SetFit with sentence-transformers/paraphrase-multilingual-mpnet-base-v2
This is a SetFit model that can be used for Text Classification. This SetFit model uses sentence-transformers/paraphrase-multilingual-mpnet-base-v2 as the Sentence Transformer embedding model. A LogisticRegression instance is used for classification.
The model has been trained using an efficient few-shot learning technique that involves:
- Fine-tuning a Sentence Transformer with contrastive learning.
- Training a classification head with features from the fine-tuned Sentence Transformer.
Model Details
Model Description
- Model Type: SetFit
- Sentence Transformer body: sentence-transformers/paraphrase-multilingual-mpnet-base-v2
- Classification head: a LogisticRegression instance
- Maximum Sequence Length: 128 tokens
- Number of Classes: 3 classes
Model Sources
- Repository: SetFit on GitHub
- Paper: Efficient Few-Shot Learning Without Prompts
- Blogpost: SetFit: Efficient Few-Shot Learning Without Prompts
Model Labels
Label | Examples |
---|---|
pos |
|
obj |
|
neg |
|
Evaluation
Metrics
Label | 0 | 1 | 2 | Accuracy | Macro Avg | Weighted Avg |
---|---|---|---|---|---|---|
all | {'precision': 0.6018099547511312, 'recall': 0.5611814345991561, 'f1-score': 0.5807860262008734, 'support': 237} | {'precision': 0.5602409638554217, 'recall': 0.6262626262626263, 'f1-score': 0.591414944356121, 'support': 297} | {'precision': 0.7932692307692307, 'recall': 0.7268722466960352, 'f1-score': 0.7586206896551724, 'support': 227} | 0.6360 | {'precision': 0.6517733831252612, 'recall': 0.6381054358526059, 'f1-score': 0.6436072200707222, 'support': 761} | {'precision': 0.642697294251897, 'recall': 0.6360052562417872, 'f1-score': 0.6379808452498016, 'support': 761} |
Uses
Direct Use for Inference
First install the SetFit library:
pip install setfit
Then you can load this model and run inference.
from setfit import SetFitModel
# Download from the 🤗 Hub
model = SetFitModel.from_pretrained("mogaio/pr_ebsa_fr_tran_merged25_e1_beginning_offsets_10_v3")
# Run inference
preds = model("Adil Hussain
Adil Hussain est reconnaissant d'avoir reçu l'enseignement de l'acteur Naseeruddin Shah à l'époque où il fréquentait l'École nationale d'art dramatique")
Training Details
Training Set Metrics
Training set | Min | Median | Max |
---|---|---|---|
Word count | 1 | 243.9997 | 2071 |
Label | Training Sample Count |
---|---|
neg | 912 |
obj | 1220 |
pos | 908 |
Training Hyperparameters
- batch_size: (8, 8)
- num_epochs: (10, 10)
- max_steps: -1
- sampling_strategy: oversampling
- num_iterations: 1
- body_learning_rate: (2e-05, 1e-05)
- head_learning_rate: 0.01
- loss: CosineSimilarityLoss
- distance_metric: cosine_distance
- margin: 0.25
- end_to_end: False
- use_amp: False
- warmup_proportion: 0.1
- seed: 42
- eval_max_steps: -1
- load_best_model_at_end: True
Training Results
Epoch | Step | Training Loss | Validation Loss |
---|---|---|---|
0.0013 | 1 | 0.3526 | - |
0.0658 | 50 | 0.3825 | - |
