benschlagman
commited on
Commit
•
493ac67
1
Parent(s):
eeb3dd5
Upload config.json
Browse files- config.json +71 -0
config.json
ADDED
@@ -0,0 +1,71 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"_name_or_path": "google/tapas-base",
|
3 |
+
"activation": "gelu",
|
4 |
+
"aggregation_labels": null,
|
5 |
+
"aggregation_loss_weight": 1.0,
|
6 |
+
"aggregation_temperature": 1.0,
|
7 |
+
"allow_empty_column_selection": false,
|
8 |
+
"answer_loss_cutoff": null,
|
9 |
+
"answer_loss_importance": 1.0,
|
10 |
+
"architectures": [
|
11 |
+
"DistilBertForSequenceClassification"
|
12 |
+
],
|
13 |
+
"attention_dropout": 0.1,
|
14 |
+
"attention_probs_dropout_prob": 0.0,
|
15 |
+
"average_approximation_function": "ratio",
|
16 |
+
"average_logits_per_cell": false,
|
17 |
+
"cell_selection_preference": null,
|
18 |
+
"dim": 768,
|
19 |
+
"disable_per_token_loss": false,
|
20 |
+
"dropout": 0.1,
|
21 |
+
"gradient_checkpointing": false,
|
22 |
+
"hidden_act": "gelu",
|
23 |
+
"hidden_dim": 3072,
|
24 |
+
"hidden_dropout_prob": 0.07,
|
25 |
+
"huber_loss_delta": null,
|
26 |
+
"init_cell_selection_weights_to_zero": false,
|
27 |
+
"initializer_range": 0.02,
|
28 |
+
"intermediate_size": 3072,
|
29 |
+
"layer_norm_eps": 1e-12,
|
30 |
+
"max_num_columns": 32,
|
31 |
+
"max_num_rows": 64,
|
32 |
+
"max_position_embeddings": 1024,
|
33 |
+
"model_type": "distilbert",
|
34 |
+
"n_heads": 12,
|
35 |
+
"n_layers": 12,
|
36 |
+
"no_aggregation_label_index": null,
|
37 |
+
"num_aggregation_labels": 0,
|
38 |
+
"pad_token_id": 0,
|
39 |
+
"positive_label_weight": 10.0,
|
40 |
+
"qa_dropout": 0.1,
|
41 |
+
"reset_position_index_per_cell": true,
|
42 |
+
"select_one_column": true,
|
43 |
+
"seq_classif_dropout": 0.2,
|
44 |
+
"sinusoidal_pos_embds": false,
|
45 |
+
"softmax_temperature": 1.0,
|
46 |
+
"torch_dtype": "float32",
|
47 |
+
"transformers_version": "4.14.0.dev0",
|
48 |
+
"type_vocab_size": [
|
49 |
+
3,
|
50 |
+
256,
|
51 |
+
256,
|
52 |
+
2,
|
53 |
+
256,
|
54 |
+
256,
|
55 |
+
10
|
56 |
+
],
|
57 |
+
"type_vocab_sizes": [
|
58 |
+
3,
|
59 |
+
256,
|
60 |
+
256,
|
61 |
+
2,
|
62 |
+
256,
|
63 |
+
256,
|
64 |
+
10
|
65 |
+
],
|
66 |
+
"use_answer_as_supervision": null,
|
67 |
+
"use_gumbel_for_aggregation": false,
|
68 |
+
"use_gumbel_for_cells": false,
|
69 |
+
"use_normalized_answer_loss": false,
|
70 |
+
"vocab_size": 30522
|
71 |
+
}
|