Update README.md
Browse files
README.md
CHANGED
|
@@ -77,8 +77,8 @@ Than you need to initialize a model and a pipeline:
|
|
| 77 |
from gliclass import GLiClassModel, ZeroShotClassificationPipeline
|
| 78 |
from transformers import AutoTokenizer
|
| 79 |
|
| 80 |
-
model = GLiClassModel.from_pretrained("knowledgator/gliclass-
|
| 81 |
-
tokenizer = AutoTokenizer.from_pretrained("knowledgator/gliclass-
|
| 82 |
pipeline = ZeroShotClassificationPipeline(model, tokenizer, classification_type='multi-label', device='cuda:0')
|
| 83 |
|
| 84 |
text = "One day I will see the world!"
|
|
@@ -88,6 +88,44 @@ for result in results:
|
|
| 88 |
print(result["label"], "=>", result["score"])
|
| 89 |
```
|
| 90 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 91 |
If you want to use it for NLI type of tasks, we recommend representing your premise as a text and hypothesis as a label, you can put several hypotheses, but the model works best with a single input hypothesis.
|
| 92 |
```python
|
| 93 |
# Initialize model and multi-label pipeline
|
|
@@ -98,40 +136,66 @@ print(results)
|
|
| 98 |
```
|
| 99 |
|
| 100 |
### Benchmarks:
|
| 101 |
-
Below, you can
|
| 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 |
| Model | Num Examples | sst5 | ag_news | emotion | **AVERAGE:** |
|
| 134 |
|------------------------------------|------------------|--------|---------|--------------|----------|
|
|
|
|
|
|
|
|
|
|
| 135 |
| gliclass-modern-large-v2.0-init | 0 | 0.1123 | 0.6933 | 0.3746 | 0.3934 |
|
| 136 |
| gliclass-modern-large-v2.0-init | 8 | 0.5098 | 0.8339 | 0.5010 | 0.6149 |
|
| 137 |
| gliclass-modern-large-v2.0-init | Weak Supervision | 0.0951 | 0.6478 | 0.4520 | 0.3983 |
|
|
|
|
| 77 |
from gliclass import GLiClassModel, ZeroShotClassificationPipeline
|
| 78 |
from transformers import AutoTokenizer
|
| 79 |
|
| 80 |
+
model = GLiClassModel.from_pretrained("knowledgator/gliclass-base-v2.0-rac-init")
|
| 81 |
+
tokenizer = AutoTokenizer.from_pretrained("knowledgator/gliclass-base-v2.0-rac-init")
|
| 82 |
pipeline = ZeroShotClassificationPipeline(model, tokenizer, classification_type='multi-label', device='cuda:0')
|
| 83 |
|
| 84 |
text = "One day I will see the world!"
|
|
|
|
| 88 |
print(result["label"], "=>", result["score"])
|
| 89 |
```
|
| 90 |
|
| 91 |
+
To use with one **RAC** example:
|
| 92 |
+
```python
|
| 93 |
+
example_1 = {
|
| 94 |
+
"text": "A recently developed machine learning platform offers robust automation for complex data analysis workflows. While it enhances productivity, users have reported difficulties in integrating it with their current data infrastructure and a need for better documentation.",
|
| 95 |
+
"all_labels": ["AI", "automation", "data_analysis", "usability", "integration"],
|
| 96 |
+
"true_labels": ["AI", "integration", 'automation']
|
| 97 |
+
}
|
| 98 |
+
|
| 99 |
+
text = "The new AI-powered tool streamlines data analysis by automating repetitive tasks, improving efficiency for data scientists. However, its steep learning curve and limited integration with existing platforms pose challenges for widespread adoption."
|
| 100 |
+
labels = ["AI", "automation", "data_analysis", "usability", "integration"]
|
| 101 |
+
|
| 102 |
+
results = pipeline(text, labels, threshold=0.1, rac_examples=[example_1])[0]
|
| 103 |
+
|
| 104 |
+
for predict in results:
|
| 105 |
+
print(predict["label"], " - ", predict["score"])
|
| 106 |
+
```
|
| 107 |
+
|
| 108 |
+
To use with several **RAC** examples:
|
| 109 |
+
```python
|
| 110 |
+
example_1 = {
|
| 111 |
+
"text": "A recently developed machine learning platform offers robust automation for complex data analysis workflows. While it enhances productivity, users have reported difficulties in integrating it with their current data infrastructure and a need for better documentation.",
|
| 112 |
+
"all_labels": ["AI", "automation", "data_analysis", "usability", "integration"],
|
| 113 |
+
"true_labels": ["AI", "integration", 'automation']
|
| 114 |
+
}
|
| 115 |
+
example_2 = {
|
| 116 |
+
"text": "A cloud-based analytics tool leverages artificial intelligence to provide real-time insights. It significantly improves workflow efficiency but struggles with compatibility across different enterprise systems, requiring additional customization efforts.",
|
| 117 |
+
"all_labels": ["AI", "automation", "data_analysis", "usability", "integration"],
|
| 118 |
+
"true_labels": ["AI", "integration", "data_analysis"]
|
| 119 |
+
}
|
| 120 |
+
text = "The new AI-powered tool streamlines data analysis by automating repetitive tasks, improving efficiency for data scientists. However, its steep learning curve and limited integration with existing platforms pose challenges for widespread adoption."
