43 lines
1.0 KiB
JSON
43 lines
1.0 KiB
JSON
|
{
|
||
|
|
"_name_or_path": "/root/.cache/torch/sentence_transformers/danielsaggau_longformer_simcse_scotus/",
|
||
|
|
"architectures": [
|
||
|
|
"LongformerModel"
|
||
|
|
],
|
||
|
|
"attention_mode": "longformer",
|
||
|
|
"attention_probs_dropout_prob": 0.1,
|
||
|
|
"attention_window": [
|
||
|
|
128,
|
||
|
|
128,
|
||
|
|
128,
|
||
|
|
128,
|
||
|
|
128,
|
||
|
|
128
|
||
|
|
],
|
||
|
|
"bos_token_id": 0,
|
||
|
|
"classifier_dropout": null,
|
||
|
|
"cls_token_id": 101,
|
||
|
|
"eos_token_id": null,
|
||
|
|
"gradient_checkpointing": false,
|
||
|
|
"hidden_act": "gelu",
|
||
|
|
"hidden_dropout_prob": 0.1,
|
||
|
|
"hidden_size": 512,
|
||
|
|
"ignore_attention_mask": false,
|
||
|
|
"initializer_range": 0.02,
|
||
|
|
"intermediate_size": 2048,
|
||
|
|
"layer_norm_eps": 1e-05,
|
||
|
|
"max_position_embeddings": 4098,
|
||
|
|
"model_max_length": 4096,
|
||
|
|
"model_type": "longformer",
|
||
|
|
"num_attention_heads": 8,
|
||
|
|
"num_hidden_layers": 6,
|
||
|
|
"onnx_export": false,
|
||
|
|
"pad_token_id": 0,
|
||
|
|
"position_embedding_type": "absolute",
|
||
|
|
"sep_token_id": 102,
|
||
|
|
"torch_dtype": "float32",
|
||
|
|
"transformers_version": "4.24.0",
|
||
|
|
"type_vocab_size": 2,
|
||
|
|
"use_cache": true,
|
||
|
|
"vocab_size": 30522
|
||
|
|
}
|