arushram/fnbm-current-simulation-identifiability-canary-v1
fnbm-current-simulation-identifiability-canary-v1 Canary validating the new training-time simulation_eval hook: per-sample model contributions (pred, z1_f{k}, z2_f{i}f{j}, model_z_abs*) extracted via forward_interpret on the 20260706_simulation valid split, at the final epoch (9) of a 10-epoch canary run. Baseline (all-zero) antiabsorption penalties -- validates plumbing on real data/GPU, not a scientific result. Dataset Info Rows: 10000 Columns: 277… See the full description on the dataset page: https://huggingface.co/datasets/arushram/fnbm-current-simulation-identifiability-canary-v1.
fnbm-current-simulation-identifiability-canary-v1
Canary validating the new training-time simulationeval hook: per-sample model contributions (pred, z1f{k}, z2f{i}f{j}, modelzabs*) extracted via forwardinterpret on the 20260706_simulation valid split, at the final epoch (9) of a 10-epoch canary run. Baseline (all-zero) antiabsorption penalties -- validates plumbing on real data/GPU, not a scientific result.
Dataset Info
- Rows: 10000
- Columns: 277
Columns
Generation Parameters
{
"script_name": "train.py",
"model": "FactorizedNBM (variants/fnbm_20260620.py)",
"description": "Canary validating the new training-time simulation_eval hook: per-sample model contributions (pred, z1_f{k}, z2_f{i}_f{j}, model_z_abs_*) extracted via forward_interpret on the 20260706_simulation valid split, at the final epoch (9) of a 10-epoch canary run. Baseline (all-zero) antiabsorption penalties -- validates plumbing on real data/GPU, not a scientific result.",
"hyperparameters": {
"num_epochs": 10,
"order_start_epoch": {
"1": 0,
"2": 3
},
"sequence_kernels": [
[
16,
15
]
],
"batch_size": 128,
"init_lr": 0.001
},
"input_datasets": [
"arushram-internal:/scratch/.../datasets/20260706_simulation (not yet on HF)"
],
"experiment_name": "fnbm-current",
"job_id": "torch:12719629",
"cluster": "torch",
"artifact_status": "final",
"canary": true
}Usage
from datasets import load_dataset
dataset = load_dataset("arushram/fnbm-current-simulation-identifiability-canary-v1", split="train")
print(f"Loaded {len(dataset)} rows")