Upload config.json
Browse files- config.json +21 -0
config.json
ADDED
|
@@ -0,0 +1,21 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{"architectures": [
|
| 2 |
+
"MegatronBertForSequenceClassification"
|
| 3 |
+
],
|
| 4 |
+
"_num_labels": 2,
|
| 5 |
+
"id2label":{"1":"similar","0":"not similar"},
|
| 6 |
+
"vocab_size": 21248,
|
| 7 |
+
"hidden_size": 2048,
|
| 8 |
+
"num_hidden_layers": 24,
|
| 9 |
+
"num_attention_heads": 8,
|
| 10 |
+
"hidden_act": "gelu_new",
|
| 11 |
+
"intermediate_size": 8192,
|
| 12 |
+
"hidden_dropout_prob": 0.1,
|
| 13 |
+
"attention_probs_dropout_prob": 0.1,
|
| 14 |
+
"max_position_embeddings": 512,
|
| 15 |
+
"type_vocab_size": 2,
|
| 16 |
+
"initializer_range": 0.2,
|
| 17 |
+
"layer_norm_eps": 1e-12,
|
| 18 |
+
"gradient_checkpointing": false,
|
| 19 |
+
"position_embedding_type": "absolute",
|
| 20 |
+
"use_cache": false,
|
| 21 |
+
"model_type": "megatron-bert"}
|