BiGS_512 / config.json
JunxiongWang's picture
Bidirectional Gated State Space Model. 346M parameters, 23 layers, max sentence length 512
1dd00da
raw
history blame contribute delete
No virus
581 Bytes
{
"architectures": [
"BiGSForMaskedLM"
],
"classifier_dropout": null,
"decode": false,
"hidden_act": "gelu",
"hidden_dropout_prob": 0.1,
"hidden_size": 1024,
"initializer_range": 0.02,
"intermediate_size": 3072,
"layer_norm_eps": 1e-12,
"max_position_embeddings": 512,
"model_type": "BiGS",
"num_hidden_layers": 23,
"num_ssm": 64,
"pad_token_id": 0,
"position_embedding_type": "absolute",
"pre_norm": true,
"transformers_version": "4.19.0.dev0",
"type_vocab_size": 2,
"use_auth_token": null,
"use_cache": true,
"vocab_size": 30522
}