0.1316 | 100 | 0.2039 | - |
0.1974 | 150 | 0.2579 | - |
0.2632 | 200 | 0.3062 | - |
0.3289 | 250 | 0.1744 | - |
0.3947 | 300 | 0.1571 | - |
0.4605 | 350 | 0.222 | - |
0.5263 | 400 | 0.2697 | - |
0.5921 | 450 | 0.2507 | - |
0.6579 | 500 | 0.311 | - |
0.7237 | 550 | 0.3169 | - |
0.7895 | 600 | 0.1766 | - |
0.8553 | 650 | 0.1756 | - |
0.9211 | 700 | 0.2497 | - |
0.9868 | 750 | 0.0889 | - |
1.0 | 760 | - | 0.2281 |
1.0526 | 800 | 0.1036 | - |
1.1184 | 850 | 0.3135 | - |
1.1842 | 900 | 0.1744 | - |
1.25 | 950 | 0.3287 | - |
1.3158 | 1000 | 0.1816 | - |
1.3816 | 1050 | 0.231 | - |
1.4474 | 1100 | 0.153 | - |
1.5132 | 1150 | 0.2131 | - |
1.5789 | 1200 | 0.1358 | - |
1.6447 | 1250 | 0.276 | - |
1.7105 | 1300 | 0.2646 | - |
1.7763 | 1350 | 0.0337 | - |
1.8421 | 1400 | 0.158 | - |
1.9079 | 1450 | 0.1123 | - |
1.9737 | 1500 | 0.0889 | - |
2.0 | 1520 | - | 0.2268 |
2.0395 | 1550 | 0.2369 | - |
2.1053 | 1600 | 0.196 | - |
2.1711 | 1650 | 0.2799 | - |
2.2368 | 1700 | 0.073 | - |
2.3026 | 1750 | 0.2392 | - |
2.3684 | 1800 | 0.1551 | - |
2.4342 | 1850 | 0.178 | - |
2.5 | 1900 | 0.1719 | - |
2.5658 | 1950 | 0.1203 | - |
2.6316 | 2000 | 0.1502 | - |
2.6974 | 2050 | 0.0963 | - |
2.7632 | 2100 | 0.1566 | - |
2.8289 | 2150 | 0.1273 | - |
2.8947 | 2200 | 0.264 | - |
2.9605 | 2250 | 0.0736 | - |
3.0 | 2280 | - | 0.2385 |
3.0263 | 2300 | 0.1577 | - |
3.0921 | 2350 | 0.1613 | - |
3.1579 | 2400 | 0.2313 | - |
3.2237 | 2450 | 0.1134 | - |
3.2895 | 2500 | 0.0593 | - |
3.3553 | 2550 | 0.0395 | - |
3.4211 | 2600 | 0.119 | - |
3.4868 | 2650 | 0.0152 | - |
3.5526 | 2700 | 0.019 | - |
3.6184 | 2750 | 0.1056 | - |
3.6842 | 2800 | 0.1355 | - |
3.75 | 2850 | 0.2262 | - |
3.8158 | 2900 | 0.102 | - |
3.8816 | 2950 | 0.2388 | - |
3.9474 | 3000 | 0.1755 | - |
4.0 | 3040 | - | 0.2576 |
4.0132 | 3050 | 0.0957 | - |
4.0789 | 3100 | 0.2034 | - |
4.1447 | 3150 | 0.0076 | - |
4.2105 | 3200 | 0.0431 | - |
4.2763 | 3250 | 0.2228 | - |
4.3421 | 3300 | 0.0124 | - |
4.4079 | 3350 | 0.2316 | - |
4.4737 | 3400 | 0.037 | - |
4.5395 | 3450 | 0.1812 | - |
4.6053 | 3500 | 0.2115 | - |
4.6711 | 3550 | 0.2534 | - |
4.7368 | 3600 | 0.1833 | - |
4.8026 | 3650 | 0.0135 | - |
4.8684 | 3700 | 0.1169 | - |
4.9342 | 3750 | 0.0093 | - |
5.0 | 3800 | 0.0728 | 0.2787 |
5.0658 | 3850 | 0.022 | - |
5.1316 | 3900 | 0.0586 | - |
5.1974 | 3950 | 0.002 | - |
5.2632 | 4000 | 0.1112 | - |
5.3289 | 4050 | 0.0902 | - |
5.3947 | 4100 | 0.0002 | - |
5.4605 | 4150 | 0.108 | - |
5.5263 | 4200 | 0.0158 | - |
5.5921 | 4250 | 0.0725 | - |
5.6579 | 4300 | 0.0014 | - |
5.7237 | 4350 | 0.2018 | - |
5.7895 | 4400 | 0.0023 | - |
5.8553 | 4450 | 0.002 | - |
5.9211 | 4500 | 0.213 | - |
5.9868 | 4550 | 0.0034 | - |
6.0 | 4560 | - | 0.2994 |
6.0526 | 4600 | 0.1203 | - |
6.1184 | 4650 | 0.1615 | - |
6.1842 | 4700 | 0.1968 | - |
6.25 | 4750 | 0.004 | - |
6.3158 | 4800 | 0.0018 | - |
6.3816 | 4850 | 0.048 | - |
6.4474 | 4900 | 0.0748 | - |
6.5132 | 4950 | 0.0007 | - |
6.5789 | 5000 | 0.0019 | - |
