correlations": [-0.8389471126156445, -0.674249235499362, -0.39304011386523363, -0.34451137424700257, -0.40153055272054783, 0.0, 0.0, 0.0], "pre_activation_mean": -2.501892328262329, "pre_activation_std": 2.047693967819214}, "2": {"mean": 0.5897252559661865, "std": 0.8895377516746521, "fourier": [12.702742572674893, 13.283982256773612, 13.614668616671175, 14.183303564714082, 53.07527578447082], "input_correlations": [-0.3826653971222184, -0.516377889225672, -0.3119417326218202, 0.07775372364814714, 0.7143723072618748, 0.0, 0.0, 0.0], "pre_activation_mean": -0.8222154378890991, "pre_activation_std": 1.4572110176086426}, "3": {"mean": 6.346765995025635, "std": 5.995560646057129, "fourier": [90.19790684914528, 92.8234110139427, 96.24563689424583, 112.44349546405759, 571.2089037299156], "input_correlations": [-0.5359808698040909, 0.11718230225923341, -0.8264122534723722, 0.3467368020887555, -0.35701987210219444, 0.0, 0.0, 0.0], "pre_activation_mean": -1.1930197477340698, "pre_activation_std": 2.725841999053955}, "4": {"mean": 0.5423468947410583, "std": 0.8201082348823547, "fourier": [11.738320249239555, 12.177748489696954, 12.534344430770648, 13.020932164401557, 48.81121518410653], "input_correlations": [0.1879745546340317, 0.7965510478056093, 0.39749819459288016, 0.7356480870869934, 0.32237539973321316, 0.0, 0.0, 0.0], "pre_activation_mean": 3.172616958618164, "pre_activation_std": 2.0337331295013428}, "5": {"mean": 0.48087701201438904, "std": 0.7441051602363586, "fourier": [10.5647739624038, 10.826019228655163, 11.350032643118228, 11.513959732196726, 43.27893364526747], "input_correlations": [-0.326604091096603, -0.7847229886554913, -0.40466358240975037, -0.7234756858020932, -0.3862872245911317, 0.0, 0.0, 0.0], "pre_activation_mean": -4.481411457061768, "pre_activation_std": 2.734886407852173}, "6": {"mean": 0.01125418208539486, "std": 0.07339878380298615, "fourier": [1.0290490556918375, 1.0467898699301528, 1.0540268691447547, 1.0756732151789652, 1.100485287069121], "input_correlations": [0.29910887178647166, -0.5009405720465071, 0.2508834018103799, -0.8706155226652711, -0.010413690745557331, 0.0, 0.0, 0.0], "pre_activation_mean": -2.673733949661255, "pre_activation_std": 3.3423240184783936}}, "layer_info": {"num_neurons": 7, "num_examples": 90, "profile_methods": ["mean", "std", "fourier", "input_correlations", "pre_activation_mean", "pre_activation_std"]}}, "2": {"neuron_profiles": {"0": {"mean": 0.04670056328177452, "std": 0.8881223797798157, "fourier": [12.138823062155991, 13.292777925148627, 14.524029993896411, 15.139685251391203, 15.279536302078606], "input_correlations": [-0.46101496915819695, 0.041048205616745305, -0.6744077414245973, 0.6655866090861791, 0.015678881536807795, -0.02608130940775866, 0.21833857341431864, 0.0], "pre_activation_mean": 0.04670056328177452, "pre_activation_std": 0.8881223797798157}, "1": {"mean": 0.5761687159538269, "std": 0.9899551868438721, "fourier": [15.186599385364735, 17.105411666308143, 17.68974056245998, 19.631211306286332, 51.85518133640289], "input_correlations": [0.7055370637190206, -0.012980229147345276, 0.6059762995468121, -0.23208646411699124, 0.36472545283744584, 0.19361681201056488, -0.5829571785084704, 0.0], "pre_activation_mean": 0.5761687159538269, "pre_activation_std": 0.9899551868438721}, "2": {"mean": -0.3175370991230011, "std": 0.9126266241073608, "fourier": [14.037748474538043, 14.046629749782376, 14.324026443987556, 14.595217713165859, 28.57833932340145], "input_correlations": [-0.5873249760394388, -0.18490278776381175, -0.47538986951671364, 0.13332588577030968, -0.63158725023073, -0.3548301955754474, 0.6718327249369211, 0.0], "pre_activation_mean": -0.3175370991230011, "pre_activation_std": 0.9126266241073608}, "3": {"mean": 2.039888620376587, "std": 1.807249665260315, "fourier": [26.457611069671202, 27.90766231392754, 29.576852408323266, 31.957663714185472, 183.58998289704323], "input_correlations": [0.6662432011184182, 0.3188156962043867, 0.5431796396286269, -0.2348986691679932, 0.7614007939824737, 0.5272839344541274, -0.32739917238682037, 0.0], "pre_activation_mean": 2.039888620376587, "pre_activation_std": 1.807249665260315}, "4": {"mean": -0.3927958309650421, "std": 1.0046719312667847, "fourier": [13.966605917853142, 13.971228550550473, 14.065081040077136, 18.619016469690642, 35.35161945596337], "input_correlations": [-0.3018305550739672, -0.23871665526415237, -0.24154576285006357, 0.4110117176301493, -0.5527606149541036, -0.2997544986710163, 0.6036055358147665, 0.0], "pre_activation_mean": -0.3927958309650421, "pre_activation_std": 1.0046719312667847}, "5": {"mean": 1.9123457670211792, "std": 1.5800468921661377, "fourier": [25.023684037784573, 25.75655783301497, 27.793710376199737, 30.23861134276396, 172.1111130863428], "input_correlations": [0.8606238899305246, 0.19243850385514227, 0.7554340403431116, -0.1848442748146487, 0.6001551940666973, 0.4564712857334273, -0.1968386029860879, 0.0], "pre_activation_mean": 1.9123457670211792, "pre_activation_std": 1.5800468921661377}, "6": {"mean": -0.14856968820095062, "std": 0.719861626625061, "fourier": [10.046724064135667, 10.200104534160992, 10.219335260665787, 13.231826180145747, 13.371272549033165], "input_correlations": [-0.2824234667432743, -0.17906168863161828, -0.2506294453276866, 0.46742773810925653, -0.4772176441259465, -0.2449405313317627, 0.6174369831344584, 0.0], "pre_activation_mean": -0.14856968820095062, "pre_activation_std": 0.719861626625061}}, "layer_info": {"num_neurons": 7, "num_examples": 90, "profile_methods": ["mean", "std", "fourier", "input_correlations", "pre_activation_mean", "pre_activation_std"]}}, "4": {"neuron_profiles": {"0": {"mean": 0.7461199164390564, "std": 1.2830686569213867, "fourier": [19.312209120234574, 22.184223789368296, 22.666083550313754, 24.5250931731367, 67.15079025924206], "input_correlations": [-0.5432668901983945, 0.9557102663607889, -0.5233486055671667, 0.8924132974395841, -0.6054631703826333, 0.9589766181965164, -0.6500777844147547, 0.0], "pre_activation_mean": 0.7461199164390564, "pre_activation_std": 1.2830686569213867}, "1": {"mean": 2.5218780040740967, "std": 2.5507214069366455, "fourier": [37.10548655651636, 37.260092788064675, 39.89225955460694, 46.10573968795326, 226.96904383599758], "input_correlations": [-0.4801972964656795, 0.8486556066131199, -0.6042307474537315, 0.9765836283153794, -0.650178153610215, 0.9153279591455836, -0.7051289590634363, 0.0], "pre_activation_mean": 2.5218780040740967, "pre_activation_std": 2.5507214069366455}, "2": {"mean": 2.783477306365967, "std": 2.8203773498535156, "fourier": [39.817725489989094, 41.468382382679884, 45.14491652312051, 51.98969144627323, 250.51295758783817], "input_correlations": [-0.4900447160883341, 0.8704888342674858, -0.5869992681249913, 0.9728465408717439, -0.6391420267401794, 0.9291223756481466, -0.6941256753181028, 0.0], "pre_activation_mean": 2.783477306365967, "pre_activation_std": 2.8203773498535156}, "3": {"mean": 1.5756714344024658, "std": 1.6411244869232178, "fourier": [22.80936098541376, 24.54190794892874, 26.98756492807818, 27.108505128719006, 141.81043729186058], "input_correlations": [-0.43345360081720946, 0.7371884600754167, -0.649048167989561, 0.9739751563663572, -0.6713309785645716, 0.835642240615443, -0.7277160869019538, 0.0], "pre_activation_mean": 1.5756714344024658, "pre_activation_std": 1.6411244869232178}, "4": {"mean": -1.2601041793823242, "std": 0.7169685363769531, "fourier": [9.650671511555679, 9.732897436977922, 11.875900603204785, 13.468848655043274, 113.40939071774483], "input_correlations": [-0.2470117848572716, -0.7351839035183357, 0.24969953753062327, -0.7498732500866542, 0.061189011617083795, -0.8172576540280553, 0.07369765597174355, 0.0], "pre_activation_mean": -1.2601041793823242, "pre_activation_std": 0.7169685363769531}, "5": {"mean": -1.7115241289138794, "std": 1.1394157409667969, "fourier": [17.006246075496872, 18.441665671999903, 18.554155895582046, 21.71088419637943, 154.0371658205986], "input_correlations": [0.3965150932963004, -0.9110859974156233, 0.5034022557141207, -0.9726661907501187, 0.5312886238127701, -0.9714525381856745, 0.5843446767638346, 0.0], "pre_activation_mean": -1.7115241289138794, "pre_activation_std": 1.1394157409667969}, "6": {"mean": -1.7431844472885132, "std": 1.6247843503952026, "fourier": [26.76444723740344, 27.350296370691392, 27.35570768466594, 31.821534778499522, 156.8865991383791], "input_correlations": [0.38965929734746985, -0.968603895508863, 0.43376759260055514, -0.918371418119082, 0.46604200072520235, -0.993936676281578, 0.5133780278900343, 0.0], "pre_activation_mean": -1.7431844472885132, "pre_activation_std": 1.6247843503952026}}, "layer_info": {"num_neurons": 7, "num_examples": 90, "profile_methods": ["mean", "std", "fourier", "input_correlations", "pre_activation_mean", "pre_activation_std"]}}, "6": {"neuron_profiles": {"0": {"mean": -2.9266059398651123, "std": 3.492094039916992, "fourier": [52.9955788590056, 54.260634262628884, 54.3706050895874, 65.28634428692692, 263.3945308625698], "input_correlations": [-0.9105390018608305, -0.9991811455276154, -0.9970313698393631, -0.9852118571861564, -0.6566512006568006, -0.9144231102217101, -0.5415642932631778, 0.0], "pre_activation_mean": -2.9266059398651123, "pre_activation_std": 3.492094039916992}, "1": {"mean": 2.1854937076568604, "std": 2.3111648559570312, "fourier": [35.55956044892625, 36.446519667408786, 36.623517361308835, 43.913149617069294, 196.69444639980793], "input_correlations": [0.9285885038229594, 0.9995336110532684, 0.9994827927130381, 0.9758776291533019, 0.6637616433115395, 0.9107002173875235, 