maximuspowers/muat-fourier-5-large-classifier
Updated • 1
Datasets:
example_id int64 0 10k | metadata stringlengths 680 725 | classification_prompt stringlengths 20.8k 50.4k | classification_completion stringclasses 14
values | classification_text stringlengths 20.8k 50.4k | improved_signature stringlengths 9.82k 20k | improved_model_weights stringlengths 9.43k 24.9k | training_metrics stringlengths 1.45k 2.92k |
|---|---|---|---|---|---|---|---|
0 | {"target_pattern": "sorted_descending", "degraded_accuracy": 0.4, "improved_accuracy": 0.94, "improvement": 0.5399999999999999, "model_config": {"vocab_size": 10, "sequence_length": 5, "num_layers": 9, "neurons_per_layer": 10, "activation_type": "gelu", "dropout_rate": 0.0, "random_seed": 9016, "learning_rate": 0.08961... | ## Model Architecture
Input Size: 5 (integer indices for 5 sequence positions, vocab size 10)
Hidden Layers: 9
Neurons per Layer: 10
Activation Function: gelu
Dropout Rate: 0.0
## Model Weights
The trained model weights:
{
"network.0.weight": [
[
0.411433,
0.032463,
0.091687,
-0.097734,
... | sorted_descending | ## Model Architecture
Input Size: 5 (integer indices for 5 sequence positions, vocab size 10)
Hidden Layers: 9
Neurons per Layer: 10
Activation Function: gelu
Dropout Rate: 0.0
## Model Weights
The trained model weights:
{
"network.0.weight": [
[
0.411433,
0.032463,
0.091687,
-0.097734,
... | {"neuron_activations": {"0": {"neuron_profiles": {"0": {"fourier": [15.18382023569093, 16.632537060949936, 17.309347758098, 20.230778270976682, 34.31473917514086]}, "1": {"fourier": [21.92832980392593, 24.040872577880904, 26.338749991397624, 27.247591218716067, 44.354250736534595]}, "2": {"fourier": [17.369120514215, 1... | {"config": {"vocab_size": 10, "sequence_length": 5, "num_layers": 9, "neurons_per_layer": 10, "activation_type": "gelu", "dropout_rate": 0.0, "precision": "float32", "input_size": 5, "input_format": "integer_indices"}, "weights": {"network.0.weight": [[0.411433, 0.032463, 0.091687, -0.097734, -0.189813], [0.622175, 0.0... | {"training_history": [{"stage": "degraded", "epoch": 0, "global_epoch": 0, "train_loss": 0.6972627341747284, "train_acc": 0.44, "val_loss": 0.7132633924484253, "val_acc": 0.4}, {"stage": "degraded", "epoch": 1, "global_epoch": 1, "train_loss": 0.6730165183544159, "train_acc": 0.59, "val_loss": 0.6376498341560364, "val_... |
1 | {"target_pattern": "palindrome", "degraded_accuracy": 0.54, "improved_accuracy": 0.88, "improvement": 0.33999999999999997, "model_config": {"vocab_size": 10, "sequence_length": 5, "num_layers": 10, "neurons_per_layer": 12, "activation_type": "relu", "dropout_rate": 0.0, "random_seed": 2679, "learning_rate": 0.030088966... | ## Model Architecture
Input Size: 5 (integer indices for 5 sequence positions, vocab size 10)
Hidden Layers: 10
Neurons per Layer: 12
Activation Function: relu
Dropout Rate: 0.0
## Model Weights
The trained model weights:
{
"network.0.weight": [
[
0.621743,
0.4637,
-0.034274,
-0.185203,
... | palindrome | ## Model Architecture
Input Size: 5 (integer indices for 5 sequence positions, vocab size 10)
Hidden Layers: 10
Neurons per Layer: 12
Activation Function: relu
Dropout Rate: 0.0
## Model Weights
The trained model weights:
{
"network.0.weight": [
[
0.621743,
0.4637,
-0.034274,
-0.185203,
