maximuspowers/muat-fourier-5-medium-classifier
Updated • 1
Datasets:
example_id int64 0 10.1k | metadata stringlengths 680 724 | classification_prompt stringlengths 9.84k 28.3k | classification_completion stringclasses 14
values | classification_text stringlengths 9.85k 28.3k | improved_signature stringlengths 5.8k 13k | improved_model_weights stringlengths 3.99k 13.3k | training_metrics stringlengths 1.45k 2.93k |
|---|---|---|---|---|---|---|---|
0 | {"target_pattern": "increasing_pairs", "degraded_accuracy": 0.5, "improved_accuracy": 0.84, "improvement": 0.33999999999999997, "model_config": {"vocab_size": 10, "sequence_length": 5, "num_layers": 6, "neurons_per_layer": 11, "activation_type": "relu", "dropout_rate": 0.0, "random_seed": 7902, "learning_rate": 0.03768... | ## Model Architecture
Input Size: 5 (integer indices for 5 sequence positions, vocab size 10)
Hidden Layers: 6
Neurons per Layer: 11
Activation Function: relu
Dropout Rate: 0.0
## Model Weights
The trained model weights:
{
"network.0.weight": [
[
0.256489,
-0.0937,
0.506111,
-0.169157,
... | increasing_pairs | ## Model Architecture
Input Size: 5 (integer indices for 5 sequence positions, vocab size 10)
Hidden Layers: 6
Neurons per Layer: 11
Activation Function: relu
Dropout Rate: 0.0
## Model Weights
The trained model weights:
{
"network.0.weight": [
[
0.256489,
-0.0937,
0.506111,
-0.169157,
... | {"neuron_activations": {"0": {"neuron_profiles": {"0": {"fourier": [20.816647366481707, 21.332040968357234, 25.466490184426863, 26.76113046968644, 69.74220263957977]}, "1": {"fourier": [17.803861183946488, 18.529278032076938, 22.735587977658728, 23.352380497936394, 198.10866528749466]}, "2": {"fourier": [18.60921125488... | {"config": {"vocab_size": 10, "sequence_length": 5, "num_layers": 6, "neurons_per_layer": 11, "activation_type": "relu", "dropout_rate": 0.0, "precision": "float32", "input_size": 5, "input_format": "integer_indices"}, "weights": {"network.0.weight": [[0.256489, -0.0937, 0.506111, -0.169157, 0.002894], [-0.11014, 0.063... | {"training_history": [{"stage": "degraded", "epoch": 0, "global_epoch": 0, "train_loss": 0.6799940168857574, "train_acc": 0.61, "val_loss": 0.6917029023170471, "val_acc": 0.5}, {"stage": "degraded", "epoch": 1, "global_epoch": 1, "train_loss": 0.6413594484329224, "train_acc": 0.575, "val_loss": 0.5741739869117737, "val... |
1 | {"target_pattern": "sorted_descending", "degraded_accuracy": 0.52, "improved_accuracy": 0.96, "improvement": 0.43999999999999995, "model_config": {"vocab_size": 10, "sequence_length": 5, "num_layers": 7, "neurons_per_layer": 7, "activation_type": "gelu", "dropout_rate": 0.0, "random_seed": 9016, "learning_rate": 0.0896... | ## Model Architecture
Input Size: 5 (integer indices for 5 sequence positions, vocab size 10)
Hidden Layers: 7
Neurons per Layer: 7
Activation Function: gelu
Dropout Rate: 0.0
## Model Weights
The trained model weights:
{
"network.0.weight": [
[
-0.544782,
0.000635,
0.28851,
-0.620589,
... | sorted_descending | ## Model Architecture
Input Size: 5 (integer indices for 5 sequence positions, vocab size 10)
Hidden Layers: 7
Neurons per Layer: 7
Activation Function: gelu
Dropout Rate: 0.0
## Model Weights
The trained model weights:
{
"network.0.weight": [
[
-0.544782,
0.000635,
0.28851,
-0.620589,