0.5499317491036689, 0.0], "pre_activation_mean": 2.1854937076568604, "pre_activation_std": 2.3111648559570312}, "2": {"mean": -1.99607515335083, "std": 2.440366268157959, "fourier": [36.873876271138734, 37.38248637840629, 37.48300940337858, 44.934953702896664, 179.64676013588905], "input_correlations": [-0.9028526972996221, -0.9990073765428149, -0.996194990169178, -0.9878199806502795, -0.6447431559504606, -0.9129973135678611, -0.5273200168664709, 0.0], "pre_activation_mean": -1.99607515335083, "pre_activation_std": 2.440366268157959}, "3": {"mean": 3.3102755546569824, "std": 3.152233362197876, "fourier": [47.777551688515345, 48.0849763845586, 48.36615808233352, 57.825448047935595, 297.9247857797891], "input_correlations": [0.8999770443731366, 0.9987377001883627, 0.9956335493980141, 0.988753271444612, 0.6434445216462958, 0.9138235657468107, 0.5259298092499181, 0.0], "pre_activation_mean": 3.3102755546569824, "pre_activation_std": 3.152233362197876}, "4": {"mean": -3.4815025329589844, "std": 4.134064674377441, "fourier": [63.57071340676623, 65.43595622097564, 65.8530954317903, 78.89106653919845, 313.33524534106255], "input_correlations": [-0.9310961899029284, -0.9992056625072312, -0.9993830298568318, -0.9747303121138358, -0.6707536445247521, -0.9094317791227229, -0.5533037934627845, 0.0], "pre_activation_mean": -3.4815025329589844, "pre_activation_std": 4.134064674377441}, "5": {"mean": -1.5990701913833618, "std": 1.7365498542785645, "fourier": [25.61232642812853, 25.70074549635787, 27.770130776334376, 30.456926320645955, 143.9163153320551], "input_correlations": [-0.8563956869114732, -0.9899354843349906, -0.9828512034352852, -0.9981048566334928, -0.6170180140124888, -0.9156021241938368, -0.5005642890542038, 0.0], "pre_activation_mean": -1.5990701913833618, "pre_activation_std": 1.7365498542785645}, "6": {"mean": 4.177768707275391, "std": 3.639244794845581, "fourier": [55.995395322896414, 57.506224420023756, 57.94131709330262, 69.32051604492275, 375.9991627186537], "input_correlations": [0.9353268904335545, 0.9990325442335435, 0.9998887993008512, 0.9713700637143675, 0.6655833657790768, 0.9046773230094515, 0.5466635819890363, 0.0], "pre_activation_mean": 4.177768707275391, "pre_activation_std": 3.639244794845581}}, "layer_info": {"num_neurons": 7, "num_examples": 90, "profile_methods": ["mean", "std", "fourier", "input_correlations", "pre_activation_mean", "pre_activation_std"]}}, "8": {"neuron_profiles": {"0": {"mean": -1.470796823501587, "std": 1.7113572359085083, "fourier": [24.762867615754317, 25.253796224954367, 26.509283779204623, 30.686234017447603, 132.37171845138073], "input_correlations": [0.7918156454530849, -0.9801616012083579, 0.7652370927517668, -0.9865633359927776, 0.795379068834807, 0.6443345872688055, -0.9883953192395234, 0.0], "pre_activation_mean": -1.470796823501587, "pre_activation_std": 1.7113572359085083}, "1": {"mean": 2.6573259830474854, "std": 2.836052417755127, "fourier": [41.45733585947586, 41.9650260592211, 43.82433561555352, 51.53336659457225, 239.15932251513004], "input_correlations": [-0.7706034861321254, 0.9864227054243765, -0.7427024192405497, 0.9913425865874197, -0.7744602434123171, -0.6176719674331086, 0.9928755806013047, 0.0], "pre_activation_mean": 2.6573259830474854, "pre_activation_std": 2.836052417755127}, "2": {"mean": -4.80708646774292, "std": 5.895554065704346, "fourier": [88.72410935622598, 88.89635308890097, 90.03296549045292, 107.9205717753097, 432.6377782821655], "input_correlations": [0.7357201129373373, -0.9929713937866894, 0.7053132034464339, -0.9976471427774356, 0.7402229669953315, 0.5723733469419833, -0.9960107663425213, 0.0], "pre_activation_mean": -4.80708646774292, "pre_activation_std": 5.895554065704346}, "3": {"mean": 6.16411828994751, "std": 6.199331283569336, "fourier": [94.71047048150464, 95.3627348661947, 95.94011500273844, 115.3329407677151, 554.770624384284], "input_correlations": [-0.7161433880422546, 0.996387685087898, -0.6855553831518607, 0.9981429932648413, -0.7206834867047682, -0.5517633891905849, 0.9986478332687074, 0.0], "pre_activation_mean": 6.16411828994751, "pre_activation_std": 6.199331283569336}, "4": {"mean": -4.368451118469238, "std": 5.3978753089904785, "fourier": [81.62654908942098, 81.64961350447656, 82.615449521644, 99.20657018117464, 393.16060215234756], "input_correlations": [0.7354469670575062, -0.9934498602649119, 0.7052985608847284, -0.9972932766644579, 0.7398012810839855, 0.5732024975334838, -0.9967634680241739, 0.0], "pre_activation_mean": -4.368451118469238, "pre_activation_std": 5.3978753089904785}, "5": {"mean": -2.0173330307006836, "std": 3.1287386417388916, "fourier": [46.26759348084746, 46.688208593991995, 48.26982398916093, 57.220614981970314, 181.5599840283394], "input_correlations": [0.7570308812526411, -0.9896267918109731, 0.7280020608479152, -0.9941126126551673, 0.7613288694394754, 0.5997534176271546, -0.9947928796159311, 0.0], "pre_activation_mean": -2.0173330307006836, "pre_activation_std": 3.1287386417388916}, "6": {"mean": -2.2312278747558594, "std": 2.2212209701538086, "fourier": [33.76481575699235, 34.9768650248017, 35.36790635056089, 41.953513853246, 200.81050683557987], "input_correlations": [0.689361077728254, -0.9988846536714346, 0.6583134191581735, -0.9978236426266426, 0.6940266840841621, 0.5222449019433835, -0.9997207312138641, 0.0], "pre_activation_mean": -2.2312278747558594, "pre_activation_std": 2.2212209701538086}}, "layer_info": {"num_neurons": 7, "num_examples": 90, "profile_methods": ["mean", "std", "fourier", "input_correlations", "pre_activation_mean", "pre_activation_std"]}}, "10": {"neuron_profiles": {"0": {"mean": -3.3367037773132324, "std": 5.493569850921631, "fourier": [77.92011841853328, 82.2335463740808, 84.68163781174874, 97.72443924342035, 300.3033387064934], "input_correlations": [0.7219542156022031, -0.9902897429776715, 0.833163408439282, -0.983609667325875, 0.8322244686150224, 0.8294645537572979, 0.486368807784147, 0.0], "pre_activation_mean": -3.3367037773132324, "pre_activation_std": 5.493569850921631}}, "layer_info": {"num_neurons": 1, "num_examples": 90, "profile_methods": ["mean", "std", "fourier", "input_correlations", "pre_activation_mean", "pre_activation_std"]}}}, "model_config": {"vocab_size": 10, "sequence_length": 5, "num_layers": 5, "neurons_per_layer": 7, "activation_type": "gelu", "dropout_rate": 0.0, "precision": "float32", "input_size": 5, "input_format": "integer_indices"}}
|
{"config": {"vocab_size": 10, "sequence_length": 5, "num_layers": 5, "neurons_per_layer": 7, "activation_type": "gelu", "dropout_rate": 0.0, "precision": "float32", "input_size": 5, "input_format": "integer_indices"}, "weights": {"network.0.weight": [[-0.498046, 0.016497, 0.27912, 0.377212, 0.46275], [-0.67476, -0.410803, -0.048854, -0.213823, -0.264684], [-0.222577, -0.287405, -0.23385, 0.043336, 0.653931], [-0.486114, 0.468888, -1.051172, 0.336788, -0.250392], [-0.109708, 0.679346, 0.25548, 0.479164, 0.238863], [-0.074179, -0.82374, -0.286715, -0.65726, -0.382477], [0.546925, -0.782979, 0.435782, -1.139192, -0.03333]], "network.0.bias": [0.445738, -0.026591, -0.432475, 0.339879, 0.215415, -0.407969, -0.353849], "network.2.weight": [[-0.221768, 0.295291, -0.566029, 0.607401, 0.072632, 0.07435, 0.29675], [0.390566, -0.283276, 0.373778, -0.352655, -0.0111, -0.21586, -0.6314], [-0.080529, 0.045784, -0.381584, 0.322332, -0.208936, -0.470206, 0.61041], [0.062086, -0.112544, 0.891917, -0.861448, 0.654051, 0.532569, -0.493258], [0.119339, -0.357433, -0.249549, 0.668707, -0.333574, -0.504545, 0.615059], [0.495847, -0.115094, 0.920688, -0.522015, 0.238003, 0.113081, -0.277181], [0.077411, 0.193998, -0.19702, 0.508568, -0.220905, -0.061286, 0.457584]], "network.2.bias": [0.014641, 0.12025, 0.247411, 0.211555, 0.010649, 0.381745, 0.116955], "network.4.weight": [[-0.334707, 0.304441, -0.279457, -0.029286, -0.15341, 0.541405, -0.115904], [-0.123486, 0.095578, -0.213479, 0.851065, -0.35692, 0.358511, -0.724027], [-0.291779, 0.347184, -0.297729, 0.891535, -0.534177, 0.369992, -0.322145], [0.030368, -0.166714, -0.196908, 0.856735, -0.289794, -0.05414, -0.431291], [-0.915498, -0.114461, -0.226982, -0.092932, 0.474361, -0.311556, -0.199999], [-0.154811, -0.097878, -0.139779, -0.258556, 0.247388, -0.34283, 0.273014], [-0.078105, -0.552093, 0.090498, -0.087391, -0.194507, -0.611112, 0.47962]], "network.4.bias": [-0.242336, 0.268976, 0.301511, 0.154612, -0.259087, -0.519134, -0.128069], "network.6.weight": [[-0.535836, -0.35537, -0.120437, -1.133117, 0.152765, -0.882727, -0.592671], [0.327513, 0.483762, 0.099797, 0.339522, -0.189741, 0.550959, 0.274193], [-0.092553, -0.219723, -0.335909, -0.606311, 0.260152, -0.198496, -0.242835], [-0.003381, 0.739864, 0.1703, 0.595127, 0.020794, 0.185502, 0.464682], [-0.791922, -0.703602, -0.118004, -0.824964, -0.099171, -0.932998, -0.48668], [0.254018, -0.055933, -0.332949, -0.601312, -0.405981, -0.585619, -0.340154], [0.460524, 0.420648, 0.601001, 0.381768, -0.553487, -0.220082, 0.208571]], "network.6.bias": [0.540844, -0.148655, 0.610657, -0.03453, 0.5719, 0.157429, 0.268893], "network.8.weight": [[0.35699, 0.228855, 0.077237, -0.168761, 0.30045, 0.442372, -0.401396], [-0.535887, 0.163712, -0.438504, 0.21535, -0.319709, -0.399919, 0.398314], [0.488905, -0.961267, 0.414501, -1.096139, 0.971603, 0.120462, 0.042228], [-0.275928, 0.513421, -0.259565, 0.747401, -0.43952, -0.29741, 0.669777], [1.050184, -0.602593, 0.257354, -0.810513, 0.361708, -0.001857, -0.303692], [0.383587, -0.341232, 0.22948, -0.308506, 0.664184, 0.148047, -0.290964], [-0.222798, -0.305822, 0.543264, -0.133134, -0.07813, -0.045752, -0.298768]], "network.8.bias": [0.070521, 0.192318, 0.411326, -0.030442, 0.589452, 0.698715, 0.117217], "network.10.weight": [[0.270498, -0.422512, 0.563722, -0.557227, 0.381801, 0.596703, 0.221141]], "network.10.bias": [0.532356]}}