|
| 121 |
+
labels = ["AI", "automation", "data_analysis", "usability", "integration"]
|
| 122 |
+
|
| 123 |
+
results = pipeline(text, labels, threshold=0.1, rac_examples=[example_1, example_2])[0]
|
| 124 |
+
|
| 125 |
+
for predict in results:
|
| 126 |
+
print(predict["label"], " - ", predict["score"])
|
| 127 |
+
```
|
| 128 |
+
|
| 129 |
If you want to use it for NLI type of tasks, we recommend representing your premise as a text and hypothesis as a label, you can put several hypotheses, but the model works best with a single input hypothesis.
|
| 130 |
```python
|
| 131 |
# Initialize model and multi-label pipeline
|
|
|
|
| 136 |
```
|
| 137 |
|
| 138 |
### Benchmarks:
|
| 139 |
+
Below, you can find a comparison with other GLiClass models:
|
| 140 |
+
|
| 141 |
+
| Dataset | gliclass-base-v1.0-init | gliclass-large-v1.0-init | gliclass-modern-base-v2.0-init | gliclass-modern-large-v2.0-init | gliclass-base-v2.0-rac-init |
|
| 142 |
+
|----------------------|-----------------------|-----------------------|---------------------|---------------------|---------------------|
|
| 143 |
+
| CR | 0.8672 | 0.8024 | 0.9041 | 0.8980 | 0.7852 |
|
| 144 |
+
| sst2 | 0.8342 | 0.8734 | 0.9011 | 0.9434 | 0.8610 |
|
| 145 |
+
| sst5 | 0.2048 | 0.1638 | 0.1972 | 0.1123 | 0.0598 |
|
| 146 |
+
| 20_news_groups | 0.2317 | 0.4151 | 0.2448 | 0.2792 | 0.4007 |
|
| 147 |
+
| spam | 0.5963 | 0.5407 | 0.5074 | 0.6364 | 0.6739 |
|
| 148 |
+
| financial_phrasebank | 0.3594 | 0.3705 | 0.2537 | 0.2562 | 0.2537 |
|
| 149 |
+
| imdb | 0.8772 | 0.8836 | 0.8255 | 0.9137 | 0.8716 |
|
| 150 |
+
| ag_news | 0.5614 | 0.7069 | 0.6050 | 0.6933 | 0.6759 |
|
| 151 |
+
| emotion | 0.2865 | 0.3840 | 0.2474 | 0.3746 | 0.4160 |
|
| 152 |
+
| cap_sotu | 0.3966 | 0.4353 | 0.2929 | 0.2919 | 0.3871 |
|
| 153 |
+
| rotten_tomatoes | 0.6626 | 0.7933 | 0.6630 | 0.5928 | 0.7739 |
|
| 154 |
+
| **AVERAGE:** | 0.5344 | 0.5790 | 0.5129 | 0.5447 | 0.5598 |
|
| 155 |
+
|
| 156 |
+
Here you can see how the performance of the model grows, providing more **RAC** examples:
|
| 157 |
+
| Dataset | 0 examples | 1 example | 2 examples | 3 examples |
|
| 158 |
+
|-------------------------------------|------------|------------|------------|------------|
|
| 159 |
+
| cap_sotu | 0.3857 | 0.4665 | 0.4935 | 0.4847 |
|
| 160 |
+
| cap_sotu (8 examples) | 0.4938 | 0.5097 | 0.4976 | 0.4894 |
|
| 161 |
+
| cap_sotu (Weak Supervision - 8) | 0.4319 | 0.4764 | 0.4488 | 0.4465 |
|
| 162 |
+
| dair-ai_emotion | 0.4472 | 0.5505 | 0.5619 | 0.5705 |
|
| 163 |
+
| dair-ai_emotion (8 examples) | 0.5088 | 0.5630 | 0.5623 | 0.5740 |
|
| 164 |
+
| dair-ai_emotion (Weak Supervision - 8) | 0.4187 | 0.5479 | 0.5693 | 0.5828 |
|
| 165 |
+
| ag_news | 0.6791 | 0.8507 | 0.8717 | 0.8866 |
|
| 166 |
+
| ag_news (8 examples) | 0.8496 | 0.9002 | 0.9072 | 0.9091 |
