Update config.json
Browse files- config.json +176 -0
config.json
ADDED
@@ -0,0 +1,176 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"architectures": [
|
3 |
+
"RagTokenForGeneration"
|
4 |
+
],
|
5 |
+
"dataset": "wiki_dpr",
|
6 |
+
"dataset_split": "train",
|
7 |
+
"decoder_start_token_id": 2,
|
8 |
+
"doc_sep": " // ",
|
9 |
+
"eos_token_id": 2,
|
10 |
+
"generator": {
|
11 |
+
"_num_labels": 3,
|
12 |
+
"activation_dropout": 0.0,
|
13 |
+
"activation_function": "gelu",
|
14 |
+
"add_bias_logits": false,
|
15 |
+
"add_cross_attention": false,
|
16 |
+
"add_final_layer_norm": false,
|
17 |
+
"architectures": [
|
18 |
+
"BartModel",
|
19 |
+
"BartForMaskedLM",
|
20 |
+
"BartForSequenceClassification"
|
21 |
+
],
|
22 |
+
"attention_dropout": 0.0,
|
23 |
+
"bad_words_ids": null,
|
24 |
+
"bos_token_id": 0,
|
25 |
+
"chunk_size_feed_forward": 0,
|
26 |
+
"classif_dropout": 0.0,
|
27 |
+
"d_model": 1024,
|
28 |
+
"decoder_attention_heads": 16,
|
29 |
+
"decoder_ffn_dim": 4096,
|
30 |
+
"decoder_layerdrop": 0.0,
|
31 |
+
"decoder_layers": 12,
|
32 |
+
"decoder_start_token_id": 2,
|
33 |
+
"do_sample": false,
|
34 |
+
"dropout": 0.1,
|
35 |
+
"early_stopping": false,
|
36 |
+
"encoder_attention_heads": 16,
|
37 |
+
"encoder_ffn_dim": 4096,
|
38 |
+
"encoder_layerdrop": 0.0,
|
39 |
+
"encoder_layers": 12,
|
40 |
+
"eos_token_id": 2,
|
41 |
+
"extra_pos_embeddings": 2,
|
42 |
+
"finetuning_task": null,
|
43 |
+
"force_bos_token_to_be_generated": false,
|
44 |
+
"id2label": {
|
45 |
+
"0": "LABEL_0",
|
46 |
+
"1": "LABEL_1",
|
47 |
+
"2": "LABEL_2"
|
48 |
+
},
|
49 |
+
"init_std": 0.02,
|
50 |
+
"is_decoder": false,
|
51 |
+
"is_encoder_decoder": true,
|
52 |
+
"label2id": {
|
53 |
+
"LABEL_0": 0,
|
54 |
+
"LABEL_1": 1,
|
55 |
+
"LABEL_2": 2
|
56 |
+
},
|
57 |
+
"length_penalty": 1.0,
|
58 |
+
"max_length": 20,
|
59 |
+
"max_position_embeddings": 1024,
|
60 |
+
"min_length": 0,
|
61 |
+
"model_type": "bart",
|
62 |
+
"no_repeat_ngram_size": 0,
|
63 |
+
"normalize_before": false,
|
64 |
+
"normalize_embedding": true,
|
65 |
+
"num_beams": 1,
|
66 |
+
"num_hidden_layers": 12,
|
67 |
+
"num_return_sequences": 1,
|
68 |
+
"output_attentions": false,
|
69 |
+
"output_hidden_states": false,
|
70 |
+
"output_past": false,
|
71 |
+
"pad_token_id": 1,
|
72 |
+
"prefix": " ",
|
73 |
+
"pruned_heads": {},
|
74 |
+
"repetition_penalty": 1.0,
|
75 |
+
"return_dict": false,
|
76 |
+
"scale_embedding": false,
|
77 |
+
"static_position_embeddings": false,
|
78 |
+
"task_specific_params": {
|
79 |
+
"summarization": {
|
80 |
+
"early_stopping": true,
|
81 |
+
"length_penalty": 2.0,
|
82 |
+
"max_length": 142,
|
83 |
+
"min_length": 56,
|
84 |
+
"no_repeat_ngram_size": 3,
|
85 |
+
"num_beams": 4
|
86 |
+
}
|
87 |
+
},
|
88 |
+