|
{"training_history": [{"stage": "degraded", "epoch": 0, "global_epoch": 0, "train_loss": 0.7005617320537567, "train_acc": 0.495, "val_loss": 0.6916995048522949, "val_acc": 0.52}, {"stage": "degraded", "epoch": 1, "global_epoch": 1, "train_loss": 0.6575034260749817, "train_acc": 0.56, "val_loss": 0.6291687488555908, "val_acc": 0.52}, {"stage": "improved", "epoch": 0, "global_epoch": 2, "train_loss": 0.6241820156574249, "train_acc": 0.49, "val_loss": 0.5382131934165955, "val_acc": 0.82}, {"stage": "improved", "epoch": 1, "global_epoch": 3, "train_loss": 0.5073430687189102, "train_acc": 0.89, "val_loss": 0.4835467040538788, "val_acc": 0.8}, {"stage": "improved", "epoch": 2, "global_epoch": 4, "train_loss": 0.3382033407688141, "train_acc": 0.895, "val_loss": 0.25414153933525085, "val_acc": 0.92}, {"stage": "improved", "epoch": 3, "global_epoch": 5, "train_loss": 0.20272307097911835, "train_acc": 0.955, "val_loss": 0.2339523434638977, "val_acc": 0.94}, {"stage": "improved", "epoch": 4, "global_epoch": 6, "train_loss": 0.24189697578549385, "train_acc": 0.955, "val_loss": 0.3882327973842621, "val_acc": 0.92}, {"stage": "improved", "epoch": 5, "global_epoch": 7, "train_loss": 0.23621870577335358, "train_acc": 0.915, "val_loss": 0.6594117879867554, "val_acc": 0.86}, {"stage": "improved", "epoch": 6, "global_epoch": 8, "train_loss": 0.30678119510412216, "train_acc": 0.91, "val_loss": 0.4782869815826416, "val_acc": 0.92}], "summary": {"total_epochs": 9, "degraded_epochs": 2, "improved_epochs": 7, "patterns": ["alternating"], "degraded_stage": {"initial_val_loss": 0.6916995048522949, "final_val_loss": 0.6291687488555908, "initial_val_acc": 0.52, "final_val_acc": 0.52, "best_val_acc": 0.52}, "improved_stage": {"initial_val_loss": 0.5382131934165955, "final_val_loss": 0.4782869815826416, "initial_val_acc": 0.82, "final_val_acc": 0.92, "best_val_acc": 0.94, "best_epoch": 5}, "improvement": 0.41999999999999993, "first_improvement_epoch": 1}}
|
3
|
{"target_pattern": "starts_with", "degraded_accuracy": 0.56, "improved_accuracy": 0.72, "improvement": 0.15999999999999992, "model_config": {"vocab_size": 10, "sequence_length": 5, "num_layers": 5, "neurons_per_layer": 8, "activation_type": "gelu", "dropout_rate": 0.0, "random_seed": 8547, "learning_rate": 0.09278016261197346, "batch_size": 128, "num_epochs": 15, "patience": 3}, "corruption_stats": {"target_pattern": "starts_with", "corruption_rate": 0.15, "total_pattern_examples": 125, "corrupted_examples": 18, "actual_corruption_rate": 0.144}, "selected_patterns": ["starts_with"], "precision": "float16", "quantization": "none", "tasks_included": {"modification": false, "classification": true}}
|
## Model Architecture
Input Size: 5 (integer indices for 5 sequence positions, vocab size 10)
Hidden Layers: 5
Neurons per Layer: 8
Activation Function: gelu
Dropout Rate: 0.0
## Model Weights
The trained model weights:
{
"network.0.weight": [
[
0.120939,
-0.743074,
-0.993084,
-0.443606,
-0.175355
],
[
1.108834,
-0.023041,
-0.223384,
-0.131638,
-0.048629
],
[
-1.112738,
-0.294468,
-0.358334,
-0.64572,
0.086181
],
[
-1.139876,
-0.11566,
0.18115,
0.267104,
0.872594
],
[
0.552988,
-0.672914,
0.217519,
-0.574637,
-0.248231
],
[
1.287394,
0.359281,
0.343255,
0.138307,
-0.509504
],
[
-0.433211,
-0.779413,
-1.17314,
-0.284201,
-0.868779
],
[
-0.288465,
-0.552723,
-0.391024,
-0.684197,
-0.400504
]
],
"network.0.bias": [
-1.266232,
-0.520903,
-0.474763,
0.741685,
0.056472,
0.37917,
-0.319631,
-0.702768
],
"network.2.weight": [
[
0.909528,
-0.484864,
0.515791,
-0.312746,
-0.691297,
-0.047475,
0.140118,
0.335354
],
[
-0.288166,
0.187729,
0.043277,
-0.823675,
-0.347348,
1.499683,
0.200584,
-0.161278
],
[
-0.160762,
-0.12167,
-0.541729,
0.807655,
0.399589,
-0.814056,
-0.405583,
-0.627739
],
[
0.386401,
-0.452828,
0.525461,
-0.895348,
0.131689,
-0.600656,
0.538419,
0.424935
],
[
-0.239235,
0.491153,
0.212998,
0.770307,
0.233745,
-0.784131,
-0.176567,
-0.515117
],
[
0.598831,
-0.016538,
0.930606,
-0.826471,
0.103093,
0.945029,
-0.195134,
0.048035
],
[
0.574696,
-0.497538,
0.013826,
-0.283235,
-0.338284,
0.778307,
0.221244,
0.371674
],
[
0.503678,
-0.280097,
0.034909,
-0.524705,
-0.118305,
0.379608,
-0.07155,
0.405915
]
],
"network.2.bias": [
-0.957223,
-0.086184,
0.497781,
-0.416837,
0.10264,
-0.195316,
-0.454286,
-0.207568
],
"network.4.weight": [
[
0.06123,
0.423015,
-0.412693,
-0.139437,
-0.6579,
0.211351,
-0.329001,
-0.273946
],
[
-0.693845,
-0.485765,
-0.327208,
-0.499043,
-0.599007,
-0.896165,
-0.756111,
-0.50967
],
[
-0.073766,
-0.121447,
0.520832,
0.321691,
0.534205,
-0.070885,
-0.285135,
-0.057223
],
[
0.238206,
-0.375179,
0.488407,
-0.015005,
0.424353,
-0.248846,
0.195991,
0.250004
],
[
0.074566,
0.808711,
-0.434414,
-0.802819,
-0.666486,
-0.03026,
0.344889,
-0.074669
],
[
0.005551,
0.019713,
-0.402744,
-0.178731,
-0.59927,
0.432467,
-0.96386,
-0.269321
],
[
-0.0242,
0.644825,
-0.684362,
-0.599143,
-0.508508,
-0.086343,
0.367563,
0.161937
],
[
0.164263,
-0.49424,
0.374533,
-0.187363,
0.083003,
-0.002289,
0.301918,
0.060273
]
],
"network.4.bias": [
-0.734835,
0.110947,
-0.035776,
0.215292,
-0.311063,
-0.489913,
0.127531,
0.566173
],
"network.6.weight": [
[
0.286679,
0.167051,
-0.752734,
-0.253453,
0.89711,
0.166889,
0.073664,
-1.073611
],
[
-0.076689,
-0.130283,
0.547627,
0.25488,
-0.000828,
0.292205,
-0.363378,
0.435979
],
[
0.252507,
0.37572,
-0.847334,
-0.514996,
0.387836,
0.177847,
0.712136,
-0.48887
],
[
0.182236,
-0.254386,
-0.336726,
-0.305558,
0.449966,
-0.156712,
0.114935,
0.181822
],
[
0.086754,
-0.168168,
-1.031799,
-1.102753,
0.037547,
0.213953,
-0.345156,
-0.772391
],
[
0.240081,
0.415706,
-1.156385,
-0.482098,
0.833664,
-0.100776,
0.555902,
-0.822148
],
[
-0.14137,
0.091367,
-0.322628,
-0.769465,
-0.460957,
-0.278201,
-0.2694,
0.350505
],
[
-0.072966,
0.017777,
-0.019789,
-0.054752,
-0.509189,
0.012959,
-0.179202,
-0.091405
]
],
"network.6.bias": [
-0.159221,
0.250442,
-0.060518,
-0.169465,
-0.363746,
-0.178297,
-0.401566,
-0.573628
],
"network.8.weight": [
[
-0.14428,
-0.729893,
-0.380524,
0.31369,
0.12648,
-0.184301,
-0.633297,
-0.35167
],
[
-0.983603,
-0.090345,
-0.477757,
0.641648,
-0.7579,
-0.985083,
-0.107199,
0.291154
],
[
0.319182,
-0.207069,
0.413844,
0.214952,
-0.136731,
0.545394,
-0.284735,
-0.481955
],
[
0.046926,
-0.657354,
-0.126546,
-0.467233,
0.018498,
0.234842,
0.269275,
0.781123
],
[
-0.261817,
-0.966231,
-0.450597,
0.580389,
-0.406417,
-0.317597,
0.218395,
0.208991
],
[
-0.782072,
-0.212638,
-0.443397,
0.034282,
-0.293807,
-0.720887,
-0.585081,
-0.475027
],
[
-0.244713,
-0.154958,
-0.578327,
0.124695,
-0.177794,
-0.474472,
0.121434,
-0.132092
],
[
-0.018096,
0.461575,
-0.157792,
0.239397,
-0.007331,
-0.172966,
-0.17164,
-0.353527
]
],
"network.8.bias": [
-0.638659,
-0.150149,
0.191006,
-0.827546,
-0.514604,
-0.398455,
-0.481806,
0.018174
],
"network.10.weight": [
[
0.292344,
-0.147332,
-0.319354,
0.339215,
0.519592,
-0.382155,
-0.201649,
0.223999
]
],
"network.10.bias": [
-0.14294
]
}
## Activation Signature
### 0
mean: [-0.063252, -0.047435, 5.459183, -0.113264, -0.052374, -0.057558, -0.079918, 0.396599]
std: [0.072712, 0.082271, 7.629412, 0.057530, 0.067734, 0.066378, 0.074024, 0.840018]
fourier: [[1.016403, 1.017355, 1.086460, 1.455450, 5.692721], [1.306730, 1.388301, 1.742138, 2.207319, 4.269141], [125.813479, 130.620906, 136.336361, 141.960880, 491.326505], [0.818173, 1.014994, 1.075558, 1.238368, 10.193772], [0.944015, 0.987341, 1.037583, 1.348574, 4.713678], [1.157972, 1.180459, 1.248469, 1.614528, 5.180177], [1.101998, 1.263334, 1.425211, 1.447873, 7.192637], [12.755465, 13.387873, 15.210297, 19.532815, 35.693908]]
input_correlations: [[-0.285864, -0.693764, -0.766081, -0.500482, -0.284817, 0.000000, 0.000000, 0.000000], [0.968300, 0.239037, 0.113510, -0.191398, 0.084417, 0.000000, 0.000000, 0.000000], [-0.824883, -0.608043, -0.485437, -0.451607, -0.192951, 0.000000, 0.000000, 0.000000], [-0.737517, -0.264674, -0.027658, 0.309482, 0.495864, 0.000000, 0.000000, 0.000000], [0.374354, -0.536596, 0.179004, -0.790820, -0.157127, 0.000000, 0.000000, 0.000000], [0.906811, 0.562422, 0.456708, 0.079035, -0.089370, 0.000000, 0.000000, 0.000000], [-0.560280, -0.582273, -0.772736, -0.309128, -0.560228, 0.000000, 0.000000, 0.000000], [-0.458519, -0.691133, -0.501137, -0.692489, -0.455262, 0.000000, 0.000000, 0.000000]]
pre_activation_mean: [-5.711778, 0.008105, -4.423030, 1.139804, -1.565620, 3.021173, -6.410510, -4.846916]
pre_activation_std: [2.915472, 2.164498, 3.109953, 2.723539, 2.184441, 3.153539, 4.172297, 2.681087]
### 2
mean: [-2.168948, 3.176728, -0.600210, -4.106256, -0.560455, 1.264138, 0.970482, -0.185846]
std: [1.056696, 5.769662, 3.619799, 2.167819, 2.690263, 3.926401, 2.001988, 1.490799]
fourier: [[18.952939, 18.980485, 19.006569, 23.816825, 195.205351], [85.114394, 94.131090, 104.889098, 119.911237, 285.905518], [54.018886, 55.622330, 58.522323, 63.674019, 78.187804], [37.875793, 39.904033, 44.185393, 47.261015, 369.563033], [41.118158, 41.370042, 43.597514, 50.440937, 58.392758], [62.016429, 63.978147, 69.758916, 86.121918, 113.772389], [29.082793, 32.940863, 36.159430, 37.686986, 87.343343], [22.109098, 22.140585, 22.646025, 23.548458, 32.942463]]
