2019-12-18 22:39:18 +00:00
|
|
|
{
|
2020-01-31 23:00:32 +00:00
|
|
|
"architectures": [
|
|
|
|
"XLMRobertaForMaskedLM"
|
|
|
|
],
|
2019-12-18 22:39:18 +00:00
|
|
|
"attention_probs_dropout_prob": 0.1,
|
2020-04-24 16:07:25 +00:00
|
|
|
"bos_token_id": 0,
|
|
|
|
"eos_token_id": 2,
|
2019-12-18 22:39:18 +00:00
|
|
|
"hidden_act": "gelu",
|
|
|
|
"hidden_dropout_prob": 0.1,
|
|
|
|
"hidden_size": 1024,
|
|
|
|
"initializer_range": 0.02,
|
|
|
|
"intermediate_size": 4096,
|
|
|
|
"layer_norm_eps": 1e-05,
|
|
|
|
"max_position_embeddings": 514,
|
2020-04-24 16:07:25 +00:00
|
|
|
"model_type": "xlm-roberta",
|
2019-12-18 22:39:18 +00:00
|
|
|
"num_attention_heads": 16,
|
|
|
|
"num_hidden_layers": 24,
|
|
|
|
"output_past": true,
|
2020-04-24 16:07:25 +00:00
|
|
|
"pad_token_id": 1,
|
2022-03-04 13:45:43 +00:00
|
|
|
"position_embedding_type": "absolute",
|
|
|
|
"transformers_version": "4.17.0.dev0",
|
2019-12-18 22:39:18 +00:00
|
|
|
"type_vocab_size": 1,
|
2022-03-04 13:45:43 +00:00
|
|
|
"use_cache": true,
|
2019-12-18 22:39:18 +00:00
|
|
|
"vocab_size": 250002
|
|
|
|
}
|