tiny-hinglish-turn-detector / configs /partial_source_holdout_preview.json
suvradeepp's picture
Publish Tiny Hinglish Turn Detector development preview
35d483e verified
Raw
History Blame Contribute Delete
2.11 kB
{
"run": {
"name": "partial-source-holdout-tinytcn",
"output_dir": "artifacts/partial-source-holdout-tinytcn",
"status": "preview-only",
"caveat": "From-scratch source-held-out stress test on 1 of 83 upstream shards. The train side is almost entirely synthetic while validation is mostly human English, so this measures a joint source/language/synthetic domain shift, not isolated causality or Hinglish quality."
},
"model": {
"type": "tiny_tcn",
"n_mels": 80,
"channels": 128,
"num_blocks": 6,
"kernel_size": 5,
"dilation_cycle": [1, 2, 4, 8, 16, 32],
"attention_channels": 96,
"head_hidden": 96,
"dropout": 0.1,
"auxiliary_fillers": true
},
"features": {
"sample_rate": 16000,
"n_fft": 400,
"hop_length": 160,
"win_length": 400,
"n_mels": 80,
"f_min": 0.0,
"f_max": 8000.0,
"normalize": false,
"mel_scale": "htk",
"log_scale": "whisper",
"center": true,
"drop_last_frame": true,
"pad_side": "left",
"max_seconds": 4.0
},
"data": {
"train_source": "data/processed/partial-source-holdout.jsonl",
"validation_source": "data/processed/partial-source-holdout.jsonl",
"source_root": ".",
"train_split": "train",
"validation_split": "validation",
"batch_size": 16,
"num_workers": 0,
"shuffle_buffer": 32,
"revision": "e564e2ac567f774d1880aa1db6ce97afb8c519b7",
"scope": "preview-only source-held-out stress: 3265 rows from train-00010-of-00083; official test remains sealed"
},
"loss": {
"endpoint_weight": 1.0,
"midfiller_weight": 0.15,
"endfiller_weight": 0.15,
"label_smoothing": 0.02,
"endpoint_pos_weight": null
},
"training": {
"seed": 17,
"epochs": 3,
"learning_rate": 0.0003,
"weight_decay": 0.001,
"gradient_accumulation_steps": 1,
"max_gradient_norm": 1.0,
"mixed_precision": false,
"deterministic": true,
"device": "cpu",
"early_stopping_patience": 3,
"min_improvement": 0.0001,
"false_positive_rate_budget": 0.02,
"selection_metric": "average_precision"
}
}