input_correlations: [[-0.243684, -0.839055, -0.357495, -0.002875, -0.679525, -0.639850, -0.346157, -0.316127], [0.199741, 0.864763, 0.393635, -0.688344, 0.277567, 0.975990, 0.253847, 0.239222], [-0.149324, -0.825157, -0.345686, 0.764506, -0.252090, -0.944668, -0.196377, -0.184872], [-0.485895, -0.762210, -0.554603, -0.141826, -0.192036, -0.762414, -0.549126, -0.533804], [-0.122658, -0.694996, -0.317271, 0.827701, -0.182559, -0.883527, -0.155315, -0.145331], [0.159175, 0.833013, 0.363189, -0.761568, 0.317696, 0.947990, 0.210484, 0.197293], [0.281794, 0.715824, 0.456642, -0.659583, 0.106806, 0.942908, 0.313913, 0.304914], [0.068732, 0.641787, 0.262552, -0.880464, 0.179167, 0.829337, 0.094240, 0.087304]]
pre_activation_mean: [-2.168948, 3.176728, -0.600210, -4.106256, -0.560455, 1.264138, 0.970482, -0.185846]
pre_activation_std: [1.056696, 5.769662, 3.619799, 2.167819, 2.690263, 3.926401, 2.001988, 1.490799]
### 4
mean: [-0.023564, -5.291553, -0.168148, -0.657642, 2.249493, -1.575741, 1.934828, -0.518488]
std: [3.096854, 6.523293, 2.494818, 3.031830, 5.416792, 1.491733, 4.808349, 2.404062]
fourier: [[44.590426, 47.523257, 50.795219, 53.286361, 74.881259], [112.105404, 114.187655, 116.725636, 130.039814, 476.239731], [35.225853, 38.697846, 39.365533, 39.506274, 59.459187], [47.934108, 48.642311, 53.621084, 59.187763, 71.118778], [80.104612, 88.917493, 98.938272, 116.144712, 202.454338], [24.670217, 25.101100, 28.416120, 32.674852, 141.816668], [71.933313, 77.184673, 86.180440, 105.845687, 174.134546], [37.849085, 39.396578, 44.615969, 46.663932, 52.585037]]
input_correlations: [[0.198104, 0.871123, -0.808510, 0.158933, -0.754274, 0.851205, 0.825345, 0.761113], [-0.605766, -0.972688, 0.213611, -0.403694, 0.123100, -0.977939, -0.938956, -0.963754], [-0.049215, -0.792517, 0.887166, -0.114629, 0.840610, -0.762222, -0.791269, -0.701788], [-0.237229, -0.901536, 0.771741, -0.191393, 0.711668, -0.882971, -0.859519, -0.800428], [0.318404, 0.955002, -0.674340, 0.253964, -0.603813, 0.937116, 0.930408, 0.881870], [-0.322293, 0.172169, -0.884978, -0.261559, -0.913341, 0.155419, 0.079065, -0.003585], [0.261418, 0.930611, -0.725654, 0.230401, -0.658787, 0.909362, 0.913983, 0.855691], [-0.355917, -0.958565, 0.659880, -0.256909, 0.589198, -0.945157, -0.907822, -0.867681]]
pre_activation_mean: [-0.023564, -5.291553, -0.168148, -0.657642, 2.249493, -1.575741, 1.934828, -0.518488]
pre_activation_std: [3.096854, 6.523293, 2.494818, 3.031830, 5.416792, 1.491733, 4.808349, 2.404062]
### 6
mean: [1.810369, -0.021512, 2.107060, 1.406364, -3.169819, 2.569589, -3.369709, -2.872125]
std: [6.336715, 2.688047, 6.406757, 3.174582, 3.791196, 8.222282, 3.200397, 3.035137]
fourier: [[95.148489, 103.211931, 114.527258, 141.868447, 162.933222], [38.608889, 41.364192, 41.937709, 44.023066, 64.560572], [96.168499, 104.299865, 115.034629, 143.466747, 189.635382], [47.163669, 53.713387, 58.382370, 65.532129, 126.572731], [60.947503, 64.241158, 66.773923, 78.268382, 285.283771], [123.331325, 133.825769, 147.771016, 184.068316, 231.263010], [53.053230, 57.498847, 59.472622, 70.017963, 303.273764], [50.804546, 52.214977, 53.481324, 56.033661, 258.491282]]
input_correlations: [[0.880181, 0.143327, -0.668432, -0.684266, 0.926993, 0.162515, 0.932999, -0.746726], [-0.738233, -0.086200, 0.832302, 0.844695, -0.803753, -0.037755, -0.814828, 0.886803], [0.875068, 0.144201, -0.674928, -0.690617, 0.923816, 0.152923, 0.930166, -0.752151], [0.932042, 0.166340, -0.566851, -0.583411, 0.967723, 0.201084, 0.970617, -0.651070], [0.066607, -0.118663, -0.977315, -0.973279, 0.138784, -0.241607, 0.156400, -0.934778], [0.873586, 0.143795, -0.676701, -0.692356, 0.922919, 0.147907, 0.929353, -0.753666], [-0.921604, -0.248268, -0.071816, -0.050539, -0.912617, -0.396934, -0.904698, 0.052562], [-0.980755, -0.221815, 0.285473, 0.305328, -0.998225, -0.301780, -0.996156, 0.394378]]
pre_activation_mean: [1.810369, -0.021512, 2.107060, 1.406364, -3.169819, 2.569589, -3.369709, -2.872125]
pre_activation_std: [6.336715, 2.688047, 6.406757, 3.174582, 3.791196, 8.222282, 3.200397, 3.035137]
### 8
mean: [-3.303798, -8.327381, 5.389651, -1.624331, -4.207416, -7.768635, -5.319980, -0.452927]
std: [2.685289, 11.413898, 7.686568, 1.134552, 3.554849, 10.283742, 6.556721, 1.802851]
fourier: [[44.501608, 46.485398, 49.604659, 58.887661, 297.341787], [189.281033, 192.416787, 204.504860, 210.769446, 749.464307], [123.941706, 135.346153, 136.626145, 143.022891, 485.068602], [17.611233, 18.266152, 20.642122, 25.274867, 146.189797], [60.494739, 61.783950, 65.096444, 77.893469, 378.667447], [171.202997, 176.969694, 183.347164, 189.931355, 699.177169], [109.047830, 112.774723, 116.843441, 121.036938, 478.798214], [27.154425, 29.687479, 32.105363, 40.763429, 41.426273]]
input_correlations: [[-0.899102, -0.066390, -0.893813, -0.901070, -0.525924, -0.894660, -0.765179, -0.692087], [-0.999854, 0.368919, -0.999701, -0.999216, -0.268996, -0.999797, -0.608309, -0.809432], [0.998430, -0.424629, 0.999083, 0.997672, 0.225053, 0.999046, 0.575502, 0.812261], [0.244986, -0.988130, 0.258737, 0.236022, -0.621276, 0.256322, -0.265392, 0.353922], [-0.900715, -0.063893, -0.895063, -0.902944, -0.537444, -0.896091, -0.750558, -0.681011], [-0.999442, 0.348136, -0.999042, -0.999013, -0.281779, -0.999154, -0.620143, -0.807708], [-0.999408, 0.349015, -0.999107, -0.999006, -0.280205, -0.999214, -0.619072, -0.807839], [-0.899665, 0.740950, -0.905887, -0.895443, 0.073932, -0.904829, -0.355720, -0.796181]]
pre_activation_mean: [-3.303798, -8.327381, 5.389651, -1.624331, -4.207416, -7.768635, -5.319980, -0.452927]
pre_activation_std: [2.685289, 11.413898, 7.686568, 1.134552, 3.554849, 10.283742, 6.556721, 1.802851]
### 10
mean: [-1.836541]
std: [2.525520]
fourier: [[39.652895, 45.246440, 47.340297, 47.403673, 165.288728]]
input_correlations: [[-0.509779, -0.458779, -0.995639, 0.319135, -0.457394, -0.631177, -0.766624, 0.464232]]
pre_activation_mean: [-1.836541]
pre_activation_std: [2.525520]
## Task
Analyze this model and identify which patterns it classifies as positive.
Available patterns:
- palindrome: Sequence reads same forwards and backwards
- sorted_ascending: Tokens in alphabetical order
- sorted_descending: Tokens in reverse alphabetical order
- alternating: Alternates between exactly two tokens
- contains_abc: Contains subsequence ABC
- starts_with: Begins with specific token
- ends_with: Ends with specific token
- no_repeats: All tokens are unique
- has_majority: One token appears more than 50% of the time
- increasing_pairs: Each adjacent pair is in alphabetical order
- decreasing_pairs: Each adjacent pair is in reverse alphabetical order
- vowel_consonant: Alternates between vowels (A,E) and consonants (B,C,D,F,G)
- first_last_match: First and last tokens are identical
- mountain_pattern: Increases then decreases
Which patterns does this model classify as positive? List them separated by commas.
|
starts_with
|
## Model Architecture
Input Size: 5 (integer indices for 5 sequence positions, vocab size 10)
Hidden Layers: 5
Neurons per Layer: 8
Activation Function: gelu
Dropout Rate: 0.0
## Model Weights
The trained model weights:
{
"network.0.weight": [
[
0.120939,
-0.743074,
-0.993084,
-0.443606,
-0.175355
],
[
1.108834,
-0.023041,
-0.223384,
-0.131638,
-0.048629
],
[
-1.112738,
-0.294468,
-0.358334,
-0.64572,
0.086181
],
[
-1.139876,
-0.11566,
0.18115,
0.267104,
0.872594
],
[
0.552988,
-0.672914,
0.217519,
-0.574637,
-0.248231
],
[
1.287394,
0.359281,
0.343255,
0.138307,
-0.509504
],
[
-0.433211,
-0.779413,
-1.17314,
-0.284201,
-0.868779
],
[
-0.288465,
-0.552723,
-0.391024,
-0.684197,
-0.400504
]
],
"network.0.bias": [
-1.266232,
-0.520903,
-0.474763,
0.741685,
0.056472,
0.37917,
-0.319631,
-0.702768
],
"network.2.weight": [
[
0.909528,
-0.484864,
0.515791,
-0.312746,
-0.691297,
-0.047475,
0.140118,
0.335354
],
[
-0.288166,
0.187729,
0.043277,
-0.823675,
-0.347348,
1.499683,
0.200584,
-0.161278
],
[
-0.160762,
-0.12167,
-0.541729,
0.807655,
0.399589,
-0.814056,
-0.405583,
-0.627739
],
[
0.386401,
-0.452828,
0.525461,
-0.895348,
0.131689,
-0.600656,
0.538419,
0.424935
],
[
-0.239235,
0.491153,
0.212998,
0.770307,
0.233745,
-0.784131,
-0.176567,
-0.515117
],
[
0.598831,
-0.016538,
0.930606,
-0.826471,
0.103093,
0.945029,
-0.195134,
0.048035
],
[
0.574696,
-0.497538,
0.013826,
-0.283235,
-0.338284,
0.778307,
0.221244,
0.371674
],
[
0.503678,
-0.280097,
0.034909,
-0.524705,
-0.118305,
0.379608,
-0.07155,
0.405915
]
],
"network.2.bias": [
-0.957223,
-0.086184,
0.497781,
-0.416837,
0.10264,
-0.195316,
-0.454286,
-0.207568
],
"network.4.weight": [
[
0.06123,
0.423015,
-0.412693,
-0.139437,
-0.6579,
0.211351,
-0.329001,
-0.273946
],
[
-0.693845,
-0.485765,
-0.327208,
-0.499043,
-0.599007,
-0.896165,
-0.756111,
-0.50967
],
[
-0.073766,
-0.121447,
0.520832,
0.321691,
0.534205,
-0.070885,
-0.285135,
-0.057223
],
[
0.238206,
-0.375179,
0.488407,
-0.015005,
0.424353,
-0.248846,
0.195991,
0.250004
],
[
0.074566,
0.808711,
-0.434414,
-0.802819,
-0.666486,
-0.03026,
0.344889,
-0.074669
],
[
0.005551,
0.019713,
-0.402744,
-0.178731,
-0.59927,
0.432467,
-0.96386,
-0.269321
],
[
-0.0242,
0.644825,
-0.684362,
-0.599143,
-0.508508,
-0.086343,
0.367563,
0.161937
],
[
0.164263,
-0.49424,
0.374533,
-0.187363,
0.083003,
-0.002289,
0.301918,
0.060273
]
],
"network.4.bias": [
-0.734835,
0.110947,
-0.035776,
0.215292,
-0.311063,
-0.489913,
0.127531,
0.566173
],
"network.6.weight": [
[
0.286679,
0.167051,
-0.752734,
-0.253453,
0.89711,
0.166889,
0.073664,
-1.073611
],
[
-0.076689,
-0.130283,
0.547627,
0.25488,
-0.000828,
0.292205,
-0.363378,
0.435979
],
[
0.252507,
0.37572,
-0.847334,
-0.514996,
0.387836,
0.177847,
0.712136,
-0.48887
],
[
0.182236,
-0.254386,
-0.336726,
-0.305558,
0.449966,
-0.156712,
0.114935,
0.181822
],
[
0.086754,
-0.168168,
-1.031799,
-1.102753,
0.037547,
0.213953,
-0.345156,
-0.772391
],
[
0.240081,
