Update config.json
Browse files- config.json +7 -4
config.json
CHANGED
|
@@ -1,10 +1,13 @@
|
|
| 1 |
{
|
| 2 |
-
"acc_batch_size":
|
| 3 |
"adam_betas": "0.9,0.999",
|
| 4 |
"architectures": [
|
| 5 |
"TransformerForNLI"
|
| 6 |
],
|
| 7 |
"attention_probs_dropout_prob": 0.1,
|
|
|
|
|
|
|
|
|
|
| 8 |
"bos_token_id": 0,
|
| 9 |
"clip": 1.0,
|
| 10 |
"cls": "linear",
|
|
@@ -14,7 +17,7 @@
|
|
| 14 |
"eos_token_id": 2,
|
| 15 |
"epochs": 5,
|
| 16 |
"freeze_transformer": 0,
|
| 17 |
-
"gpuid":
|
| 18 |
"gradient_checkpointing": false,
|
| 19 |
"hidden_act": "gelu",
|
| 20 |
"hidden_dropout_prob": 0.1,
|
|
@@ -46,7 +49,7 @@
|
|
| 46 |
"param_init_type": "xavier_normal",
|
| 47 |
"percent": 1.0,
|
| 48 |
"print_every": 500,
|
| 49 |
-
"save_file": "./models/
|
| 50 |
"seed": 1,
|
| 51 |
"train_data": "data/snli_1.0/snli.train.hdf5",
|
| 52 |
"train_res": "",
|
|
@@ -55,7 +58,7 @@
|
|
| 55 |
"val_data": "data/snli_1.0/snli.val.hdf5",
|
| 56 |
"val_res": "",
|
| 57 |
"vocab_size": 50265,
|
| 58 |
-
"warmup": "
|
| 59 |
"warmup_perc": 0.1,
|
| 60 |
"weight_decay": 0.01
|
| 61 |
}
|
|
|
|
| 1 |
{
|
| 2 |
+
"acc_batch_size": -1,
|
| 3 |
"adam_betas": "0.9,0.999",
|
| 4 |
"architectures": [
|
| 5 |
"TransformerForNLI"
|
| 6 |
],
|
| 7 |
"attention_probs_dropout_prob": 0.1,
|
| 8 |
+
"bias_proj": "",
|
| 9 |
+
"bias_v1": "",
|
| 10 |
+
"bias_v2": "",
|
| 11 |
"bos_token_id": 0,
|
| 12 |
"clip": 1.0,
|
| 13 |
"cls": "linear",
|
|
|
|
| 17 |
"eos_token_id": 2,
|
| 18 |
"epochs": 5,
|
| 19 |
"freeze_transformer": 0,
|
| 20 |
+
"gpuid": 1,
|
| 21 |
"gradient_checkpointing": false,
|
| 22 |
"hidden_act": "gelu",
|
| 23 |
"hidden_dropout_prob": 0.1,
|
|
|
|
| 49 |
"param_init_type": "xavier_normal",
|
| 50 |
"percent": 1.0,
|
| 51 |
"print_every": 500,
|
| 52 |
+
"save_file": "./models/robertabase_snli_seed1",
|
| 53 |
"seed": 1,
|
| 54 |
"train_data": "data/snli_1.0/snli.train.hdf5",
|
| 55 |
"train_res": "",
|
|
|
|
| 58 |
"val_data": "data/snli_1.0/snli.val.hdf5",
|
| 59 |
"val_res": "",
|
| 60 |
"vocab_size": 50265,
|
| 61 |
+
"warmup": "linear",
|
| 62 |
"warmup_perc": 0.1,
|
| 63 |
"weight_decay": 0.01
|
| 64 |
}
|