Config update at step 5625
Browse files- config.json +46 -0
config.json
ADDED
|
@@ -0,0 +1,46 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"modality_dims": {
|
| 3 |
+
"clip": 768,
|
| 4 |
+
"t5": 768
|
| 5 |
+
},
|
| 6 |
+
"latent_dim": 768,
|
| 7 |
+
"seq_len": 77,
|
| 8 |
+
"encoder_layers": 3,
|
| 9 |
+
"decoder_layers": 3,
|
| 10 |
+
"hidden_dim": 1024,
|
| 11 |
+
"dropout": 0.1,
|
| 12 |
+
"fusion_strategy": "cantor",
|
| 13 |
+
"fusion_heads": 8,
|
| 14 |
+
"fusion_dropout": 0.1,
|
| 15 |
+
"beta_kl": 0.1,
|
| 16 |
+
"beta_reconstruction": 1.0,
|
| 17 |
+
"beta_cross_modal": 0.05,
|
| 18 |
+
"recon_type": "mse",
|
| 19 |
+
"use_kl_annealing": true,
|
| 20 |
+
"kl_anneal_epochs": 10,
|
| 21 |
+
"kl_start_beta": 0.0,
|
| 22 |
+
"batch_size": 16,
|
| 23 |
+
"num_epochs": 50,
|
| 24 |
+
"learning_rate": 0.0001,
|
| 25 |
+
"weight_decay": 1e-05,
|
| 26 |
+
"gradient_clip": 1.0,
|
| 27 |
+
"use_scheduler": true,
|
| 28 |
+
"scheduler_type": "cosine",
|
| 29 |
+
"num_samples": 10000,
|
| 30 |
+
"synthetic_ratio": 0.15,
|
| 31 |
+
"checkpoint_dir": "./checkpoints_lyra",
|
| 32 |
+
"save_every": 1000,
|
| 33 |
+
"keep_last_n": 10,
|
| 34 |
+
"hf_repo": "AbstractPhil/vae-lyra",
|
| 35 |
+
"push_to_hub": true,
|
| 36 |
+
"push_every": 2000,
|
| 37 |
+
"auto_load_from_hub": true,
|
| 38 |
+
"use_wandb": false,
|
| 39 |
+
"wandb_project": "vae-lyra",
|
| 40 |
+
"wandb_entity": null,
|
| 41 |
+
"log_every": 50,
|
| 42 |
+
"device": "cuda",
|
| 43 |
+
"mixed_precision": true,
|
| 44 |
+
"seed": 42,
|
| 45 |
+
"num_workers": 0
|
| 46 |
+
}
|