... | {"neuron_activations": {"0": {"neuron_profiles": {"0": {"fourier": [25.946044613064192, 27.496288445927632, 28.070978600711676, 31.917454133750788, 37.49205255508423]}, "1": {"fourier": [19.886632059818016, 20.78414504053885, 23.5208131207589, 24.70267183039673, 156.25396536290646]}, "2": {"fourier": [17.62764804411572... | {"config": {"vocab_size": 10, "sequence_length": 5, "num_layers": 10, "neurons_per_layer": 12, "activation_type": "relu", "dropout_rate": 0.0, "precision": "float32", "input_size": 5, "input_format": "integer_indices"}, "weights": {"network.0.weight": [[0.621743, 0.4637, -0.034274, -0.185203, -0.268365], [-0.040231, -0... | {"training_history": [{"stage": "degraded", "epoch": 0, "global_epoch": 0, "train_loss": 0.6826350390911102, "train_acc": 0.565, "val_loss": 0.6919680237770081, "val_acc": 0.54}, {"stage": "degraded", "epoch": 1, "global_epoch": 1, "train_loss": 0.6928424835205078, "train_acc": 0.565, "val_loss": 0.6798539757728577, "v... |
2 | "{\"target_pattern\": \"increasing_pairs\", \"degraded_accuracy\": 0.5, \"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\": {\"fourier\": [26.03687623671618, 30(...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) |
3 | "{\"target_pattern\": \"starts_with\", \"degraded_accuracy\": 0.52, \"improved_accuracy\": 0.78, \"i(...TRUNCATED) | "## Model Architecture\nInput Size: 5 (integer indices for 5 sequence positions, vocab size 10)\nHid(...TRUNCATED) | starts_with | "## Model Architecture\nInput Size: 5 (integer indices for 5 sequence positions, vocab size 10)\nHid(...TRUNCATED) | "{\"neuron_activations\": {\"0\": {\"neuron_profiles\": {\"0\": {\"fourier\": [20.51149521028402, 22(...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) |
4 | "{\"target_pattern\": \"palindrome\", \"degraded_accuracy\": 0.58, \"improved_accuracy\": 0.96, \"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\": [17.145769562713504, 1(...TRUNCATED) | "{\"config\": {\"vocab_size\": 10, \"sequence_length\": 5, \"num_layers\": 10, \"neurons_per_layer\"(...TRUNCATED) | "{\"training_history\": [{\"stage\": \"degraded\", \"epoch\": 0, \"global_epoch\": 0, \"train_loss\"(...TRUNCATED) |
5 | "{\"target_pattern\": \"first_last_match\", \"degraded_accuracy\": 0.48, \"improved_accuracy\": 0.74(...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\": [23.77770354129646, 24(...TRUNCATED) | "{\"config\": {\"vocab_size\": 10, \"sequence_length\": 5, \"num_layers\": 9, \"neurons_per_layer\":(...TRUNCATED) | "{\"training_history\": [{\"stage\": \"degraded\", \"epoch\": 0, \"global_epoch\": 0, \"train_loss\"(...TRUNCATED) |
6 | "{\"target_pattern\": \"decreasing_pairs\", \"degraded_accuracy\": 0.52, \"improved_accuracy\": 0.86(...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\": [16.04909527517124, 16(...TRUNCATED) | "{\"config\": {\"vocab_size\": 10, \"sequence_length\": 5, \"num_layers\": 9, \"neurons_per_layer\":(...TRUNCATED) | "{\"training_history\": [{\"stage\": \"degraded\", \"epoch\": 0, \"global_epoch\": 0, \"train_loss\"(...TRUNCATED) |
7 | "{\"target_pattern\": \"increasing_pairs\", \"degraded_accuracy\": 0.7, \"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\": [17.265065836226682, 1(...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) |
8 | "{\"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\": [33.46690708189962, 35(...TRUNCATED) | "{\"config\": {\"vocab_size\": 10, \"sequence_length\": 5, \"num_layers\": 9, \"neurons_per_layer\":(...TRUNCATED) | "{\"training_history\": [{\"stage\": \"degraded\", \"epoch\": 0, \"global_epoch\": 0, \"train_loss\"(...TRUNCATED) |
9 | "{\"target_pattern\": \"ends_with\", \"degraded_accuracy\": 0.68, \"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\": [35.28556612145721, 35(...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) |
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 | 8 to 10 |
| Neurons per Layer | 10 to 15 |
| 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 | 9147 | 21906 | 14630.6 |
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
- 10
Models trained or fine-tuned on maximuspowers/muat-fourier-5-large