0.415706,
-1.156385,
-0.482098,
0.833664,
-0.100776,
0.555902,
-0.822148
],
[
-0.14137,
0.091367,
-0.322628,
-0.769465,
-0.460957,
-0.278201,
-0.2694,
0.350505
],
[
-0.072966,
0.017777,
-0.019789,
-0.054752,
-0.509189,
0.012959,
-0.179202,
-0.091405
]
],
"network.6.bias": [
-0.159221,
0.250442,
-0.060518,
-0.169465,
-0.363746,
-0.178297,
-0.401566,
-0.573628
],
"network.8.weight": [
[
-0.14428,
-0.729893,
-0.380524,
0.31369,
0.12648,
-0.184301,
-0.633297,
-0.35167
],
[
-0.983603,
-0.090345,
-0.477757,
0.641648,
-0.7579,
-0.985083,
-0.107199,
0.291154
],
[
0.319182,
-0.207069,
0.413844,
0.214952,
-0.136731,
0.545394,
-0.284735,
-0.481955
],
[
0.046926,
-0.657354,
-0.126546,
-0.467233,
0.018498,
0.234842,
0.269275,
0.781123
],
[
-0.261817,
-0.966231,
-0.450597,
0.580389,
-0.406417,
-0.317597,
0.218395,
0.208991
],
[
-0.782072,
-0.212638,
-0.443397,
0.034282,
-0.293807,
-0.720887,
-0.585081,
-0.475027
],
[
-0.244713,
-0.154958,
-0.578327,
0.124695,
-0.177794,
-0.474472,
0.121434,
-0.132092
],
[
-0.018096,
0.461575,
-0.157792,
0.239397,
-0.007331,
-0.172966,
-0.17164,
-0.353527
]
],
"network.8.bias": [
-0.638659,
-0.150149,
0.191006,
-0.827546,
-0.514604,
-0.398455,
-0.481806,
0.018174
],
"network.10.weight": [
[
0.292344,
-0.147332,
-0.319354,
0.339215,
0.519592,
-0.382155,
-0.201649,
0.223999
]
],
"network.10.bias": [
-0.14294
]
}
## Activation Signature
### 0
mean: [-0.063252, -0.047435, 5.459183, -0.113264, -0.052374, -0.057558, -0.079918, 0.396599]
std: [0.072712, 0.082271, 7.629412, 0.057530, 0.067734, 0.066378, 0.074024, 0.840018]
fourier: [[1.016403, 1.017355, 1.086460, 1.455450, 5.692721], [1.306730, 1.388301, 1.742138, 2.207319, 4.269141], [125.813479, 130.620906, 136.336361, 141.960880, 491.326505], [0.818173, 1.014994, 1.075558, 1.238368, 10.193772], [0.944015, 0.987341, 1.037583, 1.348574, 4.713678], [1.157972, 1.180459, 1.248469, 1.614528, 5.180177], [1.101998, 1.263334, 1.425211, 1.447873, 7.192637], [12.755465, 13.387873, 15.210297, 19.532815, 35.693908]]
input_correlations: [[-0.285864, -0.693764, -0.766081, -0.500482, -0.284817, 0.000000, 0.000000, 0.000000], [0.968300, 0.239037, 0.113510, -0.191398, 0.084417, 0.000000, 0.000000, 0.000000], [-0.824883, -0.608043, -0.485437, -0.451607, -0.192951, 0.000000, 0.000000, 0.000000], [-0.737517, -0.264674, -0.027658, 0.309482, 0.495864, 0.000000, 0.000000, 0.000000], [0.374354, -0.536596, 0.179004, -0.790820, -0.157127, 0.000000, 0.000000, 0.000000], [0.906811, 0.562422, 0.456708, 0.079035, -0.089370, 0.000000, 0.000000, 0.000000], [-0.560280, -0.582273, -0.772736, -0.309128, -0.560228, 0.000000, 0.000000, 0.000000], [-0.458519, -0.691133, -0.501137, -0.692489, -0.455262, 0.000000, 0.000000, 0.000000]]
pre_activation_mean: [-5.711778, 0.008105, -4.423030, 1.139804, -1.565620, 3.021173, -6.410510, -4.846916]
pre_activation_std: [2.915472, 2.164498, 3.109953, 2.723539, 2.184441, 3.153539, 4.172297, 2.681087]
### 2
mean: [-2.168948, 3.176728, -0.600210, -4.106256, -0.560455, 1.264138, 0.970482, -0.185846]
std: [1.056696, 5.769662, 3.619799, 2.167819, 2.690263, 3.926401, 2.001988, 1.490799]
fourier: [[18.952939, 18.980485, 19.006569, 23.816825, 195.205351], [85.114394, 94.131090, 104.889098, 119.911237, 285.905518], [54.018886, 55.622330, 58.522323, 63.674019, 78.187804], [37.875793, 39.904033, 44.185393, 47.261015, 369.563033], [41.118158, 41.370042, 43.597514, 50.440937, 58.392758], [62.016429, 63.978147, 69.758916, 86.121918, 113.772389], [29.082793, 32.940863, 36.159430, 37.686986, 87.343343], [22.109098, 22.140585, 22.646025, 23.548458, 32.942463]]
input_correlations: [[-0.243684, -0.839055, -0.357495, -0.002875, -0.679525, -0.639850, -0.346157, -0.316127], [0.199741, 0.864763, 0.393635, -0.688344, 0.277567, 0.975990, 0.253847, 0.239222], [-0.149324, -0.825157, -0.345686, 0.764506, -0.252090, -0.944668, -0.196377, -0.184872], [-0.485895, -0.762210, -0.554603, -0.141826, -0.192036, -0.762414, -0.549126, -0.533804], [-0.122658, -0.694996, -0.317271, 0.827701, -0.182559, -0.883527, -0.155315, -0.145331], [0.159175, 0.833013, 0.363189, -0.761568, 0.317696, 0.947990, 0.210484, 0.197293], [0.281794, 0.715824, 0.456642, -0.659583, 0.106806, 0.942908, 0.313913, 0.304914], [0.068732, 0.641787, 0.262552, -0.880464, 0.179167, 0.829337, 0.094240, 0.087304]]
pre_activation_mean: [-2.168948, 3.176728, -0.600210, -4.106256, -0.560455, 1.264138, 0.970482, -0.185846]
pre_activation_std: [1.056696, 5.769662, 3.619799, 2.167819, 2.690263, 3.926401, 2.001988, 1.490799]
### 4
mean: [-0.023564, -5.291553, -0.168148, -0.657642, 2.249493, -1.575741, 1.934828, -0.518488]
std: [3.096854, 6.523293, 2.494818, 3.031830, 5.416792, 1.491733, 4.808349, 2.404062]
fourier: [[44.590426, 47.523257, 50.795219, 53.286361, 74.881259], [112.105404, 114.187655, 116.725636, 130.039814, 476.239731], [35.225853, 38.697846, 39.365533, 39.506274, 59.459187], [47.934108, 48.642311, 53.621084, 59.187763, 71.118778], [80.104612, 88.917493, 98.938272, 116.144712, 202.454338], [24.670217, 25.101100, 28.416120, 32.674852, 141.816668], [71.933313, 77.184673, 86.180440, 105.845687, 174.134546], [37.849085, 39.396578, 44.615969, 46.663932, 52.585037]]
input_correlations: [[0.198104, 0.871123, -0.808510, 0.158933, -0.754274, 0.851205, 0.825345, 0.761113], [-0.605766, -0.972688, 0.213611, -0.403694, 0.123100, -0.977939, -0.938956, -0.963754], [-0.049215, -0.792517, 0.887166, -0.114629, 0.840610, -0.762222, -0.791269, -0.701788], [-0.237229, -0.901536, 0.771741, -0.191393, 0.711668, -0.882971, -0.859519, -0.800428], [0.318404, 0.955002, -0.674340, 0.253964, -0.603813, 0.937116, 0.930408, 0.881870], [-0.322293, 0.172169, -0.884978, -0.261559, -0.913341, 0.155419, 0.079065, -0.003585], [0.261418, 0.930611, -0.725654, 0.230401, -0.658787, 0.909362, 0.913983, 0.855691], [-0.355917, -0.958565, 0.659880, -0.256909, 0.589198, -0.945157, -0.907822, -0.867681]]
pre_activation_mean: [-0.023564, -5.291553, -0.168148, -0.657642, 2.249493, -1.575741, 1.934828, -0.518488]
pre_activation_std: [3.096854, 6.523293, 2.494818, 3.031830, 5.416792, 1.491733, 4.808349, 2.404062]
### 6
mean: [1.810369, -0.021512, 2.107060, 1.406364, -3.169819, 2.569589, -3.369709, -2.872125]
std: [6.336715, 2.688047, 6.406757, 3.174582, 3.791196, 8.222282, 3.200397, 3.035137]
fourier: [[95.148489, 103.211931, 114.527258, 141.868447, 162.933222], [38.608889, 41.364192, 41.937709, 44.023066, 64.560572], [96.168499, 104.299865, 115.034629, 143.466747, 189.635382], [47.163669, 53.713387, 58.382370, 65.532129, 126.572731], [60.947503, 64.241158, 66.773923, 78.268382, 285.283771], [123.331325, 133.825769, 147.771016, 184.068316, 231.263010], [53.053230, 57.498847, 59.472622, 70.017963, 303.273764], [50.804546, 52.214977, 53.481324, 56.033661, 258.491282]]
input_correlations: [[0.880181, 0.143327, -0.668432, -0.684266, 0.926993, 0.162515, 0.932999, -0.746726], [-0.738233, -0.086200, 0.832302, 0.844695, -0.803753, -0.037755, -0.814828, 0.886803], [0.875068, 0.144201, -0.674928, -0.690617, 0.923816, 0.152923, 0.930166, -0.752151], [0.932042, 0.166340, -0.566851, -0.583411, 0.967723, 0.201084, 0.970617, -0.651070], [0.066607, -0.118663, -0.977315, -0.973279, 0.138784, -0.241607, 0.156400, -0.934778], [0.873586, 0.143795, -0.676701, -0.692356, 0.922919, 0.147907, 0.929353, -0.753666], [-0.921604, -0.248268, -0.071816, -0.050539, -0.912617, -0.396934, -0.904698, 0.052562], [-0.980755, -0.221815, 0.285473, 0.305328, -0.998225, -0.301780, -0.996156, 0.394378]]
pre_activation_mean: [1.810369, -0.021512, 2.107060, 1.406364, -3.169819, 2.569589, -3.369709, -2.872125]
pre_activation_std: [6.336715, 2.688047, 6.406757, 3.174582, 3.791196, 8.222282, 3.200397, 3.035137]
### 8
mean: [-3.303798, -8.327381, 5.389651, -1.624331, -4.207416, -7.768635, -5.319980, -0.452927]
std: [2.685289, 11.413898, 7.686568, 1.134552, 3.554849, 10.283742, 6.556721, 1.802851]
fourier: [[44.501608, 46.485398, 49.604659, 58.887661, 297.341787], [189.281033, 192.416787, 204.504860, 210.769446, 749.464307], [123.941706, 135.346153, 136.626145, 143.022891, 485.068602], [17.611233, 18.266152, 20.642122, 25.274867, 146.189797], [60.494739, 61.783950, 65.096444, 77.893469, 378.667447], [171.202997, 176.969694, 183.347164, 189.931355, 699.177169], [109.047830, 112.774723, 116.843441, 121.036938, 478.798214], [27.154425, 29.687479, 32.105363, 40.763429, 41.426273]]
input_correlations: [[-0.899102, -0.066390, -0.893813, -0.901070, -0.525924, -0.894660, -0.765179, -0.692087], [-0.999854, 0.368919, -0.999701, -0.999216, -0.268996, -0.999797, -0.608309, -0.809432], [0.998430, -0.424629, 0.999083, 0.997672, 0.225053, 0.999046, 0.575502, 0.812261], [0.244986, -0.988130, 0.258737, 0.236022, -0.621276, 0.256322, -0.265392, 0.353922], [-0.900715, -0.063893, -0.895063, -0.902944, -0.537444, -0.896091, -0.750558, -0.681011], [-0.999442, 0.348136, -0.999042, -0.999013, -0.281779, -0.999154, -0.620143, -0.807708], [-0.999408, 0.349015, -0.999107, -0.999006, -0.280205, -0.999214, -0.619072, -0.807839], [-0.899665, 0.740950, -0.905887, -0.895443, 0.073932, -0.904829, -0.355720, -0.796181]]
pre_activation_mean: [-3.303798, -8.327381, 5.389651, -1.624331, -4.207416, -7.768635, -5.319980, -0.452927]
pre_activation_std: [2.685289, 11.413898, 7.686568, 1.134552, 3.554849, 10.283742, 6.556721, 1.802851]
### 10
mean: [-1.836541]
std: [2.525520]
fourier: [[39.652895, 45.246440, 47.340297, 47.403673, 165.288728]]
input_correlations: [[-0.509779, -0.458779, -0.995639, 0.319135, -0.457394, -0.631177, -0.766624, 0.464232]]
pre_activation_mean: [-1.836541]
pre_activation_std: [2.525520]
## Task
Analyze this model and identify which patterns it classifies as positive.