6.6447 | 5050 | 0.0015 | - |
6.7105 | 5100 | 0.0075 | - |
6.7763 | 5150 | 0.0012 | - |
6.8421 | 5200 | 0.0016 | - |
6.9079 | 5250 | 0.0009 | - |
6.9737 | 5300 | 0.0004 | - |
7.0 | 5320 | - | 0.2721 |
7.0395 | 5350 | 0.0142 | - |
7.1053 | 5400 | 0.0527 | - |
7.1711 | 5450 | 0.0019 | - |
7.2368 | 5500 | 0.0024 | - |
7.3026 | 5550 | 0.0002 | - |
7.3684 | 5600 | 0.0349 | - |
7.4342 | 5650 | 0.0008 | - |
7.5 | 5700 | 0.0008 | - |
7.5658 | 5750 | 0.0005 | - |
7.6316 | 5800 | 0.0002 | - |
7.6974 | 5850 | 0.0024 | - |
7.7632 | 5900 | 0.0017 | - |
7.8289 | 5950 | 0.0002 | - |
7.8947 | 6000 | 0.0028 | - |
7.9605 | 6050 | 0.0124 | - |
8.0 | 6080 | - | 0.3064 |
8.0263 | 6100 | 0.0038 | - |
8.0921 | 6150 | 0.064 | - |
8.1579 | 6200 | 0.0007 | - |
8.2237 | 6250 | 0.0022 | - |
8.2895 | 6300 | 0.0012 | - |
8.3553 | 6350 | 0.0103 | - |
8.4211 | 6400 | 0.0008 | - |
8.4868 | 6450 | 0.0058 | - |
8.5526 | 6500 | 0.0046 | - |
8.6184 | 6550 | 0.1061 | - |
8.6842 | 6600 | 0.0005 | - |
8.75 | 6650 | 0.1509 | - |
8.8158 | 6700 | 0.0027 | - |
8.8816 | 6750 | 0.0002 | - |
8.9474 | 6800 | 0.0006 | - |
9.0 | 6840 | - | 0.3014 |
9.0132 | 6850 | 0.0018 | - |
9.0789 | 6900 | 0.0006 | - |
9.1447 | 6950 | 0.0003 | - |
9.2105 | 7000 | 0.0007 | - |
9.2763 | 7050 | 0.0006 | - |
9.3421 | 7100 | 0.0003 | - |
9.4079 | 7150 | 0.1859 | - |
9.4737 | 7200 | 0.0009 | - |
9.5395 | 7250 | 0.12 | - |
9.6053 | 7300 | 0.0148 | - |
9.6711 | 7350 | 0.0002 | - |
9.7368 | 7400 | 0.0005 | - |
9.8026 | 7450 | 0.1027 | - |
9.8684 | 7500 | 0.0002 | - |
9.9342 | 7550 | 0.0004 | - |
10.0 | 7600 | 0.0003 | 0.3075 |
- The bold row denotes the saved checkpoint.
Framework Versions
- Python: 3.10.12
- SetFit: 1.0.1
- Sentence Transformers: 2.2.2
- Transformers: 4.35.2
- PyTorch: 2.1.0+cu121
- Datasets: 2.16.1
- Tokenizers: 0.15.0
Citation
BibTeX
@article{https://doi.org/10.48550/arxiv.2209.11055,
doi = {10.48550/ARXIV.2209.11055},
url = {https://arxiv.org/abs/2209.11055},
author = {Tunstall, Lewis and Reimers, Nils and Jo, Unso Eun Seo and Bates, Luke and Korat, Daniel and Wasserblat, Moshe and Pereg, Oren},
keywords = {Computation and Language (cs.CL), FOS: Computer and information sciences, FOS: Computer and information sciences},
title = {Efficient Few-Shot Learning Without Prompts},
publisher = {arXiv},
year = {2022},
copyright = {Creative Commons Attribution 4.0 International}
}
- Downloads last month
- 1
Inference Providers
NEW
This model is not currently available via any of the supported third-party Inference Providers, and
the model is not deployed on the HF Inference API.
Model tree for mogaio/pr_ebsa_fr_tran_merged25_e1_beginning_offsets_10_v3
Evaluation results
- 0 on Unknowntest set self-reported[object Object]
- 1 on Unknowntest set self-reported[object Object]
- 2 on Unknowntest set self-reported[object Object]
- Accuracy on Unknowntest set self-reported0.636
- Macro Avg on Unknowntest set self-reported[object Object]
- Weighted Avg on Unknowntest set self-reported[object Object]