Newer
Older
{
"cells": [
{
"cell_type": "markdown",
"source": [
"# Run experiment\n"
],
"metadata": {
"collapsed": false
}
},
{
"cell_type": "markdown",
"source": [
"To run any experiment steps in `import_any_dataset.ipynb` and `use_any_asr.ipynb` must be done.\n",
"Experiment is managed by `ExperimentRepository`. Below is example of simple experiment."
],
"metadata": {
"collapsed": false
}
},
{
"cell_type": "code",
"execution_count": null,
"outputs": [],
"source": [
"# imports\n",
"from sziszapangma.integration.task.flair_upos_multi_transformers_wer_processor_base import \\\n",
"from sziszapangma.integration.task.wikineural_multilingual_ner_transformers_wer_processor_base import \\\n",
" WikineuralMultilingualNerTransformersWerProcessorBase\n",
"from sziszapangma.integration.task.spacy_pos_sentence_dep_tag_processor import SpacyDepTagSentenceWerProcessor\n",
"from sziszapangma.integration.task.spacy_ner_sentence_wer_processor import SpacyNerSentenceWerProcessor\n",
"from sziszapangma.integration.task.spacy_pos_sentence_wer_processor import SpacyPosSentenceWerProcessor\n",
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
"from sziszapangma.core.transformer.fasttext_embedding_transformer import FasttextEmbeddingTransformer\n",
"from sziszapangma.integration.task.embedding_wer_metrics_task import EmbeddingWerMetricsTask\n",
"from sziszapangma.integration.task.classic_wer_metric_task import ClassicWerMetricTask\n",
"from experiment.hf_asr.wav2vec2_hf import Wav2Vec2AsrProcessor\n",
"from experiment.utils.property_helper import PropertyHelper\n",
"from sziszapangma.integration.task.asr_task import AsrTask\n",
"from sziszapangma.integration.experiment_manager import ExperimentManager\n",
"from sziszapangma.integration.audio_repository.local_audio_record_repository import LocalAudioRecordRepository\n",
"from experiment.utils.loaded_dataset_helper import LoadedDatasetHelper\n",
"from pathlib import Path\n",
"from sziszapangma.integration.repository.multi_files_experiment_repository import MultiFilesExperimentRepository"
],
"metadata": {
"collapsed": false
}
},
{
"cell_type": "code",
"execution_count": null,
"outputs": [],
"source": [
"# globals\n",
"DATASET_NAME = 'dataset_name'\n",
"ASR_NAME = 'asr_name'\n",
"REPOSITORY_ROOT_PATH = Path.home() / 'asr-benchmarks-repository'\n",
"AUDIO_ROOT_PATH = Path.home() / '.cache/asr-benchmarks'\n",
"FASTTEXT_LANGUAGE_CODE = 'pl'\n",
"WIKINEURAL = \"wikineural\"\n",
"SPACY_MODEL_NAME = 'pl_core_news_lg'\n",
"FLAIR_UPOS_MULTI = 'flair_upos_multi'"
],
"metadata": {
"collapsed": false
}
},
{
"cell_type": "code",
"execution_count": null,
"outputs": [],
"source": [
"# data providers\n",
"repository = MultiFilesExperimentRepository(REPOSITORY_ROOT_PATH, DATASET_NAME)\n",
"record_provider = LoadedDatasetHelper(\n",
" repository, LocalAudioRecordRepository(AUDIO_ROOT_PATH, DATASET_NAME), DATASET_NAME\n",
")"
],
"metadata": {
"collapsed": false
}
},
{
"cell_type": "code",
"execution_count": null,
"outputs": [],
"source": [
"# experiment definition\n",
"experiment_processor = ExperimentManager(\n",
" record_id_iterator=record_provider,\n",
" processing_tasks=[\n",
" AsrTask(\n",
" asr_property_name=PropertyHelper.asr_result(ASR_NAME),\n",
" task_name=f\"AsrTask___{DATASET_NAME}___{ASR_NAME}\",\n",
" require_update=False,\n",
" asr_processor=Wav2Vec2AsrProcessor(\"facebook/wav2vec2-large-xlsr-53-polish\"),\n",
" record_path_provider=record_provider,\n",
" ),\n",
" ClassicWerMetricTask(\n",
" task_name=f\"ClassicWerMetricTask___{DATASET_NAME}___{ASR_NAME}\",\n",
" asr_property_name=PropertyHelper.asr_result(ASR_NAME),\n",
" gold_transcript_property_name=PropertyHelper.get_gold_transcript_words(),\n",
" metrics_property_name=PropertyHelper.word_wer_classic_metrics(ASR_NAME),\n",
" require_update=False,\n",
" alignment_property_name=PropertyHelper.word_wer_classic_alignment(ASR_NAME),\n",
" ),\n",