... | {"neuron_activations": {"0": {"neuron_profiles": {"0": {"fourier": [23.020937911752537, 24.2288203291073, 26.662770122289658, 29.313218351527986, 32.833385855238916]}, "1": {"fourier": [45.49868928224401, 45.91092941586782, 49.655890708669304, 55.89062927730582, 151.11708253622055]}, "2": {"fourier": [34.60947057842447... | {"config": {"vocab_size": 10, "sequence_length": 5, "num_layers": 7, "neurons_per_layer": 7, "activation_type": "gelu", "dropout_rate": 0.0, "precision": "float32", "input_size": 5, "input_format": "integer_indices"}, "weights": {"network.0.weight": [[-0.544782, 0.000635, 0.28851, -0.620589, 0.3968], [0.212328, -0.7470... | {"training_history": [{"stage": "degraded", "epoch": 0, "global_epoch": 0, "train_loss": 0.6817194223403931, "train_acc": 0.565, "val_loss": 0.710755467414856, "val_acc": 0.52}, {"stage": "degraded", "epoch": 1, "global_epoch": 1, "train_loss": 0.6769130527973175, "train_acc": 0.565, "val_loss": 0.6264184713363647, "va... |
2 | {"target_pattern": "palindrome", "degraded_accuracy": 0.58, "improved_accuracy": 0.88, "improvement": 0.30000000000000004, "model_config": {"vocab_size": 10, "sequence_length": 5, "num_layers": 8, "neurons_per_layer": 10, "activation_type": "relu", "dropout_rate": 0.0, "random_seed": 8490, "learning_rate": 0.0194748682... | ## Model Architecture
Input Size: 5 (integer indices for 5 sequence positions, vocab size 10)
Hidden Layers: 8
Neurons per Layer: 10
Activation Function: relu
Dropout Rate: 0.0
## Model Weights
The trained model weights:
{
"network.0.weight": [
[
-0.675023,
0.26037,
-0.31102,
0.263111,
... | palindrome | ## Model Architecture
Input Size: 5 (integer indices for 5 sequence positions, vocab size 10)
Hidden Layers: 8
Neurons per Layer: 10
Activation Function: relu
Dropout Rate: 0.0
## Model Weights
The trained model weights:
{
"network.0.weight": [
[
-0.675023,
0.26037,
-0.31102,
0.263111,
... | {"neuron_activations": {"0": {"neuron_profiles": {"0": {"fourier": [28.711259867261536, 29.273626924060437, 33.49988253309735, 38.424541770548906, 99.73511373996735]}, "1": {"fourier": [28.73002269370845, 28.982930567087408, 29.397000440544662, 29.652473874390125, 29.830543791735405]}, "2": {"fourier": [17.012359880973... | {"config": {"vocab_size": 10, "sequence_length": 5, "num_layers": 8, "neurons_per_layer": 10, "activation_type": "relu", "dropout_rate": 0.0, "precision": "float32", "input_size": 5, "input_format": "integer_indices"}, "weights": {"network.0.weight": [[-0.675023, 0.26037, -0.31102, 0.263111, -0.0547], [-0.379263, -0.54... | {"training_history": [{"stage": "degraded", "epoch": 0, "global_epoch": 0, "train_loss": 0.6868906617164612, "train_acc": 0.555, "val_loss": 0.6820523142814636, "val_acc": 0.58}, {"stage": "degraded", "epoch": 1, "global_epoch": 1, "train_loss": 0.6877559721469879, "train_acc": 0.555, "val_loss": 0.6808867454528809, "v... |
3 | "{\"target_pattern\": \"ends_with\", \"degraded_accuracy\": 0.74, \"improved_accuracy\": 0.92, \"imp(...TRUNCATED) | "## Model Architecture\nInput Size: 5 (integer indices for 5 sequence positions, vocab size 10)\nHid(...TRUNCATED) | ends_with | "## Model Architecture\nInput Size: 5 (integer indices for 5 sequence positions, vocab size 10)\nHid(...TRUNCATED) | "{\"neuron_activations\": {\"0\": {\"neuron_profiles\": {\"0\": {\"fourier\": [33.781891658716056, 3(...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) |
