Upload config.json with huggingface_hub
Browse files- config.json +47 -0
config.json
ADDED
|
@@ -0,0 +1,47 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"adaptive": true,
|
| 3 |
+
"architectures": [
|
| 4 |
+
"LSGRobertaForMaskedLM"
|
| 5 |
+
],
|
| 6 |
+
"attention_probs_dropout_prob": 0.1,
|
| 7 |
+
"auto_map": {
|
| 8 |
+
"AutoConfig": "modeling_lsg_roberta.LSGRobertaConfig",
|
| 9 |
+
"AutoModel": "modeling_lsg_roberta.LSGRobertaModel",
|
| 10 |
+
"AutoModelForCausalLM": "modeling_lsg_roberta.LSGRobertaForCausalLM",
|
| 11 |
+
"AutoModelForMaskedLM": "modeling_lsg_roberta.LSGRobertaForMaskedLM",
|
| 12 |
+
"AutoModelForMultipleChoice": "modeling_lsg_roberta.LSGRobertaForMultipleChoice",
|
| 13 |
+
"AutoModelForQuestionAnswering": "modeling_lsg_roberta.LSGRobertaForQuestionAnswering",
|
| 14 |
+
"AutoModelForSequenceClassification": "modeling_lsg_roberta.LSGRobertaForSequenceClassification",
|
| 15 |
+
"AutoModelForTokenClassification": "modeling_lsg_roberta.LSGRobertaForTokenClassification"
|
| 16 |
+
},
|
| 17 |
+
"base_model_prefix": "lsg",
|
| 18 |
+
"block_size": 128,
|
| 19 |
+
"bos_token_id": 0,
|
| 20 |
+
"classifier_dropout": null,
|
| 21 |
+
"eos_token_id": 2,
|
| 22 |
+
"hidden_act": "gelu",
|
| 23 |
+
"hidden_dropout_prob": 0.1,
|
| 24 |
+
"hidden_size": 768,
|
| 25 |
+
"initializer_range": 0.02,
|
| 26 |
+
"intermediate_size": 3072,
|
| 27 |
+
"layer_norm_eps": 1e-05,
|
| 28 |
+
"lsh_num_pre_rounds": 1,
|
| 29 |
+
"mask_first_token": false,
|
| 30 |
+
"max_position_embeddings": 4098,
|
| 31 |
+
"model_type": "roberta",
|
| 32 |
+
"num_attention_heads": 12,
|
| 33 |
+
"num_global_tokens": 1,
|
| 34 |
+
"num_hidden_layers": 6,
|
| 35 |
+
"pad_token_id": 1,
|
| 36 |
+
"pool_with_global": true,
|
| 37 |
+
"position_embedding_type": "absolute",
|
| 38 |
+
"sparse_block_size": 128,
|
| 39 |
+
"sparsity_factor": 2,
|
| 40 |
+
"sparsity_type": "norm",
|
| 41 |
+
"torch_dtype": "float32",
|
| 42 |
+
"transformers_version": "4.30.2",
|
| 43 |
+
"type_vocab_size": 1,
|
| 44 |
+
"use_cache": true,
|
| 45 |
+
"vocab_size": 50265
|
| 46 |
+
}
|
| 47 |
+
|