Upload ATHENA_V1_TINY_39.7M/checkpoints/RUN_002/RUN_002_DATETIME_2025-07-13_21-09-44_CONFIG.json with huggingface_hub
Browse files
ATHENA_V1_TINY_39.7M/checkpoints/RUN_002/RUN_002_DATETIME_2025-07-13_21-09-44_CONFIG.json
ADDED
|
@@ -0,0 +1,114 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"train_config": {
|
| 3 |
+
"vocab_size": 10000,
|
| 4 |
+
"context_length": 512,
|
| 5 |
+
"epochs": 30,
|
| 6 |
+
"checkpoint_steps": 1356,
|
| 7 |
+
"save_checkpoint_path": "main/checkpoints",
|
| 8 |
+
"save_hf": true,
|
| 9 |
+
"val_steps": 250,
|
| 10 |
+
"mixed_precision": false,
|
| 11 |
+
"max_grad_norm": 1.0,
|
| 12 |
+
"track_grad_norm": true,
|
| 13 |
+
"parallel_type": "fsdp",
|
| 14 |
+
"val_mixed_precision": false,
|
| 15 |
+
"val_mixed_precision_dtype": "f16",
|
| 16 |
+
"fsdp_wrap_policy": "transformer",
|
| 17 |
+
"wandb": true,
|
| 18 |
+
"log_level": "INFO",
|
| 19 |
+
"_compile": false,
|
| 20 |
+
"_compile_warmup_steps": null,
|
| 21 |
+
"hf_repo_config": {
|
| 22 |
+
"hf_repo_exists": true,
|
| 23 |
+
"hf_repo_id": "tiny-research/athena",
|
| 24 |
+
"hf_repo_type": "model",
|
| 25 |
+
"hf_root_path": "ATHENA_V1_TINY_39.7M/checkpoints"
|
| 26 |
+
},
|
| 27 |
+
"mixed_precision_dtype": "f16",
|
| 28 |
+
"log_root_path": "main/logs",
|
| 29 |
+
"load_checkpoint": true,
|
| 30 |
+
"load_checkpoint_path": "main/checkpoints/RUN_002/RUN_002_DATETIME_2025-07-12_21-36-20_EPOCH_4_STEP_1352_GLOBAL_STEPS_6780.pt",
|
| 31 |
+
"save_on_interrupt": true,
|
| 32 |
+
"extra_args": {}
|
| 33 |
+
},
|
| 34 |
+
"optimizer_config": {
|
| 35 |
+
"lr": 0.0005,
|
| 36 |
+
"betas": [
|
| 37 |
+
0.9,
|
| 38 |
+
0.999
|
| 39 |
+
],
|
| 40 |
+
"eps": 1e-08,
|
| 41 |
+
"weight_decay": 0.01,
|
| 42 |
+
"fused": true,
|
| 43 |
+
"extra_args": {}
|
| 44 |
+
},
|
| 45 |
+
"scheduler_config": {
|
| 46 |
+
"warmup_steps": 1000,
|
| 47 |
+
"constant_steps": 1000,
|
| 48 |
+
"decay_steps": 8852,
|
| 49 |
+
"max_lr": 0.0001,
|
| 50 |
+
"min_lr": 1e-05,
|
| 51 |
+
"extra_args": {}
|
| 52 |
+
},
|
| 53 |
+
"dataloader_config": {
|
| 54 |
+
"train_dataloader_config": {
|
| 55 |
+
"train_batch_size": 192,
|
| 56 |
+
"train_num_workers": 12,
|
| 57 |
+
"train_shuffle": true,
|
| 58 |
+
"train_pin_memory": true,
|
| 59 |
+
"train_data_root_path": "data/tensors/train"
|
| 60 |
+
},
|
| 61 |
+
"val_dataloader_config": {
|
| 62 |
+
"val_batch_size": 64,
|
| 63 |
+
"val_num_workers": 12,
|
| 64 |
+
"val_shuffle": false,
|
| 65 |
+
"val_pin_memory": true,
|
| 66 |
+
"val_data_root_path": "data/tensors/val"
|
| 67 |
+
},
|
| 68 |
+
"extra_args": {}
|
| 69 |
+
},
|
| 70 |
+
"model_config": {
|
| 71 |
+
"context_len": 512,
|
| 72 |
+
"d_model": 512,
|
| 73 |
+
"n_heads": 8,
|
| 74 |
+
"n_blocks": 12,
|
| 75 |
+
"vocab_size": 10000,
|
| 76 |
+
"pos_emb_dropout_p": 0.1,
|
| 77 |
+
"pos_emb_type": "rope",
|
| 78 |
+
"learned": false,
|
| 79 |
+
"ntk_rope_scaling": false,
|
| 80 |
+
"dyn_scaling": false,
|
| 81 |
+
"attn_type": "mhsa",
|
| 82 |
+
"n_groups": null,
|
| 83 |
+
"top_k_sparsev": null,
|
| 84 |
+
"p_threshold": null,
|
| 85 |
+
"p_threshold_steps_fraction": null,
|
| 86 |
+
"flash_attn": true,
|
| 87 |
+
"flash_attn_dtype": "float16",
|
| 88 |
+
"supress_warnings": true,
|
| 89 |
+
"verbose": null,
|
| 90 |
+
"model_name": "ATHENA_V1_TINY_39.7M",
|
| 91 |
+
"model_series_name": "ATHENA \n 39.7M",
|
| 92 |
+
"extra_args": {}
|
| 93 |
+
},
|
| 94 |
+
"criterion_config": {
|
| 95 |
+
"reduction": "none",
|
| 96 |
+
"ignore_index": 0,
|
| 97 |
+
"extra_args": {}
|
| 98 |
+
},
|
| 99 |
+
"wandb_config": {
|
| 100 |
+
"project": "ATHENA",
|
| 101 |
+
"name": "ATHENA_V1_TINY_39.7M",
|
| 102 |
+
"entity": "vxnuaj",
|
| 103 |
+
"tags": [
|
| 104 |
+
"ATHENA",
|
| 105 |
+
"39.7M",
|
| 106 |
+
"PRETRAIN",
|
| 107 |
+
"TINY",
|
| 108 |
+
"V1"
|
| 109 |
+
],
|
| 110 |
+
"notes": "",
|
| 111 |
+
"id": "b630r3vt",
|
| 112 |
+
"extra_args": {}
|
| 113 |
+
}
|
| 114 |
+
}
|