|
| 167 |
+
| ag_news (Weak Supervision - 8) | 0.6546 | 0.8623 | 0.8841 | 0.8978 |
|
| 168 |
+
| sst5 | 0.0599 | 0.0675 | 0.1163 | 0.1267 |
|
| 169 |
+
| sst5 (8 examples) | 0.2887 | 0.2690 | 0.2642 | 0.2394 |
|
| 170 |
+
| sst5 (Weak Supervision - 8) | 0.0744 | 0.2780 | 0.2897 | 0.2912 |
|
| 171 |
+
| ScienceQA | 0.1142 | 0.4035 | 0.4534 | 0.4495 |
|
| 172 |
+
| ScienceQA (8 examples) | 0.6493 | 0.6547 | 0.6956 | 0.6770 |
|
| 173 |
+
| ScienceQA (Weak Supervision - 8) | 0.2987 | 0.5919 | 0.5998 | 0.5674 |
|
| 174 |
+
| Malicious_code_classification | 0.3717 | 0.6260 | 0.9672 | 0.9788 |
|
| 175 |
+
| Malicious_code_classification (8 examples) | 0.8444 | 0.9722 | 0.9788 | 0.9772 |
|
| 176 |
+
| Malicious_code_classification (Weak Supervision - 8) | 0.3745 | 0.9216 | 0.9788 | 0.9772 |
|
| 177 |
+
| twitter-financial-news-topic | 0.2594 | 0.6249 | 0.6408 | 0.6427 |
|
| 178 |
+
| twitter-financial-news-topic (8 examples) | 0.6137 | 0.7072 | 0.7099 | 0.6948 |
|
| 179 |
+
| twitter-financial-news-topic (Weak Supervision - 8) | 0.4032 | 0.6651 | 0.6316 | 0.6114 |
|
| 180 |
+
| 20_newsgroups | 0.3211 | 0.1339 | 0.0906 | 0.1005 |
|
| 181 |
+
| 20_newsgroups (8 examples) | 0.0959 | 0.0657 | 0.0440 | 0.0445 |
|
| 182 |
+
| 20_newsgroups (Weak Supervision - 8) | 0.4765 | 0.1035 | 0.0775 | 0.0777 |
|
| 183 |
+
| ChemProt | 0.2024 | 0.1911 | 0.1568 | 0.1329 |
|
| 184 |
+
| ChemProt (8 examples) | 0.2985 | 0.3479 | 0.3636 | 0.3538 |
|
| 185 |
+
| ChemProt (Weak Supervision - 8) | 0.2369 | 0.2067 | 0.1911 | 0.1780 |
|
| 186 |
+
|
| 187 |
+
| **AVERAGE:** | **0 examples** | **1 example** | **2 examples** | **3 examples** |
|
| 188 |
+
|-------------------------------------|---------------|---------------|---------------|---------------|
|
| 189 |
+
| Standard | 0.3090 | 0.4275 | 0.4707 | 0.4718 |
|
| 190 |
+
| 8 examples | 0.4838 | 0.5245 | 0.5288 | 0.5244 |
|
| 191 |
+
| Weak Supervision - 8 | 0.3661 | 0.4862 | 0.4868 | 0.4821 |
|
| 192 |
+
|
| 193 |
+
Here you can see how the performance of the model grows, providing more examples in comparison to other models:
|
| 194 |
| Model | Num Examples | sst5 | ag_news | emotion | **AVERAGE:** |
|
| 195 |
|------------------------------------|------------------|--------|---------|--------------|----------|
|
| 196 |
+
| gliclass-base-v2.0-rac-init | 0 | 0.0599 | 0.6791 | 0.4472 | 0.3934 |
|
| 197 |
+
| gliclass-base-v2.0-rac-init | 8 | 0.2887 | 0.8496 | 0.5088 | 0.6149 |
|
| 198 |
+
| gliclass-base-v2.0-rac-init | Weak Supervision | 0.0744 | 0.6546 | 0.4187 | 0.3983 |
|
| 199 |
| gliclass-modern-large-v2.0-init | 0 | 0.1123 | 0.6933 | 0.3746 | 0.3934 |
|
| 200 |
| gliclass-modern-large-v2.0-init | 8 | 0.5098 | 0.8339 | 0.5010 | 0.6149 |
|
| 201 |
| gliclass-modern-large-v2.0-init | Weak Supervision | 0.0951 | 0.6478 | 0.4520 | 0.3983 |
|