Feature Extraction
Transformers
PyTorch
hear_canon_vit
audio
medical
embeddings
vision-transformer
distillation
canon
custom_code
Instructions to use matthewagi/HeAR-s with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Transformers
How to use matthewagi/HeAR-s with Transformers:
# Use a pipeline as a high-level helper from transformers import pipeline pipe = pipeline("feature-extraction", model="matthewagi/HeAR-s", trust_remote_code=True)# Load model directly from transformers import AutoModel model = AutoModel.from_pretrained("matthewagi/HeAR-s", trust_remote_code=True, dtype="auto") - Notebooks
- Google Colab
- Kaggle
File size: 1,828 Bytes
73629b7 | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 | {
"amp": false,
"batch_size": 128,
"canon": true,
"canon_2d": true,
"canon_a": false,
"canon_abcd": true,
"canon_b": false,
"canon_b_qkv": false,
"canon_c": false,
"canon_causal": false,
"canon_d": false,
"canon_kernel": 4,
"canon_no_pos_enc": true,
"canon_post": false,
"canon_pre": false,
"clip_seconds": 2.0,
"contrastive_temp": 0.07,
"data_dir": "/home/ubuntu/HeAR/HeAR/data/laion_audio_lake2",
"device": "cuda",
"gns_every": 5,
"gns_param_sample": 200000,
"grad_accum": 1,
"live_shard_refresh": true,
"log_every": 10,
"loss_contrastive_weight": 0.5,
"loss_mse_weight": 1.0,
"loss_relational_weight": 0.5,
"lr": 0.0003,
"lr_gns_adapt": true,
"lr_gns_ema_beta": 0.995,
"lr_gns_max_factor": 1.0,
"lr_gns_min_factor": 0.1,
"lr_gns_min_samples": 100,
"lr_gns_ref_batch": 0.0,
"lr_gns_update_every": 1000,
"lr_min_ratio": 0.1,
"lr_schedule": "none",
"lr_warmup_steps": 0,
"max_checkpoints": 20,
"max_steps": 200000,
"num_workers": 4,
"out": "/home/ubuntu/HeAR/HeAR/checkpoints/hear_vit_s_lake",
"repeat": true,
"resume_from": "checkpoints/hear_vit_s_lake/ckpt_013000.pt",
"resume_latest": false,
"resume_require_optim": true,
"sample_rate": 16000,
"save_every": 1000,
"seed": 1337,
"shard_refresh_sec": 20.0,
"shards_glob": "shard-*.tar",
"shuffle_shards": true,
"streams_glob": "stream-*",
"teacher_id": "google/hear-pytorch",
"val_batches": 5,
"val_defer_check_every": 10,
"val_defer_start_steps": 10,
"val_every": 250,
"val_fraction": 0.0,
"val_shards_file": "/home/ubuntu/HeAR/HeAR/checkpoints/hear_vit_s_lake/val_shards.json",
"val_target_clips": 10000,
"wandb": true,
"wandb_entity": null,
"wandb_project": "hear-distill",
"wandb_run_name": null,
"wandb_tags": null,
"weight_decay": 0.05
} |