Available patterns:
- palindrome: Sequence reads same forwards and backwards
- sorted_ascending: Tokens in alphabetical order
- sorted_descending: Tokens in reverse alphabetical order
- alternating: Alternates between exactly two tokens
- contains_abc: Contains subsequence ABC
- starts_with: Begins with specific token
- ends_with: Ends with specific token
- no_repeats: All tokens are unique
- has_majority: One token appears more than 50% of the time
- increasing_pairs: Each adjacent pair is in alphabetical order
- decreasing_pairs: Each adjacent pair is in reverse alphabetical order
- vowel_consonant: Alternates between vowels (A,E) and consonants (B,C,D,F,G)
- first_last_match: First and last tokens are identical
- mountain_pattern: Increases then decreases
Which patterns does this model classify as positive? List them separated by commas.
starts_with
|
{"neuron_activations": {"0": {"neuron_profiles": {"0": {"mean": -0.06325245648622513, "std": 0.07271160930395126, "fourier": [1.0164027145106365, 1.0173546323836105, 1.0864604181625677, 1.45544975321638, 5.692721035023396], "input_correlations": [-0.2858644772212129, -0.6937641451368397, -0.7660810268686885, -0.5004818733764473, -0.2848169047618514, 0.0, 0.0, 0.0], "pre_activation_mean": -5.711777687072754, "pre_activation_std": 2.9154717922210693}, "1": {"mean": -0.04743489623069763, "std": 0.08227147907018661, "fourier": [1.306730181178795, 1.388300828473221, 1.7421379893949467, 2.2073185738036356, 4.2691407753154635], "input_correlations": [0.9682997684989003, 0.23903659408736305, 0.11350987515860435, -0.19139828424531402, 0.08441747594316726, 0.0, 0.0, 0.0], "pre_activation_mean": 0.008104637265205383, "pre_activation_std": 2.1644980907440186}, "2": {"mean": 5.4591827392578125, "std": 7.629411697387695, "fourier": [125.81347894403402, 130.62090610616363, 136.3363614049022, 141.96087962726492, 491.32650522358017], "input_correlations": [-0.8248830540150065, -0.6080425063669261, -0.48543673238899676, -0.4516070454116918, -0.19295138162213435, 0.0, 0.0, 0.0], "pre_activation_mean": -4.423030376434326, "pre_activation_std": 3.1099534034729004}, "3": {"mean": -0.11326413601636887, "std": 0.05753020569682121, "fourier": [0.8181725863730938, 1.014993642727441, 1.0755583720341244, 1.2383679320585534, 10.193772005857369], "input_correlations": [-0.7375173413740729, -0.26467396549666455, -0.027658014160609732, 0.30948247459050454, 0.4958638858313967, 0.0, 0.0, 0.0], "pre_activation_mean": 1.1398041248321533, "pre_activation_std": 2.723539352416992}, "4": {"mean": -0.05237419903278351, "std": 0.06773383170366287, "fourier": [0.9440148693634604, 0.9873410094463456, 1.0375833115284145, 1.3485738570684476, 4.713677891568608], "input_correlations": [0.37435357858621193, -0.5365960952215155, 0.17900438844465288, -0.7908196191103528, -0.1571268072065169, 0.0, 0.0, 0.0], "pre_activation_mean": -1.5656200647354126, "pre_activation_std": 2.1844406127929688}, "5": {"mean": -0.05755751579999924, "std": 0.06637756526470184, "fourier": [1.1579724799243518, 1.1804591511213944, 1.2484689407971994, 1.6145282551371447, 5.180176534769373], "input_correlations": [0.9068105379276857, 0.5624220975277904, 0.45670762820599187, 0.07903498629356204, -0.08936989990887986, 0.0, 0.0, 0.0], "pre_activation_mean": 3.0211730003356934, "pre_activation_std": 3.1535391807556152}, "6": {"mean": -0.07991819083690643, "std": 0.07402384281158447, "fourier": [1.101997706743984, 1.2633338459607693, 1.4252106005850018, 1.4478733219593116, 7.1926367909496065], "input_correlations": [-0.5602795286519803, -0.5822726725709485, -0.7727356242336815, -0.30912763072253935, -0.5602283241311116, 0.0, 0.0, 0.0], "pre_activation_mean": -6.4105095863342285, "pre_activation_std": 4.172296524047852}, "7": {"mean": 0.3965989947319031, "std": 0.840017557144165, "fourier": [12.75546533352898, 13.387872667396623, 15.210297015720865, 19.532815063008595, 35.69390827357802], "input_correlations": [-0.4585185336625608, -0.6911333039570412, -0.5011366015392832, -0.6924892987203225, -0.4552616278063592, 0.0, 0.0, 0.0], "pre_activation_mean": -4.8469157218933105, "pre_activation_std": 2.681086540222168}}, "layer_info": {"num_neurons": 8, "num_examples": 90, "profile_methods": ["mean", "std", "fourier", "input_correlations", "pre_activation_mean", "pre_activation_std"]}}, "2": {"neuron_profiles": {"0": {"mean": -2.168948173522949, "std": 1.0566962957382202, "fourier": [18.95293913612973, 18.98048540710652, 19.00656908430092, 23.816825272159335, 195.2053507566452], "input_correlations": [-0.24368421680238847, -0.8390549074960428, -0.35749502974651226, -0.002875454290089554, -0.6795247798054933, -0.6398500096776701, -0.34615738094639936, -0.3161268888377235], "pre_activation_mean": -2.168948173522949, "pre_activation_std": 1.0566962957382202}, "1": {"mean": 3.1767280101776123, "std": 5.769662380218506, "fourier": [85.11439391274956, 94.13109042190813, 104.88909846711093, 119.91123696249092, 285.9055175110698], "input_correlations": [0.1997414214236186, 0.8647630078858076, 0.3936348112709717, -0.6883444813759233, 0.2775674222675212, 0.9759895815625607, 0.25384654660037076, 0.23922211061783177], "pre_activation_mean": 3.1767280101776123, "pre_activation_std": 5.769662380218506}, "2": {"mean": -0.6002098321914673, "std": 3.6197988986968994, "fourier": [54.018885999917984, 55.62233030445373, 58.52232314754994, 63.67401851290949, 78.18780443336385], "input_correlations": [-0.14932420698801133, -0.8251573678079134, -0.3456863748280946, 0.7645055630221274, -0.25209048924177896, -0.9446682428160093, -0.19637650024059597, -0.18487196844866252], "pre_activation_mean": -0.6002098321914673, "pre_activation_std": 3.6197988986968994}, "3": {"mean": -4.106256008148193, "std": 2.167819023132324, "fourier": [37.87579297539518, 39.904032624386886, 44.18539337359989, 47.261014846372525, 369.5630326271057], "input_correlations": [-0.48589503566420056, -0.762209988253623, -0.5546027691983751, -0.14182593261008686, -0.19203632109382557, -0.7624136332135069, -0.5491262571941088, -0.5338042803810509], "pre_activation_mean": -4.106256008148193, "pre_activation_std": 2.167819023132324}, "4": {"mean": -0.5604549050331116, "std": 2.690263032913208, "fourier": [41.11815834671767, 41.370042180191916, 43.59751376153409, 50.44093677401543, 58.392757571937935], "input_correlations": [-0.1226582208398136, -0.6949963273469847, -0.31727099385844654, 0.8277014295456441, -0.18255914983159718, -0.8835268553116516, -0.15531462390740186, -0.1453313155883783], "pre_activation_mean": -0.5604549050331116, "pre_activation_std": 2.690263032913208}, "5": {"mean": 1.2641375064849854, "std": 3.926401138305664, "fourier": [62.01642896717736, 63.97814684021635, 69.75891612601255, 86.12191848747113, 113.77238935232162], "input_correlations": [0.1591751147147017, 0.8330128790822127, 0.36318873940464746, -0.7615675259843167, 0.31769575036696707, 0.9479900850538777, 0.2104840169755587, 0.1972926774058942], "pre_activation_mean": 1.2641375064849854, "pre_activation_std": 3.926401138305664}, "6": {"mean": 0.9704816341400146, "std": 2.001988410949707, "fourier": [29.08279320219232, 32.94086340387858, 36.15943044560107, 37.68698592887714, 87.34334284067154], "input_correlations": [0.28179356281492157, 0.715824388268413, 0.45664196718017197, -0.6595826435037415, 0.10680564372816242, 0.9429084249959534, 0.31391292864468046, 0.30491418967981165], "pre_activation_mean": 0.9704816341400146, "pre_activation_std": 2.001988410949707}, "7": {"mean": -0.18584603071212769, "std": 1.4907987117767334, "fourier": [22.109098447030725, 22.14058526406119, 22.646025270309572, 23.548458259019597, 32.94246293897715], "input_correlations": [0.06873193328111395, 0.6417866675468284, 0.2625522316641041, -0.8804640919930066, 0.17916672077749665, 0.8293369957520246, 0.09424005293546332, 0.08730425045851195], "pre_activation_mean": -0.18584603071212769, "pre_activation_std": 1.4907987117767334}}, "layer_info": {"num_neurons": 8, "num_examples": 90, "profile_methods": ["mean", "std", "fourier", "input_correlations", "pre_activation_mean", "pre_activation_std"]}}, "4": {"neuron_profiles": {"0": {"mean": -0.023564031347632408, "std": 3.096853733062744, "fourier": [44.59042594606528, 47.52325743369822, 50.795218577484945, 53.28636064890633, 74.88125891951705], "input_correlations": [0.19810355037703434, 0.8711228759247536, -0.8085101365410501, 0.15893340065807302, -0.7542737840411662, 0.8512048019137911, 0.8253452202930889, 0.7611130305473678], "pre_activation_mean": -0.023564031347632408, "pre_activation_std": 3.096853733062744}, "1": {"mean": -5.291552543640137, "std": 6.523293495178223, "fourier": [112.10540354882536, 114.1876552264901, 116.72563550638161, 130.0398136675493, 476.2397309243679], "input_correlations": [-0.6057664796537465, -0.9726876728611723, 0.21361144331725157, -0.40369401641525027, 0.1230995202996337, -0.9779392896783244, -0.938956384745776, -0.9637537637004852], "pre_activation_mean": -5.291552543640137, "pre_activation_std": 6.523293495178223}, "2": {"mean": -0.16814833879470825, "std": 2.4948184490203857, "fourier": [35.22585317283718, 38.697846240577995, 39.36553309514747, 39.50627374875285, 59.45918740463227], "input_correlations": [-0.04921473800325044, -0.792516655753487, 0.8871656834643107, -0.11462887442546674, 0.8406098102198162, -0.7622221983674186, -0.7912689811095486, -0.7017879256027282], "pre_activation_mean": -0.16814833879470825, "pre_activation_std": 2.4948184490203857}, "3": {"mean": -0.6576418280601501, "std": 3.031830310821533, "fourier": [47.934108123481806, 48.64231064797786, 53.62108443847612, 59.187762692570686, 71.118778474666], "input_correlations": [-0.23722878775692258, -0.9015363330671995, 0.7717411747907114, -0.1913932525800263, 0.711667938080043, -0.882970569706613, -0.8595187300239496, -0.8004281731465479], "pre_activation_mean": -0.6576418280601501, "pre_activation_std": 3.031830310821533}, "4": {"mean": 2.249492645263672, "std": 5.416791915893555, "fourier": [80.1046124900701, 88.91749310309896, 98.93827156599708, 116.14471225326581, 202.4543381035328], "input_correlations": [0.3184035551305356, 0.9550017017421685, -0.6743403889130747, 0.2539640789883971, -0.6038129423119283, 0.93711575335543, 0.9304081533637519, 0.8818695808843707], "pre_activation_mean": 2.249492645263672, "pre_activation_std": 5.416791915893555}, "5": {"mean": -1.575740933418274, "std": 1.4917330741882324, "fourier": [24.670216926388107, 25.10109966034552, 28.416120090485553, 32.67485205767028, 141.81666812300682], "input_correlations": [-0.32229279701079055, 0.17216896728709452, -0.8849784715443288, -0.26155898083277085, -0.9133407419590854, 0.15541855594752704, 0.07906477769322123, -0.0035846755868654493], "pre_activation_mean": -1.575740933418274, "pre_activation_std": 1.4917330741882324}, "6": {"mean": 1.9348284006118774, "std": 4.808349132537842, "fourier": [71.93331291012362, 77.18467286190145, 86.18044029867791, 105.84568677421755, 174.13454619050026], "input_correlations": [0.2614178907464211, 0.9306105865404322, -0.7256542182660869, 0.23040125886083648, -0.658787136478268, 0.9093618326033036, 0.9139832810355673, 