bart-large-mnli/config.json

50 lines
1.1 KiB
JSON
Raw Normal View History

2020-03-02 23:35:06 +00:00
{
2020-04-24 15:58:44 +00:00
"_num_labels": 3,
2020-03-02 23:35:06 +00:00
"activation_dropout": 0.0,
2020-04-24 15:58:44 +00:00
"activation_function": "gelu",
"add_final_layer_norm": false,
2021-06-14 07:34:28 +00:00
"architectures": [
"BartForSequenceClassification"
],
2020-03-02 23:35:06 +00:00
"attention_dropout": 0.0,
2020-04-24 15:58:44 +00:00
"bos_token_id": 0,
2020-03-02 23:35:06 +00:00
"classif_dropout": 0.0,
2021-06-14 07:34:28 +00:00
"classifier_dropout": 0.0,
2020-03-02 23:35:06 +00:00
"d_model": 1024,
"decoder_attention_heads": 16,
"decoder_ffn_dim": 4096,
"decoder_layerdrop": 0.0,
"decoder_layers": 12,
2021-06-14 07:34:28 +00:00
"decoder_start_token_id": 2,
2020-03-02 23:35:06 +00:00
"dropout": 0.1,
"encoder_attention_heads": 16,
"encoder_ffn_dim": 4096,
"encoder_layerdrop": 0.0,
"encoder_layers": 12,
2020-04-24 15:58:44 +00:00
"eos_token_id": 2,
2021-06-14 07:34:28 +00:00
"forced_eos_token_id": 2,
"gradient_checkpointing": false,
2020-03-02 23:35:06 +00:00
"id2label": {
"0": "contradiction",
"1": "neutral",
"2": "entailment"
},
"init_std": 0.02,
2020-04-24 15:58:44 +00:00
"is_encoder_decoder": true,
2020-03-02 23:35:06 +00:00
"label2id": {
"contradiction": 0,
"entailment": 2,
"neutral": 1
},
"max_position_embeddings": 1024,
"model_type": "bart",
2020-04-24 15:58:44 +00:00
"normalize_before": false,
2020-03-02 23:35:06 +00:00
"num_hidden_layers": 12,
"output_past": false,
2020-04-24 15:58:44 +00:00
"pad_token_id": 1,
"scale_embedding": false,
2021-06-14 07:34:28 +00:00
"transformers_version": "4.7.0.dev0",
"use_cache": true,
2020-03-02 23:35:06 +00:00
"vocab_size": 50265
}