Create config.json
Browse files- config.json +37 -0
config.json
ADDED
|
@@ -0,0 +1,37 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"architectures": ["RobertaForSentencePrediction"],
|
| 3 |
+
"model_type": "roberta",
|
| 4 |
+
"num_classes": 2,
|
| 5 |
+
"num_attention_heads": 16,
|
| 6 |
+
"num_hidden_layers": 24,
|
| 7 |
+
"hidden_size": 1024,
|
| 8 |
+
"intermediate_size": 4096,
|
| 9 |
+
"hidden_act": "gelu",
|
| 10 |
+
"hidden_dropout_prob": 0.1,
|
| 11 |
+
"attention_probs_dropout_prob": 0.1,
|
| 12 |
+
"max_position_embeddings": 514,
|
| 13 |
+
"type_vocab_size": 1,
|
| 14 |
+
"initializer_range": 0.02,
|
| 15 |
+
"layer_norm_eps": 1e-05,
|
| 16 |
+
"classifier_dropout": null,
|
| 17 |
+
"task_specific_params": {
|
| 18 |
+
"sentence_prediction": {
|
| 19 |
+
"dropout": 0.1,
|
| 20 |
+
"classifier_dropout": 0.1
|
| 21 |
+
}
|
| 22 |
+
},
|
| 23 |
+
"lr": 5e-05,
|
| 24 |
+
"total_num_updates": 7812,
|
| 25 |
+
"warmup_updates": 469,
|
| 26 |
+
"classification_head_name": "imdb_head",
|
| 27 |
+
"optimizer": "adam",
|
| 28 |
+
"adam_betas": [0.9, 0.98],
|
| 29 |
+
"adam_eps": 1e-06,
|
| 30 |
+
"weight_decay": 0.1,
|
| 31 |
+
"learning_rate_scheduler": "polynomial_decay",
|
| 32 |
+
"max_sentences": 32,
|
| 33 |
+
"max_tokens": 4400,
|
| 34 |
+
"batch_size": 32,
|
| 35 |
+
"init_token": 0,
|
| 36 |
+
"separator_token": 2
|
| 37 |
+
}
|