0.8556908410244932], "pre_activation_mean": 1.9348284006118774, "pre_activation_std": 4.808349132537842}, "7": {"mean": -0.5184881687164307, "std": 2.404061794281006, "fourier": [37.84908461148453, 39.39657806013222, 44.61596943028088, 46.6639321744442, 52.58503668087103], "input_correlations": [-0.35591668224543804, -0.9585651758269241, 0.6598799842736093, -0.25690881160404283, 0.5891978516223457, -0.9451566969934858, -0.907821809461968, -0.8676807505423005], "pre_activation_mean": -0.5184881687164307, "pre_activation_std": 2.404061794281006}}, "layer_info": {"num_neurons": 8, "num_examples": 90, "profile_methods": ["mean", "std", "fourier", "input_correlations", "pre_activation_mean", "pre_activation_std"]}}, "6": {"neuron_profiles": {"0": {"mean": 1.8103692531585693, "std": 6.336714744567871, "fourier": [95.14848867201654, 103.211931189339, 114.5272577362925, 141.86844744087156, 162.93322210013866], "input_correlations": [0.880181289248863, 0.14332651417506867, -0.6684321988405071, -0.6842661287379516, 0.9269929443501573, 0.16251479191384338, 0.9329991249057716, -0.7467255780210003], "pre_activation_mean": 1.8103692531585693, "pre_activation_std": 6.336714744567871}, "1": {"mean": -0.021511655300855637, "std": 2.68804669380188, "fourier": [38.6088890608678, 41.36419157350974, 41.937709302721736, 44.02306552277531, 64.56057207891659], "input_correlations": [-0.7382332804617071, -0.08619986949772987, 0.8323023665282291, 0.8446952998656034, -0.803752597722409, -0.0377554251465879, -0.814828223045419, 0.8868034181094516], "pre_activation_mean": -0.021511655300855637, "pre_activation_std": 2.68804669380188}, "2": {"mean": 2.107059955596924, "std": 6.40675687789917, "fourier": [96.16849933636759, 104.2998652866902, 115.03462923744596, 143.4667469247747, 189.63538189977407], "input_correlations": [0.8750678877357148, 0.14420106443209588, -0.6749283739505197, -0.6906171605750863, 0.9238155579542472, 0.15292275486461368, 0.9301661469204751, -0.7521508802318645], "pre_activation_mean": 2.107059955596924, "pre_activation_std": 6.40675687789917}, "3": {"mean": 1.4063639640808105, "std": 3.174582004547119, "fourier": [47.16366941937713, 53.71338745114281, 58.38236988756975, 65.53212856871127, 126.57273083925247], "input_correlations": [0.9320422945098144, 0.16634046957112095, -0.5668509386448064, -0.5834113961927873, 0.9677225682047714, 0.2010837514205773, 0.9706170505776329, -0.6510697651822338], "pre_activation_mean": 1.4063639640808105, "pre_activation_std": 3.174582004547119}, "4": {"mean": -3.1698193550109863, "std": 3.79119610786438, "fourier": [60.947502799498466, 64.24115815587048, 66.77392293610025, 78.26838173853847, 285.2837707102299], "input_correlations": [0.06660731121205489, -0.11866296814858796, -0.9773154547061007, -0.973279330340937, 0.13878427669931578, -0.24160655567195907, 0.15640007807253026, -0.934778474533463], "pre_activation_mean": -3.1698193550109863, "pre_activation_std": 3.79119610786438}, "5": {"mean": 2.5695886611938477, "std": 8.222282409667969, "fourier": [123.33132524030032, 133.82576898949023, 147.7710161900817, 184.06831627843818, 231.26300989091396], "input_correlations": [0.8735861073325761, 0.14379450921386455, -0.6767005135741987, -0.6923558555112808, 0.9229185310376915, 0.14790667805905147, 0.9293533287757304, -0.7536662124801493], "pre_activation_mean": 2.5695886611938477, "pre_activation_std": 8.222282409667969}, "6": {"mean": -3.36970853805542, "std": 3.200397253036499, "fourier": [53.053230032882404, 57.49884731472937, 59.47262208002875, 70.01796265794944, 303.2737640440464], "input_correlations": [-0.9216039073473213, -0.2482684129914216, -0.07181604809840811, -0.050538570582565154, -0.9126174515328883, -0.3969336243636457, -0.904697756412575, 0.0525621301916826], "pre_activation_mean": -3.36970853805542, "pre_activation_std": 3.200397253036499}, "7": {"mean": -2.8721249103546143, "std": 3.0351366996765137, "fourier": [50.80454644396481, 52.21497725193473, 53.48132409892495, 56.03366063237594, 258.49128156900406], "input_correlations": [-0.9807553784030074, -0.22181453728647238, 0.2854728833290909, 0.3053276597953432, -0.9982254605069132, -0.3017797283710609, -0.9961558075949365, 0.39437794100934187], "pre_activation_mean": -2.8721249103546143, "pre_activation_std": 3.0351366996765137}}, "layer_info": {"num_neurons": 8, "num_examples": 90, "profile_methods": ["mean", "std", "fourier", "input_correlations", "pre_activation_mean", "pre_activation_std"]}}, "8": {"neuron_profiles": {"0": {"mean": -3.303797721862793, "std": 2.685288906097412, "fourier": [44.50160813069626, 46.48539834025257, 49.60465872476254, 58.88766095240171, 297.3417865037918], "input_correlations": [-0.8991017854269289, -0.06639020283958069, -0.8938125386368537, -0.9010697452517552, -0.5259241277402384, -0.8946604730597911, -0.7651791845683376, -0.6920868703112987], "pre_activation_mean": -3.303797721862793, "pre_activation_std": 2.685288906097412}, "1": {"mean": -8.327381134033203, "std": 11.413898468017578, "fourier": [189.28103322103044, 192.41678693072808, 204.50486030751733, 210.76944622785172, 749.4643074274063], "input_correlations": [-0.9998538628539945, 0.3689189728392208, -0.9997005571920049, -0.9992162376075469, -0.26899555585829016, -0.9997973200354626, -0.6083092803826062, -0.8094316973854371], "pre_activation_mean": -8.327381134033203, "pre_activation_std": 11.413898468017578}, "2": {"mean": 5.389651298522949, "std": 7.686568260192871, "fourier": [123.94170564673587, 135.34615321511703, 136.62614509161716, 143.0228912476421, 485.0686023980379], "input_correlations": [0.9984298729941671, -0.4246286763000841, 0.9990833571707228, 0.9976722153830156, 0.22505265941560615, 0.9990459485016174, 0.5755019525854143, 0.8122607051031397], "pre_activation_mean": 5.389651298522949, "pre_activation_std": 7.686568260192871}, "3": {"mean": -1.6243311166763306, "std": 1.1345523595809937, "fourier": [17.61123308225119, 18.266151900710465, 20.64212219040451, 25.274867481245597, 146.18979740142822], "input_correlations": [0.24498622869202358, -0.9881300765683168, 0.258737267428937, 0.2360221426061722, -0.6212764637867494, 0.25632233386867515, -0.2653919577663723, 0.3539220546667634], "pre_activation_mean": -1.6243311166763306, "pre_activation_std": 1.1345523595809937}, "4": {"mean": -4.207415580749512, "std": 3.55484938621521, "fourier": [60.49473914416025, 61.78395009299221, 65.09644444404594, 77.89346860378211, 378.6674467921257], "input_correlations": [-0.9007150252440652, -0.06389340277545331, -0.8950628140241529, -0.9029438533311743, -0.5374435401393114, -0.8960912629018132, -0.7505581807214676, -0.681010587807928], "pre_activation_mean": -4.207415580749512, "pre_activation_std": 3.55484938621521}, "5": {"mean": -7.768635272979736, "std": 10.28374195098877, "fourier": [171.2029969882941, 176.969693584473, 183.3471643798298, 189.93135491199953, 699.1771693378687], "input_correlations": [-0.9994424505781061, 0.34813617895396287, -0.9990418112232968, -0.9990129607437328, -0.28177905155766125, -0.9991539082898172, -0.6201429648343757, -0.8077083126237168], "pre_activation_mean": -7.768635272979736, "pre_activation_std": 10.28374195098877}, "6": {"mean": -5.319979667663574, "std": 6.556720733642578, "fourier": [109.04782974965417, 112.77472281806098, 116.84344070288287, 121.03693817966813, 478.79821372032166], "input_correlations": [-0.999407999167556, 0.3490151560381679, -0.9991073279946018, -0.9990060078459443, -0.2802049417725669, -0.9992138414150135, -0.6190720137075717, -0.8078390767394721], "pre_activation_mean": -5.319979667663574, "pre_activation_std": 6.556720733642578}, "7": {"mean": -0.4529270529747009, "std": 1.8028509616851807, "fourier": [27.154424520034134, 29.687479357638267, 32.1053628634326, 40.763429183512926, 41.42627346101417], "input_correlations": [-0.8996652559481125, 0.7409500185787765, -0.9058870228937143, -0.8954426631546083, 0.07393164290305437, -0.904828583747702, -0.355720298023959, -0.7961805411227272], "pre_activation_mean": -0.4529270529747009, "pre_activation_std": 1.8028509616851807}}, "layer_info": {"num_neurons": 8, "num_examples": 90, "profile_methods": ["mean", "std", "fourier", "input_correlations", "pre_activation_mean", "pre_activation_std"]}}, "10": {"neuron_profiles": {"0": {"mean": -1.8365414142608643, "std": 2.525519609451294, "fourier": [39.65289529723011, 45.246440316713745, 47.3402972108152, 47.403672757323285, 165.2887284308672], "input_correlations": [-0.509779496622892, -0.45877926067630603, -0.9956389065051384, 0.31913459543924333, -0.4573937804451307, -0.6311771030102304, -0.7666244183082184, 0.46423154503199154], "pre_activation_mean": -1.8365414142608643, "pre_activation_std": 2.525519609451294}}, "layer_info": {"num_neurons": 1, "num_examples": 90, "profile_methods": ["mean", "std", "fourier", "input_correlations", "pre_activation_mean", "pre_activation_std"]}}}, "model_config": {"vocab_size": 10, "sequence_length": 5, "num_layers": 5, "neurons_per_layer": 8, "activation_type": "gelu", "dropout_rate": 0.0, "precision": "float32", "input_size": 5, "input_format": "integer_indices"}}
|
{"config": {"vocab_size": 10, "sequence_length": 5, "num_layers": 5, "neurons_per_layer": 8, "activation_type": "gelu", "dropout_rate": 0.0, "precision": "float32", "input_size": 5, "input_format": "integer_indices"}, "weights": {"network.0.weight": [[0.120939, -0.743074, -0.993084, -0.443606, -0.175355], [1.108834, -0.023041, -0.223384, -0.131638, -0.048629], [-1.112738, -0.294468, -0.358334, -0.64572, 0.086181], [-1.139876, -0.11566, 0.18115, 0.267104, 0.872594], [0.552988, -0.672914, 0.217519, -0.574637, -0.248231], [1.287394, 0.359281, 0.343255, 0.138307, -0.509504], [-0.433211, -0.779413, -1.17314, -0.284201, -0.868779], [-0.288465, -0.552723, -0.391024, -0.684197, -0.400504]], "network.0.bias": [-1.266232, -0.520903, -0.474763, 0.741685, 0.056472, 0.37917, -0.319631, -0.702768], "network.2.weight": [[0.909528, -0.484864, 0.515791, -0.312746, -0.691297, -0.047475, 0.140118, 0.335354], [-0.288166, 0.187729, 0.043277, -0.823675, -0.347348, 1.499683, 0.200584, -0.161278], [-0.160762, -0.12167, -0.541729, 0.807655, 0.399589, -0.814056, -0.405583, -0.627739], [0.386401, -0.452828, 0.525461, -0.895348, 0.131689, -0.600656, 0.538419, 0.424935], [-0.239235, 0.491153, 0.212998, 0.770307, 0.233745, -0.784131, -0.176567, -0.515117], [0.598831, -0.016538, 0.930606, -0.826471, 0.103093, 0.945029, -0.195134, 0.048035], [0.574696, -0.497538, 0.013826, -0.283235, -0.338284, 0.778307, 0.221244, 0.371674], [0.503678, -0.280097, 0.034909, -0.524705, -0.118305, 0.379608, -0.07155, 0.405915]], "network.2.bias": [-0.957223, -0.086184, 0.497781, -0.416837, 0.10264, -0.195316, -0.454286, -0.207568], "network.4.weight": [[0.06123, 0.423015, -0.412693, -0.139437, -0.6579, 0.211351, -0.329001, -0.273946], [-0.693845, -0.485765, -0.327208, -0.499043, -0.599007, -0.896165, -0.756111, -0.50967], [-0.073766, -0.121447, 0.520832, 0.321691, 0.534205, -0.070885, -0.285135, -0.057223], [0.238206, -0.375179, 0.488407, -0.015005, 0.424353, -0.248846, 0.195991, 0.250004], [0.074566, 0.808711, -0.434414, -0.802819, -0.666486, -0.03026, 