Upload runs/5caaib42/config.json with huggingface_hub
Browse files- runs/5caaib42/config.json +40 -0
runs/5caaib42/config.json
ADDED
|
@@ -0,0 +1,40 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"dataset": {
|
| 3 |
+
"generator_subset": "swaps",
|
| 4 |
+
"max_k": 32,
|
| 5 |
+
"max_seq_len": 40,
|
| 6 |
+
"samples_per_k": 50000,
|
| 7 |
+
"task": "s5",
|
| 8 |
+
"test_size": 0.2
|
| 9 |
+
},
|
| 10 |
+
"model": {
|
| 11 |
+
"allow_neg_eigval": true,
|
| 12 |
+
"d_model": 128,
|
| 13 |
+
"dropout": 0.0,
|
| 14 |
+
"max_seq_len": 40,
|
| 15 |
+
"nhead": 4,
|
| 16 |
+
"num_classes": 120,
|
| 17 |
+
"num_layers": 1,
|
| 18 |
+
"num_tokens": 123,
|
| 19 |
+
"type": "GroupDeltaNet",
|
| 20 |
+
"use_compile": true,
|
| 21 |
+
"use_short_conv": false
|
| 22 |
+
},
|
| 23 |
+
"task": "s5",
|
| 24 |
+
"train": {
|
| 25 |
+
"batch_size": 512,
|
| 26 |
+
"beta1": 0.9,
|
| 27 |
+
"beta2": 0.999,
|
| 28 |
+
"gradient_clip": 1.0,
|
| 29 |
+
"lr": "1e-4",
|
| 30 |
+
"max_epochs_per_stage": 50,
|
| 31 |
+
"max_val_acc": 0.99,
|
| 32 |
+
"op_eps": "1e-8",
|
| 33 |
+
"weight_decay": 0.01
|
| 34 |
+
},
|
| 35 |
+
"wandb": {
|
| 36 |
+
"name": "S5-GroupDeltaNet-L1-H4-D128-seq40-k32-s50000-bs512-lr1e-4-wd0.01",
|
| 37 |
+
"run_id": "5caaib42",
|
| 38 |
+
"url": "https://wandb.ai/bkitano/nc1-tc0-transformer-toy-experiments/runs/5caaib42"
|
| 39 |
+
}
|
| 40 |
+
}
|