uw-madison-yoso-4096 / config.json
daze-unlv's picture
Training in progress, epoch 1
d2c2086 verified
{
"_name_or_path": "uw-madison/yoso-4096",
"architectures": [
"YosoForMultipleChoice"
],
"attention_probs_dropout_prob": 0.1,
"bos_token_id": 0,
"conv_window": null,
"eos_token_id": 2,
"hash_code_len": 9,
"hidden_act": "gelu",
"hidden_dropout_prob": 0.1,
"hidden_size": 768,
"initializer_range": 0.02,
"intermediate_size": 3072,
"layer_norm_eps": 1e-12,
"lsh_backward": true,
"max_position_embeddings": 4096,
"model_type": "yoso",
"num_attention_heads": 12,
"num_hash": 64,
"num_hidden_layers": 12,
"pad_token_id": 1,
"position_embedding_type": "absolute",
"tokenizer_class": "AlbertTokenizer",
"torch_dtype": "float32",
"transformers_version": "4.39.0.dev0",
"type_vocab_size": 1,
"use_expectation": true,
"use_fast_hash": true,
"vocab_size": 50265
}