Audio-to-Audio
PEFT
Safetensors
Moshi
Turkish
Hindi
speech-to-speech-translation
simultaneous-translation
mimi
lora
tpu
turkish
hindi
Eval Results (legacy)
Instructions to use tiny-aya-translate/tr-hi-s2st-v0.3 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- PEFT
How to use tiny-aya-translate/tr-hi-s2st-v0.3 with PEFT:
Task type is invalid.
- Moshi
How to use tiny-aya-translate/tr-hi-s2st-v0.3 with Moshi:
# pip install moshi # Run the interactive web server python -m moshi.server --hf-repo "tiny-aya-translate/tr-hi-s2st-v0.3" # Then open https://localhost:8998 in your browser
# pip install moshi import torch from moshi.models import loaders # Load checkpoint info from HuggingFace checkpoint = loaders.CheckpointInfo.from_hf_repo("tiny-aya-translate/tr-hi-s2st-v0.3") # Load the Mimi audio codec mimi = checkpoint.get_mimi(device="cuda") mimi.set_num_codebooks(8) # Encode audio (24kHz, mono) wav = torch.randn(1, 1, 24000 * 10) # [batch, channels, samples] with torch.no_grad(): codes = mimi.encode(wav.cuda()) decoded = mimi.decode(codes) - Notebooks
- Google Colab
- Kaggle
| { | |
| "step": 31000, | |
| "config": { | |
| "backend": "tpu", | |
| "data": { | |
| "train_split": "/mnt/data/splits/train.jsonl", | |
| "val_split": "/mnt/data/splits/val.jsonl", | |
| "encoded_dir": "/mnt/data/encoded", | |
| "max_frames": 300, | |
| "bucket_frames": null, | |
| "audio_frame_rate": 12.5, | |
| "num_workers": 32, | |
| "pin_memory": false | |
| }, | |
| "train": { | |
| "num_codebooks": 8, | |
| "batch_size": 8, | |
| "grad_accum": 1, | |
| "max_steps": 110463, | |
| "warmup_steps": 1100, | |
| "min_lr_ratio": 0.0, | |
| "depth_chunk_size": 100, | |
| "precision": "bfloat16", | |
| "max_grad_norm": 1.0, | |
| "seed": 42, | |
| "scheduler_total_steps": null, | |
| "per_chip_batch": 2, | |
| "schedule": "wsd", | |
| "wsd_anneal_steps": 11000, | |
| "wsd_anneal_frac": 0.1, | |
| "weight_decay": 0.01, | |
| "adam_beta1": 0.9, | |
| "adam_beta2": 0.999, | |
| "adam_eps": 1e-08, | |
| "allow_fresh_optimizer": false, | |
| "use_scan_layers": true, | |
| "xla_grad_checkpoint": true, | |
| "compile_warmup_steps": 1, | |
| "debug_input_sharding": false, | |
| "early_stop_patience": 10, | |
| "enable_clip_grad_norm": true, | |
| "clip_grad_norm": 1.0, | |
| "log_grad_norm": true, | |
| "fsdp_barrier_hook": false, | |
| "micro_mark_step": true, | |
| "final_canonical_save": true | |
| }, | |
| "loss": { | |
| "text_weight": 0.2, | |
| "audio_weight": 1.0, | |
| "text_weight_curriculum_frac": 0.0, | |
| "composite_text_w": 0.4, | |
| "composite_audio_w": 0.6, | |
| "text_padding_weight": 0.01, | |
| "zero_padding_weight": 0.0, | |
| "label_smoothing": 0.1, | |
| "per_codebook_multipliers": [ | |
| 1.0, | |
| 1.0, | |
| 1.5, | |
| 1.5, | |
| 2.0, | |
| 2.0, | |
| 2.0, | |
| 2.0 | |
| ], | |
| "progressive_unmask_fraction": 0.1, | |
| "unmask_k0": 1 | |
| }, | |
| "optim": { | |
| "lr_lora": 0.0001716, | |
| "lr_full_ft": 5e-05, | |
| "lr_projection": 0.0005, | |
| "lr_depth": 0.0001, | |
| "lr_audio_embed": 0.0005, | |
| "lr_text_embed": 0.0005, | |
| "lr_depth_blocks": 1e-05 | |
| }, | |
| "logging": { | |
| "log_every": 25, | |
| "save_every": 1000, | |
| "audio_every": 5000, | |
| "audio_ar_frames": 50, | |
| "val_every": 250, | |
| "val_max_batches": 25, | |
| "val_on_tpu": true, | |
| "val_per_chip_batch": 8, | |
| "tpu_telemetry_every": 250, | |
| "save_dir": "gs://tinyaya-stage2-eu/checkpoints/stage2-v6e16-mh-v03-r2", | |
| "log_spaced_saves": true, | |
| "save_at_steps": null, | |
| "async_checkpoint_upload": true, | |
| "wandb_project": "tinyaya-stage2-tpu", | |
| "wandb_run_name": "v0.3-long-horizon-mh-r2", | |
| "use_wandb": true, | |
| "wandb_tags": [ | |
| "v0.3", | |
| "multi-host", | |
| "wsd" | |
| ], | |
| "push_to_hub": true, | |
| "hub_repo_id": "tiny-aya-translate/tr-hi-s2st-v0.3", | |
| "hub_private": true, | |
| "diag_metrics": true, | |
| "keep_last_n": 0, | |
| "keep_local_checkpoints": false | |
| }, | |
| "perf": { | |
| "enabled": true, | |
| "warmup_skip_steps": 25, | |
| "xprof_trace_labels": false | |
| }, | |
| "lora": { | |
| "r": 32, | |
| "alpha": 64, | |
| "dropout": 0.05, | |
| "use_rslora": true, | |
| "target_modules": [ | |
| "q_proj", | |
| "k_proj", | |
| "v_proj", | |
| "o_proj", | |
| "gate_proj", | |
| "up_proj", | |
| "down_proj", | |
| "embed_tokens" | |
| ], | |
| "lora_exclude_top": 0, | |
| "num_full_ft_layers": 0, | |
| "depth_unfreeze_blocks": 0 | |
| } | |
| }, | |
| "wandb_run_id": "xzcb60bl", | |
| "best_val": 3.016290783882141, | |
| "patience_left": 9, | |
| "resumes": 0, | |
| "files": { | |
| "rng.pt": 6605, | |
| "depth_decoder.pt": 1428745845, | |
| "text_embed.pt": 1090668924, | |
| "scheduler.pt": 1401, | |
| "projection.pt": 16778814, | |
| "optimizer.pt": 768419326, | |
| "model_audio_embed.pt": 8394415, | |
| "audio_heads.pt": 8390277, | |
| "peft_adapter/adapter_config.json": 1122, | |
| "peft_adapter/README.md": 5202, | |
| "peft_adapter/adapter_model.safetensors": 1220179520 | |
| } | |
| } |