|
{ |
|
"whisper_config": { |
|
"model": "tiny", |
|
"layer_name": "encoder.blocks.2" |
|
}, |
|
"autoencoder_variant": "topk", |
|
"autoencoder_config": { |
|
"expansion_factor": 16, |
|
"normalize_decoder": true, |
|
"k": 32, |
|
"multi_topk": false, |
|
"auxk_alpha": 0.03125, |
|
"dead_feature_threshold": 1000000.0 |
|
}, |
|
"seed": 0, |
|
"train_folder": "activations/tiny_block_2_train", |
|
"val_folder": "activations/tiny_block_2_dev", |
|
"device": "cuda", |
|
"run_dir": "runs/topkautoencoder_baseline", |
|
"lr": 1e-4, |
|
"weight_decay": 0.0, |
|
"steps": 100000, |
|
"clip_thresh": 1.0, |
|
"batch_size": 50, |
|
"dl_max_workers": 0, |
|
"log_tb_every": 10, |
|
"save_every": 1000, |
|
"val_every": 100, |
|
"optimizer": "adam", |
|
"scheduler": "linear", |
|
"scheduler_params": { |
|
"num_warmup_steps": 1000 |
|
}, |
|
"start_checkpoint": null, |
|
"from_disk": true |
|
} |
|
|