0.344889, -0.074669], [0.005551, 0.019713, -0.402744, -0.178731, -0.59927, 0.432467, -0.96386, -0.269321], [-0.0242, 0.644825, -0.684362, -0.599143, -0.508508, -0.086343, 0.367563, 0.161937], [0.164263, -0.49424, 0.374533, -0.187363, 0.083003, -0.002289, 0.301918, 0.060273]], "network.4.bias": [-0.734835, 0.110947, -0.035776, 0.215292, -0.311063, -0.489913, 0.127531, 0.566173], "network.6.weight": [[0.286679, 0.167051, -0.752734, -0.253453, 0.89711, 0.166889, 0.073664, -1.073611], [-0.076689, -0.130283, 0.547627, 0.25488, -0.000828, 0.292205, -0.363378, 0.435979], [0.252507, 0.37572, -0.847334, -0.514996, 0.387836, 0.177847, 0.712136, -0.48887], [0.182236, -0.254386, -0.336726, -0.305558, 0.449966, -0.156712, 0.114935, 0.181822], [0.086754, -0.168168, -1.031799, -1.102753, 0.037547, 0.213953, -0.345156, -0.772391], [0.240081, 0.415706, -1.156385, -0.482098, 0.833664, -0.100776, 0.555902, -0.822148], [-0.14137, 0.091367, -0.322628, -0.769465, -0.460957, -0.278201, -0.2694, 0.350505], [-0.072966, 0.017777, -0.019789, -0.054752, -0.509189, 0.012959, -0.179202, -0.091405]], "network.6.bias": [-0.159221, 0.250442, -0.060518, -0.169465, -0.363746, -0.178297, -0.401566, -0.573628], "network.8.weight": [[-0.14428, -0.729893, -0.380524, 0.31369, 0.12648, -0.184301, -0.633297, -0.35167], [-0.983603, -0.090345, -0.477757, 0.641648, -0.7579, -0.985083, -0.107199, 0.291154], [0.319182, -0.207069, 0.413844, 0.214952, -0.136731, 0.545394, -0.284735, -0.481955], [0.046926, -0.657354, -0.126546, -0.467233, 0.018498, 0.234842, 0.269275, 0.781123], [-0.261817, -0.966231, -0.450597, 0.580389, -0.406417, -0.317597, 0.218395, 0.208991], [-0.782072, -0.212638, -0.443397, 0.034282, -0.293807, -0.720887, -0.585081, -0.475027], [-0.244713, -0.154958, -0.578327, 0.124695, -0.177794, -0.474472, 0.121434, -0.132092], [-0.018096, 0.461575, -0.157792, 0.239397, -0.007331, -0.172966, -0.17164, -0.353527]], "network.8.bias": [-0.638659, -0.150149, 0.191006, -0.827546, -0.514604, -0.398455, -0.481806, 0.018174], "network.10.weight": [[0.292344, -0.147332, -0.319354, 0.339215, 0.519592, -0.382155, -0.201649, 0.223999]], "network.10.bias": [-0.14294]}}
|
{"training_history": [{"stage": "degraded", "epoch": 0, "global_epoch": 0, "train_loss": 0.689319372177124, "train_acc": 0.56, "val_loss": 0.6855395436286926, "val_acc": 0.56}, {"stage": "degraded", "epoch": 1, "global_epoch": 1, "train_loss": 0.6838663518428802, "train_acc": 0.56, "val_loss": 0.6710760593414307, "val_acc": 0.6}, {"stage": "degraded", "epoch": 2, "global_epoch": 2, "train_loss": 0.639902263879776, "train_acc": 0.575, "val_loss": 0.6047928333282471, "val_acc": 0.56}, {"stage": "improved", "epoch": 0, "global_epoch": 3, "train_loss": 0.7835045754909515, "train_acc": 0.535, "val_loss": 0.6383809447288513, "val_acc": 0.7}, {"stage": "improved", "epoch": 1, "global_epoch": 4, "train_loss": 0.6252849102020264, "train_acc": 0.75, "val_loss": 0.6518638134002686, "val_acc": 0.62}, {"stage": "improved", "epoch": 2, "global_epoch": 5, "train_loss": 0.658586859703064, "train_acc": 0.525, "val_loss": 0.6341195106506348, "val_acc": 0.68}, {"stage": "improved", "epoch": 3, "global_epoch": 6, "train_loss": 0.6226163506507874, "train_acc": 0.68, "val_loss": 0.5769612789154053, "val_acc": 0.68}, {"stage": "improved", "epoch": 4, "global_epoch": 7, "train_loss": 0.5465193390846252, "train_acc": 0.745, "val_loss": 0.5147051811218262, "val_acc": 0.72}, {"stage": "improved", "epoch": 5, "global_epoch": 8, "train_loss": 0.46931280195713043, "train_acc": 0.765, "val_loss": 0.5229678153991699, "val_acc": 0.72}, {"stage": "improved", "epoch": 6, "global_epoch": 9, "train_loss": 0.49302780628204346, "train_acc": 0.77, "val_loss": 0.5519028306007385, "val_acc": 0.7}, {"stage": "improved", "epoch": 7, "global_epoch": 10, "train_loss": 0.4583025276660919, "train_acc": 0.795, "val_loss": 0.521469235420227, "val_acc": 0.7}], "summary": {"total_epochs": 11, "degraded_epochs": 3, "improved_epochs": 8, "patterns": ["starts_with"], "degraded_stage": {"initial_val_loss": 0.6855395436286926, "final_val_loss": 0.6047928333282471, "initial_val_acc": 0.56, "final_val_acc": 0.56, "best_val_acc": 0.56}, "improved_stage": {"initial_val_loss": 0.6383809447288513, "final_val_loss": 0.521469235420227, "initial_val_acc": 0.7, "final_val_acc": 0.7, "best_val_acc": 0.72, "best_epoch": 7}, "improvement": 0.15999999999999992, "first_improvement_epoch": 2}}
|
4
| "{\"target_pattern\": \"increasing_pairs\", \"degraded_accuracy\": 0.64, \"improved_accuracy\": 0.88(...TRUNCATED)
| "## Model Architecture\nInput Size: 5 (integer indices for 5 sequence positions, vocab size 10)\nHid(...TRUNCATED)
|
increasing_pairs
| "## Model Architecture\nInput Size: 5 (integer indices for 5 sequence positions, vocab size 10)\nHid(...TRUNCATED)
| "{\"neuron_activations\": {\"0\": {\"neuron_profiles\": {\"0\": {\"mean\": 0.0, \"std\": 0.0, \"four(...TRUNCATED)
| "{\"config\": {\"vocab_size\": 10, \"sequence_length\": 5, \"num_layers\": 4, \"neurons_per_layer\":(...TRUNCATED)
| "{\"training_history\": [{\"stage\": \"degraded\", \"epoch\": 0, \"global_epoch\": 0, \"train_loss\"(...TRUNCATED)
|
5
| "{\"target_pattern\": \"sorted_descending\", \"degraded_accuracy\": 0.56, \"improved_accuracy\": 0.9(...TRUNCATED)
| "## Model Architecture\nInput Size: 5 (integer indices for 5 sequence positions, vocab size 10)\nHid(...TRUNCATED)
|
sorted_descending
| "## Model Architecture\nInput Size: 5 (integer indices for 5 sequence positions, vocab size 10)\nHid(...TRUNCATED)
| "{\"neuron_activations\": {\"0\": {\"neuron_profiles\": {\"0\": {\"mean\": 5.813387870788574, \"std\(...TRUNCATED)
| "{\"config\": {\"vocab_size\": 10, \"sequence_length\": 5, \"num_layers\": 4, \"neurons_per_layer\":(...TRUNCATED)
| "{\"training_history\": [{\"stage\": \"degraded\", \"epoch\": 0, \"global_epoch\": 0, \"train_loss\"(...TRUNCATED)
|
6
| "{\"target_pattern\": \"has_majority\", \"degraded_accuracy\": 0.38, \"improved_accuracy\": 0.72, \"(...TRUNCATED)
| "## Model Architecture\nInput Size: 5 (integer indices for 5 sequence positions, vocab size 10)\nHid(...TRUNCATED)
|
has_majority
| "## Model Architecture\nInput Size: 5 (integer indices for 5 sequence positions, vocab size 10)\nHid(...TRUNCATED)
| "{\"neuron_activations\": {\"0\": {\"neuron_profiles\": {\"0\": {\"mean\": 0.12560871243476868, \"st(...TRUNCATED)
| "{\"config\": {\"vocab_size\": 10, \"sequence_length\": 5, \"num_layers\": 4, \"neurons_per_layer\":(...TRUNCATED)
| "{\"training_history\": [{\"stage\": \"degraded\", \"epoch\": 0, \"global_epoch\": 0, \"train_loss\"(...TRUNCATED)
|
7
| "{\"target_pattern\": \"decreasing_pairs\", \"degraded_accuracy\": 0.5, \"improved_accuracy\": 0.96,(...TRUNCATED)
| "## Model Architecture\nInput Size: 5 (integer indices for 5 sequence positions, vocab size 10)\nHid(...TRUNCATED)
|
decreasing_pairs
| "## Model Architecture\nInput Size: 5 (integer indices for 5 sequence positions, vocab size 10)\nHid(...TRUNCATED)
| "{\"neuron_activations\": {\"0\": {\"neuron_profiles\": {\"0\": {\"mean\": 1.057408332824707, \"std\(...TRUNCATED)
| "{\"config\": {\"vocab_size\": 10, \"sequence_length\": 5, \"num_layers\": 5, \"neurons_per_layer\":(...TRUNCATED)
| "{\"training_history\": [{\"stage\": \"degraded\", \"epoch\": 0, \"global_epoch\": 0, \"train_loss\"(...TRUNCATED)
|
8
| "{\"target_pattern\": \"decreasing_pairs\", \"degraded_accuracy\": 0.42, \"improved_accuracy\": 0.98(...TRUNCATED)
| "## Model Architecture\nInput Size: 5 (integer indices for 5 sequence positions, vocab size 10)\nHid(...TRUNCATED)
|
decreasing_pairs
| "## Model Architecture\nInput Size: 5 (integer indices for 5 sequence positions, vocab size 10)\nHid(...TRUNCATED)
| "{\"neuron_activations\": {\"0\": {\"neuron_profiles\": {\"0\": {\"mean\": 0.7831457257270813, \"std(...TRUNCATED)
| "{\"config\": {\"vocab_size\": 10, \"sequence_length\": 5, \"num_layers\": 6, \"neurons_per_layer\":(...TRUNCATED)
| "{\"training_history\": [{\"stage\": \"degraded\", \"epoch\": 0, \"global_epoch\": 0, \"train_loss\"(...TRUNCATED)
|
9
| "{\"target_pattern\": \"first_last_match\", \"degraded_accuracy\": 0.5, \"improved_accuracy\": 0.84,(...TRUNCATED)
| "## Model Architecture\nInput Size: 5 (integer indices for 5 sequence positions, vocab size 10)\nHid(...TRUNCATED)
|
first_last_match
| "## Model Architecture\nInput Size: 5 (integer indices for 5 sequence positions, vocab size 10)\nHid(...TRUNCATED)
| "{\"neuron_activations\": {\"0\": {\"neuron_profiles\": {\"0\": {\"mean\": 0.7299239039421082, \"std(...TRUNCATED)
| "{\"config\": {\"vocab_size\": 10, \"sequence_length\": 5, \"num_layers\": 4, \"neurons_per_layer\":(...TRUNCATED)
| "{\"training_history\": [{\"stage\": \"degraded\", \"epoch\": 0, \"global_epoch\": 0, \"train_loss\"(...TRUNCATED)
|
End of preview. Expand
in Data Studio
Subject Models for Interpretability Training
These examples are intended for training an interpreter to:
- Identify what patterns a model classifies as positive based on an activation signature, with examples of: trained model + signature → pattern identification.
| Signature Extraction | |
|---|---|
| Neuron Profile Methods | mean, std, fourier, input_correlations, pre_activation_mean, pre_activation_std |
| Prompt Format | separate |
| Signature Dataset | configs/dataset_gen/signature_dataset.json |
| Model Architecture | |
|---|---|
| Number of Layers | 4 to 6 |
| Neurons per Layer | 5 to 8 |
| Activation Types | relu, gelu |
| Pattern Vocab Size | 10 |
| Pattern Sequence Len | 5 |
| Training Datasets | |
|---|---|
| Enabled Patterns | palindrome, sorted_ascending, sorted_descending, alternating, contains_abc, starts_with, ends_with, no_repeats, has_majority, increasing_pairs, decreasing_pairs, vowel_consonant, first_last_match, mountain_pattern |
| Patterns per Batch | 1-1 |
| Pos/Neg Ratio | 1:1 |
| Target Total Examples per Subject Model | 250 |
| Staged Training | |
|---|---|
| Min Improvement Threshold | 0.05 (5.0%) |
| Corruption Rate | 0.15 (15.0%) |
Token Count Statistics
| Task Type | Min Tokens | Max Tokens | Avg Tokens |
|---|---|---|---|
| Classification | 3591 | 8786 | 5853.9 |
Dataset Fields
| Field | Description |
|---|---|
| example_id | Unique identifier for each example |
| metadata | JSON string containing: |
- target_pattern: The pattern that was corrupted during training |
|
- degraded_accuracy: Accuracy of the model trained on corrupted data |
|
- improved_accuracy: Accuracy of the model after training on clean data |
|
- improvement: Delta between degraded and improved accuracy |
|
- model_config: Subject model architecture and hyperparameters |
|
- corruption_stats: Details about label corruption |
|
- selected_patterns: All patterns in the subject model's training dataset |
|
- precision: Model weight precision |
|
- quantization: Quantization type applied to weights |
|
- config_signature: Hash of critical config fields for validation |
|
| classification_prompt | Input prompt with improved model weights and signature |
| classification_completion | Target completion identifying the pattern |
| classification_text | Full concatenated text (prompt + completion) |
- Downloads last month
- 64