File size: 6,324 Bytes
20602ca c51bcf6 271a169 c51bcf6 b576b83 8ac7290 79f06c5 8ac7290 79f06c5 4204e97 79f06c5 c51bcf6 79f06c5 b576b83 79f06c5 c51bcf6 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 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 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 |
---
license: mit
---
# diwank/silicone-merged
> Merged and simplified dialog act datasets from the [silicone collection](https://huggingface.co/datasets/silicone/)
All of the subsets of the original collection have been filtered (for errors and ambiguous classes), merged together and grouped into pairs of dialog turns. It is hypothesized that training dialog act classifier by including the previous utterance can help models pick up additional contextual cues and be better at inference esp if an utterance pair is provided.
## Example training script
```python
from datasets import load_dataset
from simpletransformers.classification import (
ClassificationModel, ClassificationArgs
)
# Get data
silicone_merged = load_dataset("diwank/silicone-merged")
train_df = silicone_merged["train"]
eval_df = silicone_merged["validation"]
model_args = ClassificationArgs(
num_train_epochs=8,
model_type="deberta",
model_name="microsoft/deberta-large",
use_multiprocessing=False,
evaluate_during_training=True,
)
# Create a ClassificationModel
model = ClassificationModel("deberta", "microsoft/deberta-large", args=model_args, num_labels=11) # 11 labels in this dataset
# Train model
model.train_model(train_df, eval_df=eval_df)
```
## Balanced variant of the training set
**Note**: This dataset is highly imbalanced and it is recommended to use a library like [imbalanced-learn](https://imbalanced-learn.org/stable/) before proceeding with training.
Since, balancing can be complicated and resource-intensive, we have shared a balanced variant of the train set that was created via oversampling using the _imbalanced-learn_ library. The balancing used the `SMOTEN` algorithm to deal with categorical data clustering and was resampled on a 16-core, 60GB RAM machine. You can access it using:
```load_dataset("diwank/silicone-merged", "balanced")```
## Feature description
- `text_a`: The utterance prior to the utterance being classified. (Say for dialog with turns 1-2-3, if we are trying to find the dialog act for 2, text_a is 1)
- `text_b`: The utterance to be classified
- `labels`: Dialog act label (as integer between 0-10, as mapped below)
## Labels map
```python
[
(0, 'acknowledge')
(1, 'answer')
(2, 'backchannel')
(3, 'reply_yes')
(4, 'exclaim')
(5, 'say')
(6, 'reply_no')
(7, 'hold')
(8, 'ask')
(9, 'intent')
(10, 'ask_yes_no')
]
```
*****
## Appendix
### How the original datasets were mapped:
```python
mapping = {
"acknowledge": {
"swda": [
"aap_am",
"b",
"bk"
],
"mrda": [],
"oasis": [
"ackn",
"accept",
"complete"
],
"maptask": [
"acknowledge",
"align"
],
"dyda_da": [
"commissive"
]
},
"answer": {
"swda": [
"bf",
],
"mrda": [],
"oasis": [
"answ",
"informCont",
"inform",
"answElab",
"directElab",
"refer"
],
"maptask": [
"reply_w",
"explain"
],
"dyda_da": [
"inform"
]
},
"backchannel": {
"swda": [
"ad",
"bh",
"bd",
"b^m"
],
"mrda": [
"b"
],
"oasis": [
"backch",
"selfTalk",
"init"
],
"maptask": ["ready"],
"dyda_da": []
},
"reply_yes": {
"swda": [
"na",
"aa"
],
"mrda": [],
"oasis": [
"confirm"
],
"maptask": [
"reply_y"
],
"dyda_da": []
},
"exclaim": {
"swda": [
"ft",
"fa",
"fc",
"fp"
],
"mrda": [],
"oasis": [
"appreciate",
"bye",
"exclaim",
"greet",
"thank",
"pardon",
"thank-identitySelf",
"expressRegret"
],
"maptask": [],
"dyda_da": []
},
"say": {
"swda": [
"qh",
"sd"
],
"mrda": ["s"],
"oasis": [
"expressPossibility",
"expressOpinion",
"suggest"
],
"maptask": [],
"dyda_da": []
},
"reply_no": {
"swda": [
"nn",
"ng",
"ar"
],
"mrda": [],
"oasis": [
"refuse",
"negate"
],
"maptask": [
"reply_n"
],
"dyda_da": []
},
"hold": {
"swda": [
"^h",
"t1"
],
"mrda": [
"f"
],
"oasis": [
"hold"
],
"maptask": [],
"dyda_da": []
},
"ask": {
"swda": [
"qw",
"qo",
"qw^d",
"br",
"qrr"
],
"mrda": [
"q"
],
"oasis": [
"reqInfo",
"reqDirect",
"offer"
],
"maptask": [
"query_w"
],
"dyda_da": [
"question"
]
},
"intent": {
"swda": [],
"mrda": [],
"oasis": [
"informIntent",
"informIntent-hold",
"expressWish",
"direct",
"raiseIssue",
"correct"
],
"maptask": [
"instruct",
"clarify"
],
"dyda_da": [
"directive"
]
},
"ask_yes_no": {
"swda": [
"qy^d",
"^g"
],
"mrda": [],
"oasis": [
"reqModal"
],
"maptask": [
"query_yn",
"check"
],
"dyda_da": []
}
}
``` |