" EmbeddingWerMetricsTask(\n",
" task_name=f\"EmbeddingWerMetricsTask___{DATASET_NAME}___{ASR_NAME}\",\n",
" asr_property_name=PropertyHelper.asr_result(ASR_NAME),\n",
" gold_transcript_property_name=PropertyHelper.get_gold_transcript_words(),\n",
" require_update=False,\n",
" embedding_transformer=FasttextEmbeddingTransformer(FASTTEXT_LANGUAGE_CODE),\n",
" embeddings_alignment_property_name=PropertyHelper.word_wer_embeddings_alignment(ASR_NAME),\n",
" embeddings_metrics_property_name=PropertyHelper.word_wer_embeddings_metrics(ASR_NAME),\n",
" soft_alignment_property_name=PropertyHelper.word_wer_soft_alignment(ASR_NAME),\n",
" soft_metrics_property_name=PropertyHelper.word_wer_soft_metrics(ASR_NAME),\n",
" ),\n",
" SpacyPosSentenceWerProcessor(\n",
" model_name=SPACY_MODEL_NAME,\n",
" gold_transcript_property_name=PropertyHelper.get_gold_transcript_raw(),\n",
" asr_property_name=PropertyHelper.asr_result(ASR_NAME),\n",
" alignment_property_name=PropertyHelper.pos_alignment(ASR_NAME, SPACY_MODEL_NAME),\n",
" wer_property_name=PropertyHelper.pos_metrics(ASR_NAME, SPACY_MODEL_NAME),\n",
" task_name=f\"SpacyPosSentenceWerProcessor___{DATASET_NAME}___{ASR_NAME}\",\n",
" require_update=False,\n",
" ),\n",
" SpacyNerSentenceWerProcessor(\n",
" model_name=SPACY_MODEL_NAME,\n",
" gold_transcript_property_name=PropertyHelper.get_gold_transcript_raw(),\n",
" asr_property_name=PropertyHelper.asr_result(ASR_NAME),\n",
" alignment_property_name=PropertyHelper.ner_alignment(\n",
" ASR_NAME, SPACY_MODEL_NAME\n",
" ),\n",
" wer_property_name=PropertyHelper.ner_metrics(ASR_NAME, SPACY_MODEL_NAME),\n",
" task_name=f\"SpacyNerSentenceWerProcessor___{DATASET_NAME}___{ASR_NAME}\",\n",
" require_update=False,\n",
" ),\n",
" SpacyDepTagSentenceWerProcessor(\n",
" model_name=SPACY_MODEL_NAME,\n",
" gold_transcript_property_name=PropertyHelper.get_gold_transcript_raw(),\n",
" asr_property_name=PropertyHelper.asr_result(ASR_NAME),\n",
" alignment_property_name=PropertyHelper.dep_tag_alignment(\n",
" ASR_NAME, SPACY_MODEL_NAME\n",
" ),\n",
" wer_property_name=PropertyHelper.dep_tag_metrics(ASR_NAME, SPACY_MODEL_NAME),\n",
" task_name=f\"SpacyDepTagSentenceWerProcessor___{DATASET_NAME}___{ASR_NAME}\",\n",
" require_update=False,\n",
" ),\n",
" WikineuralMultilingualNerTransformersWerProcessorBase(\n",
" gold_transcript_property_name=PropertyHelper.get_gold_transcript_raw(),\n",
" asr_property_name=PropertyHelper.asr_result(ASR_NAME),\n",
" alignment_property_name=PropertyHelper.ner_alignment(\n",
" ASR_NAME, WIKINEURAL\n",
" ),\n",
" wer_property_name=PropertyHelper.ner_metrics(ASR_NAME, WIKINEURAL),\n",
" task_name=f\"WikineuralMultilingualNerTransformersWerProcessorBase___{DATASET_NAME}___{ASR_NAME}\",\n",
" require_update=False,\n",
" ),\n",
" FlairUposMultiTransformersWerProcessorBase(\n",
" gold_transcript_property_name=PropertyHelper.get_gold_transcript_raw(),\n",
" asr_property_name=PropertyHelper.asr_result(ASR_NAME),\n",
" alignment_property_name=PropertyHelper.pos_alignment(\n",
" ASR_NAME, FLAIR_UPOS_MULTI\n",
" ),\n",
" wer_property_name=PropertyHelper.pos_metrics(\n",
" ASR_NAME, FLAIR_UPOS_MULTI\n",
" ),\n",
" task_name=f\"FlairUposMultiTransformersWerProcessorBase___{DATASET_NAME}___{ASR_NAME}\",\n",
" require_update=False,\n",
" )\n",
" ],\n",
" experiment_repository=repository,\n",
")"
],
"metadata": {
"collapsed": false
}
},
{
"cell_type": "code",
"execution_count": null,
"outputs": [],
"source": [
"# run experiment\n",
"experiment_processor.process()\n"
],
"metadata": {
"collapsed": false
}
}
],
"metadata": {
"kernelspec": {
"display_name": "Python 3",
"language": "python",
"name": "python3"
},
"language_info": {
"codemirror_mode": {
"name": "ipython",
"version": 2
},
"file_extension": ".py",
"mimetype": "text/x-python",
"name": "python",
"nbconvert_exporter": "python",
"pygments_lexer": "ipython2",
"version": "2.7.6"
}
},
"nbformat": 4,
"nbformat_minor": 0
}