diff --git a/classifier_bare_huggingface.py b/classifier_bare_huggingface.py
index 5b60babafe421c7236832bb2ae021509fd188ec2..4824ebccb8ee91f3a6820f0a4dc74c219bb0e664 100644
--- a/classifier_bare_huggingface.py
+++ b/classifier_bare_huggingface.py
@@ -152,8 +152,11 @@ trainer.train()
print('\nDev results:')
for corpus in encoded_dev_dataset:
print()
- dev_results = get_predictions_huggingface(trainer, corpus,
- encoded_dev_dataset[corpus])
+ dev_results = better_predictions_huggingface(trainer,
+ corpus,
+ encoded_dev_dataset[corpus],
+ framework_labels[corpus.split('.')[1]]
+ )
print(dev_results)
@@ -171,17 +174,18 @@ for corpus in encoded_dev_dataset:
# Test results
-print('\ntest results:')
-for corpus in encoded_test_dataset:
- print()
- test_results = get_predictions_huggingface(trainer,
- corpus,
- encoded_test_dataset[corpus])
+# print('\ntest results:')
+# for corpus in encoded_test_dataset:
+# print()
+# test_results = get_predictions_huggingface(trainer,
+# corpus,
+# framework_labels[corpus.split('.')[1]],
+# encoded_test_dataset[corpus])
- path_results = 'results/test/' + args.transformer_model + '_' + str(args.num_epochs)
- if not os.path.exists(path_results):
- os.makedirs(path_results)
+# path_results = 'results/test/' + args.transformer_model + '_' + str(args.num_epochs)
+# if not os.path.exists(path_results):
+# os.makedirs(path_results)
# print_results_to_file(corpus,
# test_dict_sentences[corpus],
diff --git a/make_mappings_zero-shot.py b/make_mappings_zero-shot.py
index cacad68a8ccd94ee3f8c6622a00259cbd69c3d94..9f57d2ce825d5ccb0605c6b9c45ef6aac9ca2c81 100644
--- a/make_mappings_zero-shot.py
+++ b/make_mappings_zero-shot.py
@@ -72,7 +72,7 @@ for f in rel_files:
dict_labels = dict(enumerate(list(set(good_rels))))
corpora_labels = {v:k for k, v in dict_labels.items()}
-corpora_labels['unk'] = len(corpora_labels)
+
leftovers = []
@@ -90,9 +90,11 @@ for sub in sub_rels:
try:
corpora_labels[sub] = corpora_labels[subs[sub]]
except KeyError:
- corpora_labels[subs[sub]] = max(list(corpora_labels.values())) + 1
+ corpora_labels[subs[sub]] = max(list(corpora_labels.values())) + 1
corpora_labels[sub] = corpora_labels[subs[sub]]
+corpora_labels['unk'] = max(list(corpora_labels.values())) + 1
+
# print(corpora_labels)
with open(args.mappings_file, 'w') as f:
diff --git a/mappings/mappings_final.tsv b/mappings/mappings_final.tsv
new file mode 100644
index 0000000000000000000000000000000000000000..81ff02718d9a452dd2e0d3e0e77a5c84bc7329b7
--- /dev/null
+++ b/mappings/mappings_final.tsv
@@ -0,0 +1,158 @@
+LABEL MAPPING
+result 0
+organization-phatic 1
+expansion.correction 2
+expansion.equivalence 3
+volitional-cause 4
+causal-result 5
+preparation 6
+summary 7
+elaboration-additional 8
+attribution 9
+temporal.asynchronous 10
+reason 11
+enablement 12
+restatement-repetition 13
+nonvolitional-cause-e 14
+contingency.purpose 15
+solutionhood 16
+topic-solutionhood 17
+temporal 18
+expansion.instantiation 19
+progression 20
+purpose-goal 21
+comment 22
+narration 23
+evaluation-comment 24
+conclusion 25
+explanation* 26
+contingency.cause 27
+temporal.synchronous 28
+comparison.concession 29
+contingency.cause+speechact 30
+continuation 31
+organization-heading 32
+acknowledgement 33
+circumstance 34
+nonvolitional-result 35
+contingency.condition+speechact 36
+temploc 37
+elab-addition 38
+elab-enumember 39
+effect 40
+restatement-mn 41
+condition 42
+topic-comment 43
+nonvolitional-cause 44
+joint-list 45
+concession 46
+clarification_question 47
+antithesis 48
+attribution-negative 49
+parenthetical 50
+purpose-attribute 51
+expansion.manner 52
+explanation-motivation 53
+q_elab 54
+context-background 55
+contingency.negative-cause 56
+conjunction 57
+frame 58
+contingency.negative 59
+bg-goal 60
+expansion.substitution 61
+nonvolitional-result-e 62
+contingency.condition 63
+volitional-result 64
+hypophora 65
+comparison.contrast 66
+adversative-antithesis 67
+joint-other 68
+mode-manner 69
+contingency-condition 70
+elab-example 71
+contingency.negative-condition+speechact 72
+joint-sequence 73
+expansion.level-of-detail 74