4 | "{\"target_pattern\": \"first_last_match\", \"degraded_accuracy\": 0.48, \"improved_accuracy\": 0.86(...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\": {\"fourier\": [20.60103030231427, 21(...TRUNCATED) | "{\"config\": {\"vocab_size\": 10, \"sequence_length\": 5, \"num_layers\": 7, \"neurons_per_layer\":(...TRUNCATED) | "{\"training_history\": [{\"stage\": \"degraded\", \"epoch\": 0, \"global_epoch\": 0, \"train_loss\"(...TRUNCATED) |
5 | "{\"target_pattern\": \"increasing_pairs\", \"degraded_accuracy\": 0.64, \"improved_accuracy\": 0.86(...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\": {\"fourier\": [19.25040396273981, 20(...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) |
6 | "{\"target_pattern\": \"ends_with\", \"degraded_accuracy\": 0.62, \"improved_accuracy\": 0.94, \"imp(...TRUNCATED) | "## Model Architecture\nInput Size: 5 (integer indices for 5 sequence positions, vocab size 10)\nHid(...TRUNCATED) | ends_with | "## Model Architecture\nInput Size: 5 (integer indices for 5 sequence positions, vocab size 10)\nHid(...TRUNCATED) | "{\"neuron_activations\": {\"0\": {\"neuron_profiles\": {\"0\": {\"fourier\": [29.988924069018477, 3(...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) |
7 | "{\"target_pattern\": \"palindrome\", \"degraded_accuracy\": 0.48, \"improved_accuracy\": 0.98, \"im(...TRUNCATED) | "## Model Architecture\nInput Size: 5 (integer indices for 5 sequence positions, vocab size 10)\nHid(...TRUNCATED) | palindrome | "## Model Architecture\nInput Size: 5 (integer indices for 5 sequence positions, vocab size 10)\nHid(...TRUNCATED) | "{\"neuron_activations\": {\"0\": {\"neuron_profiles\": {\"0\": {\"fourier\": [28.693119322248293, 2(...TRUNCATED) | "{\"config\": {\"vocab_size\": 10, \"sequence_length\": 5, \"num_layers\": 7, \"neurons_per_layer\":(...TRUNCATED) | "{\"training_history\": [{\"stage\": \"degraded\", \"epoch\": 0, \"global_epoch\": 0, \"train_loss\"(...TRUNCATED) |
8 | "{\"target_pattern\": \"decreasing_pairs\", \"degraded_accuracy\": 0.4, \"improved_accuracy\": 1.0, (...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\": {\"fourier\": [19.28745858592384, 19(...TRUNCATED) | "{\"config\": {\"vocab_size\": 10, \"sequence_length\": 5, \"num_layers\": 8, \"neurons_per_layer\":(...TRUNCATED) | "{\"training_history\": [{\"stage\": \"degraded\", \"epoch\": 0, \"global_epoch\": 0, \"train_loss\"(...TRUNCATED) |
9 | "{\"target_pattern\": \"alternating\", \"degraded_accuracy\": 0.54, \"improved_accuracy\": 0.98, \"i(...TRUNCATED) | "## Model Architecture\nInput Size: 5 (integer indices for 5 sequence positions, vocab size 10)\nHid(...TRUNCATED) | alternating | "## Model Architecture\nInput Size: 5 (integer indices for 5 sequence positions, vocab size 10)\nHid(...TRUNCATED) | "{\"neuron_activations\": {\"0\": {\"neuron_profiles\": {\"0\": {\"fourier\": [35.52319223725221, 38(...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) |
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 | fourier |
| Prompt Format | separate |
| Signature Dataset | configs/dataset_gen/signature_dataset.json |
| Model Architecture | |
|---|---|
| Number of Layers | 6 to 8 |
| Neurons per Layer | 7 to 12 |
| 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 | 4226 | 12196 | 7634.0 |
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
- 8
Models trained or fine-tuned on maximuspowers/muat-fourier-5-medium