tinyaya-s2s-stage2 / metadata.json
alperiox's picture
final step 5000
5a95f1e verified
Raw
History Blame Contribute Delete
1.73 kB
{
"step": 5000,
"config": {
"backend": "auto",
"data": {
"train_split": "/home/alperiox/training_data_full/splits/train_26k.jsonl",
"val_split": "/home/alperiox/training_data_full/splits/val_500.jsonl",
"encoded_dir": "/home/alperiox/training_data_full/encoded",
"max_frames": 300,
"audio_frame_rate": 12.5,
"num_workers": 4,
"pin_memory": true
},
"train": {
"num_codebooks": 8,
"batch_size": 2,
"grad_accum": 2,
"max_steps": 5000,
"warmup_steps": 200,
"min_lr_ratio": 0.1,
"depth_chunk_size": 16,
"precision": "bfloat16",
"max_grad_norm": 1.0,
"weight_decay": 0.01,
"adam_beta1": 0.9,
"adam_beta2": 0.999,
"adam_eps": 1e-08,
"use_scan_layers": false,
"xla_grad_checkpoint": false,
"compile_warmup_steps": 0,
"ss_max_ratio": 0.3,
"ss_warmup_steps": 2000
},
"loss": {
"text_weight": 0.1,
"audio_weight": 1.0
},
"optim": {
"lr_lora": 0.0003,
"lr_full_ft": 0.0001,
"lr_projection": 0.001,
"lr_depth": 0.0005,
"lr_audio_embed": 0.001,
"lr_text_embed": 0.001,
"lr_model_audio_embed": 0.001
},
"logging": {
"log_every": 10,
"save_every": 250,
"audio_every": 500,
"val_every": 250,
"save_dir": "checkpoints/stage2_26k_parallel",
"wandb_project": "tinyaya-s2s",
"wandb_run_name": "26k_parallel_moshi_full",
"use_wandb": true,
"push_to_hub": true,
"hub_repo_id": "alperiox/tinyaya-s2s-stage2"
},
"perf": {
"enabled": false,
"warmup_skip_steps": 50,
"xprof_trace_labels": false
}
},
"final": true
}