+explanation-evidence 75
+restatement-partial 76
+elab-aspect 77
+expansion.exception 78
+exp-reason 79
+evaluation-n 80
+topic-question 81
+parallel 82
+contingency.negative-condition 83
+evaluation 84
+attribution-positive 85
+topic-change 86
+bg-compare 87
+joint 88
+expansion.disjunction 89
+interpretation-evaluation 90
+elab-definition 91
+context-circumstance 92
+adversative-contrast 93
+causal-cause 94
+topic-drift 95
+elaboration 96
+expansion.restatement 97
+contingency.goal 98
+manner-means 99
+background 100
+mode-means 101
+comparison.similarity 102
+means 103
+comparison 104
+flashback 105
+interpretation 106
+explanation-justify 107
+organization-preparation 108
+elaboration-attribute 109
+contrast 110
+adversative-concession 111
+comparison.degree 112
+root 113
+e-elaboration 114
+expansion.alternative 115
+evaluation-s 116
+comparison.concession+speechact 117
+contingency.cause+belief 118
+purpose 119
+interrupted 120
+expansion 121
+cause 122
+question_answer_pair 123
+elab-process_step 124
+explanation 125
+cause-effect 126
+unk 127
+list 45
+evidence 75
+sequence 73
+disjunction 89
+justify 107
+motivation 53
+restatement 97
+expansion.genexpansion 121
+expansion.conjunction 57
+joint-disjunction 89
+goal 21
+alternation 115
+conditional 42
+adversative 128
+otherwise 128
+correction 2
+unconditional 89
+unless 110
+bg-general 100
+exp-evidence 75
+organization 129
+textual-organization 129
+alternative 115
+temporal.synchrony 28
+repetition 13
+expansion.level 74
+qap.hypophora 65
+qap 123
+causation 122
diff --git a/utils.py b/utils.py
index 0e13b338804381cadccc829c381745955127c9c0..e3b17afaa7222453307a17f7249bcbdfe62f2228 100644
--- a/utils.py
+++ b/utils.py
@@ -142,11 +142,11 @@ def open_file_with_lang(filename, mappings_dict):
# flip them if different direction
if args.normalize_direction == 'yes':
if l[9] == '1>2':
- lines.append(l + [[lang, fullname, framework] + sent_1 + [SEP_token] + sent_2, encode_label(l[11], mappings_dict)])
+ lines.append(l + [[lang, fullname] + sent_1 + [SEP_token] + sent_2, encode_label(l[11], mappings_dict)])
else:
- lines.append(l + [[lang, fullname, framework] + sent_2 + [SEP_token] + sent_1, encode_label(l[11], mappings_dict)])
+ lines.append(l + [[lang, fullname] + sent_2 + [SEP_token] + sent_1, encode_label(l[11], mappings_dict)])
else:
- lines.append(l + [[lang, fullname, framework] + sent_1 + [SEP_token] + sent_2, encode_label(l[11], mappings_dict)])
+ lines.append(l + [[lang, fullname] + sent_1 + [SEP_token] + sent_2, encode_label(l[11], mappings_dict)])
return lines
@@ -391,6 +391,45 @@ def get_predictions_huggingface(trainer,
return preds
+
+def better_predictions_huggingface(trainer,
+ corpus,
+ test_set,
+ corpus_labels,
+ print_results=True):
+
+ ''' SPECIFI FUNCTION FOR THE HUGGINGFACE TRAINER.
+ Function to get the model's predictions for one corpus' test set.
+ Can print accuracy using scikit-learn.
+ Also works with dev sets -- just don't save the outputs.
+ Returns: list of predictions that match test file's lines.
+ '''
+
+ results = trainer.predict(test_set)
+ preds = np.argmax(results.predictions, axis=1)
+ orig_labels = results.label_ids
+ test_acc = round(accuracy_score(top_preds, orig_labels), 4)
+
+ if print_results:
+ print(corpus + '\t' + str(test_acc) + '\n', flush='True')
+
+ print(type(results.predictions))
+ # try to make the better prediction bit
+ best_labels = []
+ for n, result in enumerate(results.predictions.tolist()):
+ orig_label = results.label_ids[n]
+ best_prob = -1000
+ best_label = -1
+ if orig_label in corpus_labels:
+ if result > best_prob:
+ best_prob = result
+ best_label = n
+ best_labels.append(n)
+ test_acc = round(accuracy_score(best_labels, orig_labels), 4)
+ print('better:\t' + str(test_acc) + '\n', flush='True')
+
+ return best_labels
+
def get_better_predictions(model,
corpus,
@@ -443,7 +482,7 @@ def get_better_predictions(model,
test_acc = round(accuracy_score(all_labels, top_preds), 4)
if print_results:
- print('better:', '\t', test_acc, flush='')
+ print('better:\t' + str(test_acc), flush='True')
return all_labels, all_preds