usr256864 commited on
Commit
8552779
·
verified ·
1 Parent(s): f71b467

Upload 5 files

Browse files
config_kwargs.json ADDED
@@ -0,0 +1,42 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "monitor_metric": "eval_MRR@NM",
3
+ "trainer_kwargs": {
4
+ "class_name": "IR"
5
+ },
6
+ "trainee_kwargs": {
7
+ "class_name": "DPRBiEncoder",
8
+ "loss": {
9
+ "class_name": "NLLLoss"
10
+ }
11
+ },
12
+ "data_module_kwargs": {
13
+ "class_name": "mono_modal_DataModule",
14
+ "data_processor": {
15
+ "class_name": "infoseek_data_processor",
16
+ "dataset_path": "../../data/infoseek/",
17
+ "kb_path": "../../data/infoseek/passages",
18
+ "entity_kb_path": "../../data/infoseek/kb"
19
+ },
20
+ "use_image": false,
21
+ "relevant_indices_key": "BM25_provenance_indices",
22
+ "irrelevant_indices_key": "BM25_irrelevant_indices",
23
+ "input_key": "question",
24
+ "passage_key": "passage",
25
+ "dataloader_kwargs": {
26
+ "num_workers": 6,
27
+ "prefetch_factor": 2
28
+ }
29
+ },
30
+ "question_model_kwargs": {
31
+ "class_name": "DPRQuestionEncoder",
32
+ "checkpoint_name": "question_model",
33
+ "pretrained_model_name_or_path": "/home/data/meerqat/my_transformers_cache/dpr_question_encoder_triviaqa_without_viquae",
34
+ "inference_path": "saved_models/DPR_infoseek/question_model"
35
+ },
36
+ "context_model_kwargs": {
37
+ "class_name": "DPRContextEncoder",
38
+ "checkpoint_name": "context_model",
39
+ "pretrained_model_name_or_path": "/home/data/meerqat/my_transformers_cache/dpr_context_encoder_triviaqa_without_viquae",
40
+ "inference_path": "saved_models/DPR_infoseek/context_model"
41
+ }
42
+ }
experiment_params.json ADDED
@@ -0,0 +1,151 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "subparser": null,
3
+ "max_seq_length": 256,
4
+ "question_max_seq_length": 256,
5
+ "RC": false,
6
+ "coco": false,
7
+ "flickr": false,
8
+ "aokvqa": false,
9
+ "grad_check": true,
10
+ "MJL": false,
11
+ "debug_run_time": false,
12
+ "only_neighbors": false,
13
+ "full_kb": false,
14
+ "image_eval": false,
15
+ "embedding_column": "my_DPR_few_shot",
16
+ "output_ds": null,
17
+ "output_kb": null,
18
+ "embed": false,
19
+ "data_path": null,
20
+ "train_batch_size": 512,
21
+ "eval_batch_size": 1000,
22
+ "transformer_model_name": "bert-base-uncased",
23
+ "gradient_accumulation_steps": 1,
24
+ "warmup_proportion": 0.1,
25
+ "weight_decay": 0.01,
26
+ "adam_beta1": 0.9,
27
+ "adam_beta2": 0.999,
28
+ "warmup_steps": 4,
29
+ "adam_epsilon": 1e-08,
30
+ "num_train_epochs": 15,
31
+ "learning_rate": 2e-05,
32
+ "dropout": 0.5,
33
+ "bert_hidden_size": 768,
34
+ "MM": null,
35
+ "use_graph_P": null,
36
+ "use_graph_Q": null,
37
+ "syntactic_P": null,
38
+ "syntactic_Q": null,
39
+ "NS": null,
40
+ "IC_Q": false,
41
+ "IC_P": false,
42
+ "gcn_lr": 2e-05,
43
+ "head_lr": 2e-06,
44
+ "func": "cat",
45
+ "graph_pooling": "mean",
46
+ "max_nbr_nodes": 2,
47
+ "num_neib": 1,
48
+ "neib_depth": 1,
49
+ "undirect": false,
50
+ "node_hidden_size": 128,
51
+ "max_num_relations": null,
52
+ "max_num_syntactic_relations": null,
53
+ "map_wikidataRelCode_to_idx": null,
54
+ "graph_layout": "fr",
55
+ "residual": false,
56
+ "num_gcn": null,
57
+ "node_alias_max_len": 32,
58
+ "attention": false,
59
+ "num_paths": 1,
60
+ "kg_embed": false,
61
+ "graph_only": false,
62
+ "freeze": false,
63
+ "sanity_run": false,
64
+ "enhanced_question": false,
65
+ "use_entity_type": false,
66
+ "filter_edges": false,
67
+ "draw": false,
68
+ "layer_norm": false,
69
+ "filters_3": 768,
70
+ "tensor_neurons": 16,
71
+ "output_dir": "saved_models/DPR_infoseek",
72
+ "main_dir": "/home/data/meerqat/ViQuAE/meerqat/GP-VQA",
73
+ "experiment_dir": "None",
74
+ "experiment_name": "DPR_infoseek",
75
+ "search": null,
76
+ "xlnet": false,
77
+ "tune_loss": false,
78
+ "tune_dev": true,
79
+ "tune_valid_loss": false,
80
+ "sanity_val_steps": 0,
81
+ "cpu": false,
82
+ "N": 0,
83
+ "nbr_workers": 1,
84
+ "num_proc": 1,
85
+ "local_cache": "/home/data/meerqat/my_transformers_cache",
86
+ "IMAGE_PATH": null,
87
+ "transformer_path": null,
88
+ "resume_from": null,
89
+ "checkpoint": null,
90
+ "linear_MEP_checkpoint": null,
91
+ "linear_MLM_checkpoint": null,
92
+ "config": "experiments/ir/infoseek/dpr/config.json",
93
+ "context_checkpoint": null,
94
+ "grad_accum": 1,
95
+ "cls_token": "[CLS]",
96
+ "sep_token": "[SEP]",
97
+ "pad_token": "[PAD]",
98
+ "cls_token_at_end": false,
99
+ "mask_padding_with_zero": true,
100
+ "pad_on_left": false,
101
+ "sequence_a_segment_id": 0,
102
+ "pad_token_segment_id": 0,
103
+ "cls_token_segment_id": 0,
104
+ "pad_token_label_id": 0,
105
+ "pad_token_id": 0,
106
+ "past_index": -1,
107
+ "world_size": 1,
108
+ "use_lstm": false,
109
+ "use_entity": false,
110
+ "use_question_graph": false,
111
+ "use_question_objects": false,
112
+ "EFeat": null,
113
+ "NFeat": null,
114
+ "ltn": false,
115
+ "use_kelm": false,
116
+ "unshared": false,
117
+ "mlm": false,
118
+ "test": false,
119
+ "split_dpr": false,
120
+ "rename_model": false,
121
+ "mask_rate": 0.15,
122
+ "pretrained_triviaq": false,
123
+ "error_analysis": false,
124
+ "debug": false,
125
+ "stop_debug": false,
126
+ "entity_linking": false,
127
+ "entity_linking_split": false,
128
+ "build_entity_paths": null,
129
+ "object_detection": null,
130
+ "update_wikidata_ids": null,
131
+ "update_None_ids": null,
132
+ "dependency_parsing": null,
133
+ "image_captionning": null,
134
+ "build_graph": false,
135
+ "from_bert": false,
136
+ "fast": false,
137
+ "get_examples": false,
138
+ "get_statistics": null,
139
+ "word_vocab_size": 0,
140
+ "char_vocab_size": 0,
141
+ "max_word_len": 30,
142
+ "device": null,
143
+ "embedding_matrix": null,
144
+ "w2v_file": "word_vector_200d.vec",
145
+ "word_emb_dim": 200,
146
+ "char_lstm": false,
147
+ "char_cnn": false,
148
+ "seed": 32,
149
+ "tune_batch_size": false,
150
+ "model_names": null
151
+ }
model-epoch=02-eval_MRR@NM=0.2249_sd=32.ckpt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:2d57f39cedbf8c74fb38cf1e399febd48af4a206d6091b2145e1192ca806e080
3
+ size 2613940077
results.txt ADDED
@@ -0,0 +1,2 @@
 