"temperature": 1.0,
|
89 |
+
"tie_encoder_decoder": false,
|
90 |
+
"tie_word_embeddings": true,
|
91 |
+
"tokenizer_class": null,
|
92 |
+
"top_k": 50,
|
93 |
+
"top_p": 1.0,
|
94 |
+
"torchscript": false,
|
95 |
+
"use_bfloat16": false,
|
96 |
+
"use_cache": true,
|
97 |
+
"vocab_size": 50265,
|
98 |
+
"xla_device": null
|
99 |
+
},
|
100 |
+
"index_name": "exact",
|
101 |
+
"index_path": null,
|
102 |
+
"is_encoder_decoder": true,
|
103 |
+
"max_combined_length": 300,
|
104 |
+
"model_type": "rag",
|
105 |
+
"n_docs": 5,
|
106 |
+
"pad_token_id": 1,
|
107 |
+
"passages_path": null,
|
108 |
+
"question_encoder": {
|
109 |
+
"add_cross_attention": false,
|
110 |
+
"architectures": [
|
111 |
+
"DPRQuestionEncoder"
|
112 |
+
],
|
113 |
+
"attention_probs_dropout_prob": 0.1,
|
114 |
+
"bad_words_ids": null,
|
115 |
+
"bos_token_id": null,
|
116 |
+
"chunk_size_feed_forward": 0,
|
117 |
+
"decoder_start_token_id": null,
|
118 |
+
"do_sample": false,
|
119 |
+
"early_stopping": false,
|
120 |
+
"eos_token_id": null,
|
121 |
+
"finetuning_task": null,
|
122 |
+
"gradient_checkpointing": false,
|
123 |
+
"hidden_act": "gelu",
|
124 |
+
"hidden_dropout_prob": 0.1,
|
125 |
+
"hidden_size": 768,
|
126 |
+
"id2label": {
|
127 |
+
"0": "LABEL_0",
|
128 |
+
"1": "LABEL_1"
|
129 |
+
},
|
130 |
+
"initializer_range": 0.02,
|
131 |
+
"intermediate_size": 3072,
|
132 |
+
"is_decoder": false,
|
133 |
+
"is_encoder_decoder": false,
|
134 |
+
"label2id": {
|
135 |
+
"LABEL_0": 0,
|
136 |
+
"LABEL_1": 1
|
137 |
+
},
|
138 |
+
"layer_norm_eps": 1e-12,
|
139 |
+
"length_penalty": 1.0,
|
140 |
+
"max_length": 20,
|
141 |
+
"max_position_embeddings": 512,
|
142 |
+
"min_length": 0,
|
143 |
+
"model_type": "dpr",
|
144 |
+
"no_repeat_ngram_size": 0,
|
145 |
+
"num_attention_heads": 12,
|
146 |
+
"num_beams": 1,
|
147 |
+
"num_hidden_layers": 12,
|
148 |
+
"num_return_sequences": 1,
|
149 |
+
"output_attentions": false,
|
150 |
+
"output_hidden_states": false,
|
151 |
+
"pad_token_id": 0,
|
152 |
+
"prefix": null,
|
153 |
+
"projection_dim": 0,
|
154 |
+
"pruned_heads": {},
|
155 |
+
"repetition_penalty": 1.0,
|
156 |
+
"return_dict": false,
|
157 |
+
"task_specific_params": null,
|
158 |
+
"temperature": 1.0,
|
159 |
+
"tie_encoder_decoder": false,
|
160 |
+
"tie_word_embeddings": true,
|
161 |
+
"tokenizer_class": null,
|
162 |
+
"top_k": 50,
|
163 |
+
"top_p": 1.0,
|
164 |
+
"torchscript": false,
|
165 |
+
"type_vocab_size": 2,
|
166 |
+
"use_bfloat16": false,
|
167 |
+
"use_cache": true,
|
168 |
+
"vocab_size": 30522,
|
169 |
+
"xla_device": null
|
170 |
+
},
|
171 |
+
"retrieval_batch_size": 8,
|
172 |
+
"retrieval_vector_size": 768,
|
173 |
+
"title_sep": " / ",
|
174 |
+
"use_dummy_dataset": false,
|
175 |
+
"vocab_size": null
|
176 |
+
}
|