system HF staff commited on
Commit
1965c5e
1 Parent(s): 63bb5e8

Update config.json

Browse files
Files changed (1) hide show
  1. config.json +162 -8
config.json CHANGED
@@ -1,21 +1,175 @@
1
  {
 
 
 
2
  "bos_token_id": 0,
 
 
3
  "decoder_start_token_id": 2,
4
  "doc_sep": " // ",
5
  "eos_token_id": 2,
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
6
  "is_encoder_decoder": true,
7
  "max_combined_length": 300,
8
  "model_type": "rag",
9
  "n_docs": 5,
10
  "pad_token_id": 1,
11
- "pretrained_generator_tokenizer_name_or_path": "facebook/bart-large",
12
- "pretrained_question_encoder_tokenizer_name_or_path": "facebook/dpr-question_encoder-single-nq-base",
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
13
  "retrieval_batch_size": 8,
14
  "retrieval_vector_size": 768,
15
- "retriever_type": "hf_retriever",
16
- "dataset": "wiki_dpr",
17
- "dataset_split": "train",
18
- "index_name": "embeddings",
19
  "title_sep": " / ",
20
- "vocab_size": 50265
21
- }
 
 
1
  {
2
+ "architectures": [
3
+ "RagTokenForGeneration"
4
+ ],
5
  "bos_token_id": 0,
6
+ "dataset": "wiki_dpr",
7
+ "dataset_split": "train",
8
  "decoder_start_token_id": 2,
9
  "doc_sep": " // ",
10
  "eos_token_id": 2,
11
+ "generator": {
12
+ "_num_labels": 3,
13
+ "activation_dropout": 0.0,
14
+ "activation_function": "gelu",
15
+ "add_bias_logits": false,
16
+ "add_cross_attention": false,
17
+ "add_final_layer_norm": false,
18
+ "architectures": [
19
+ "BartForConditionalGeneration"
20
+ ],
21
+ "attention_dropout": 0.0,
22
+ "bad_words_ids": null,
23
+ "bos_token_id": 0,
24
+ "chunk_size_feed_forward": 0,
25
+ "classif_dropout": 0.0,
26
+ "d_model": 1024,
27
+ "decoder_attention_heads": 16,
28
+ "decoder_ffn_dim": 4096,
29
+ "decoder_layerdrop": 0.0,
30
+ "decoder_layers": 12,
31
+ "decoder_start_token_id": 2,
32
+ "do_sample": false,
33
+ "dropout": 0.1,
34
+ "early_stopping": false,
35
+ "encoder_attention_heads": 16,
36
+ "encoder_ffn_dim": 4096,
37
+ "encoder_layerdrop": 0.0,
38
+ "encoder_layers": 12,
39
+ "eos_token_id": 2,
40
+ "extra_pos_embeddings": 2,
41
+ "finetuning_task": null,
42
+ "force_bos_token_to_be_generated": true,
43
+ "id2label": {
44
+ "0": "LABEL_0",
45
+ "1": "LABEL_1",
46
+ "2": "LABEL_2"
47
+ },
48
+ "init_std": 0.02,
49
+ "is_decoder": false,
50
+ "is_encoder_decoder": true,
51
+ "label2id": {
52
+ "LABEL_0": 0,
53
+ "LABEL_1": 1,
54
+ "LABEL_2": 2
55
+ },
56
+ "length_penalty": 1.0,
57
+ "max_length": 20,
58
+ "max_position_embeddings": 1024,
59
+ "min_length": 0,
60
+ "model_type": "bart",
61
+ "no_repeat_ngram_size": 0,
62
+ "normalize_before": false,
63
+ "normalize_embedding": true,
64
+ "num_beams": 1,
65
+ "num_hidden_layers": 12,
66
+ "num_return_sequences": 1,
67
+ "output_attentions": false,
68
+ "output_hidden_states": false,
69
+ "output_past": false,
70
+ "pad_token_id": 1,