 
 
1
+ -------- dev---------
2
+ [0.22488734126091003]
validation_metrics.txt ADDED
@@ -0,0 +1,15 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {'MRR@NM': 0.10558550525976783, 'hits@1': 0.040221297185754436}
2
+ {'MRR@NM': 0.2161168630014627, 'hits@1': 0.09654178674351585}
3
+ {'MRR@NM': 0.22488734429883234, 'hits@1': 0.10273241541253068}
4
+ {'MRR@NM': 0.22140587399232292, 'hits@1': 0.101024655779699}
5
+ {'MRR@NM': 0.21979681045055613, 'hits@1': 0.10275020457537268}
6
+ {'MRR@NM': 0.2177612861961234, 'hits@1': 0.10168285480485288}
7
+ {'MRR@NM': 0.21682980275282598, 'hits@1': 0.10225210801579677}
8
+ {'MRR@NM': 0.21654957124706536, 'hits@1': 0.10273241541253068}
9
+ {'MRR@NM': 0.20927489700441432, 'hits@1': 0.0989433237271854}
10
+ {'MRR@NM': 0.20752238784589702, 'hits@1': 0.09864090795887145}
11
+ {'MRR@NM': 0.21108217116175781, 'hits@1': 0.10043761340591312}
12
+ {'MRR@NM': 0.20421901194846445, 'hits@1': 0.09764471483971965}
13
+ {'MRR@NM': 0.19860714370345037, 'hits@1': 0.09497634041342014}
14
+ {'MRR@NM': 0.19831797864640366, 'hits@1': 0.09689757000035579}
15
+ {'MRR@NM': 0.1992387042524054, 'hits@1': 0.09648841925498985}