71
+ "prefix": null,
72
+ "pruned_heads": {},
73
+ "repetition_penalty": 1.0,
74
+ "return_dict": true,
75
+ "scale_embedding": false,
76
+ "static_position_embeddings": false,
77
+ "task_specific_params": {
78
+ "summarization": {
79
+ "early_stopping": true,
80
+ "length_penalty": 2.0,
81
+ "max_length": 142,
82
+ "min_length": 56,
83
+ "no_repeat_ngram_size": 3,
84
+ "num_beams": 4
85
+ }
86
+ },
87
+ "temperature": 1.0,
88
+ "tie_encoder_decoder": false,
89
+ "tie_word_embeddings": true,
90
+ "tokenizer_class": null,
91
+ "top_k": 50,
92
+ "top_p": 1.0,
93
+ "torchscript": false,
94
+ "use_bfloat16": false,
95
+ "use_cache": true,
96
+ "vocab_size": 50265,
97
+ "xla_device": null
98
+ },
99
+ "index_name": "exact",
100
+ "index_path": null,
101
  "is_encoder_decoder": true,
102
  "max_combined_length": 300,
103
  "model_type": "rag",
104
  "n_docs": 5,
105
  "pad_token_id": 1,
106
+ "passages_path": null,
107
+ "question_encoder": {
108
+ "add_cross_attention": false,
109
+ "architectures": [
110
+ "DPRQuestionEncoder"
111
+ ],
112
+ "attention_probs_dropout_prob": 0.1,
113
+ "bad_words_ids": null,
114
+ "bos_token_id": null,
115
+ "chunk_size_feed_forward": 0,
116
+ "decoder_start_token_id": null,
117
+ "do_sample": false,
118
+ "early_stopping": false,
119
+ "eos_token_id": null,
120
+ "finetuning_task": null,
121
+ "gradient_checkpointing": false,
122
+ "hidden_act": "gelu",
123
+ "hidden_dropout_prob": 0.1,
124
+ "hidden_size": 768,
125
+ "id2label": {
126
+ "0": "LABEL_0",
127
+ "1": "LABEL_1"
128
+ },
129
+ "initializer_range": 0.02,
130
+ "intermediate_size": 3072,
131
+ "is_decoder": false,
132
+ "is_encoder_decoder": false,
133
+ "label2id": {
134
+ "LABEL_0": 0,
135
+ "LABEL_1": 1
136
+ },
137
+ "layer_norm_eps": 1e-12,
138
+ "length_penalty": 1.0,
139
+ "max_length": 20,
140
+ "max_position_embeddings": 512,
141
+ "min_length": 0,
142
+ "model_type": "dpr",
143
+ "no_repeat_ngram_size": 0,
144
+ "num_attention_heads": 12,
145
+ "num_beams": 1,
146
+ "num_hidden_layers": 12,
147
+ "num_return_sequences": 1,
148
+ "output_attentions": false,
149
+ "output_hidden_states": false,
150
+ "pad_token_id": 0,
151
+ "prefix": null,
152
+ "projection_dim": 0,
153
+ "pruned_heads": {},
154
+ "repetition_penalty": 1.0,
155
+ "return_dict": false,
156
+ "task_specific_params": null,
157
+ "temperature": 1.0,
158
+ "tie_encoder_decoder": false,
159
+ "tie_word_embeddings": true,
160
+ "tokenizer_class": null,
161
+ "top_k": 50,
162
+ "top_p": 1.0,
163
+ "torchscript": false,
164
+ "type_vocab_size": 2,
165
+ "use_bfloat16": false,
166
+ "use_cache": true,
167
+ "vocab_size": 30522,
168
+ "xla_device": null
169
+ },
170
  "retrieval_batch_size": 8,
171
  "retrieval_vector_size": 768,
 
 
 
 
172
  "title_sep": " / ",
173
+ "use_dummy_dataset": false,
174
+ "vocab_size": null
175
+ }