SetFit with mini1013/master_domain
This is a SetFit model that can be used for Text Classification. This SetFit model uses mini1013/master_domain 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: mini1013/master_domain
- Classification head: a LogisticRegression instance
- Maximum Sequence Length: 512 tokens
- Number of Classes: 4 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 |
---|---|
3 |
|
0 |
|
2 |
|
1 |
|
Evaluation
Metrics
Label | Accuracy |
---|---|
all | 0.5104 |
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("mini1013/master_cate_bt2_test_flat_top_cate")
# Run inference
preds = model("메디힐 워터마이드 하이드롭 에센셜 마스크 REX 24ml 홈>전체상품;(#M)홈>스킨케어>마스크팩 Naverstore > 화장품/미용 > 마스크/팩 > 마스크시트")
Training Details
Training Set Metrics
Training set | Min | Median | Max |
---|---|---|---|
Word count | 11 | 23.615 | 91 |
Label | Training Sample Count |
---|---|
0 | 50 |
1 | 50 |
2 | 50 |
3 | 50 |
Training Hyperparameters
- batch_size: (64, 64)
- num_epochs: (30, 30)
- max_steps: -1
- sampling_strategy: oversampling
- num_iterations: 100
- 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
- l2_weight: 0.01
- seed: 42
- eval_max_steps: -1
- load_best_model_at_end: False
Training Results
Epoch | Step | Training Loss | Validation Loss |
---|---|---|---|
0.0032 | 1 | 0.4519 | - |
0.1597 | 50 | 0.4406 | - |
0.3195 | 100 | 0.4089 | - |
0.4792 | 150 | 0.3854 | - |
0.6390 | 200 | 0.3414 | - |
0.7987 | 250 | 0.2792 | - |
0.9585 | 300 | 0.2485 | - |
1.1182 | 350 | 0.2268 | - |
1.2780 | 400 | 0.1526 | - |
1.4377 | 450 | 0.1375 | - |
1.5974 | 500 | 0.1239 | - |
1.7572 | 550 | 0.123 | - |
1.9169 | 600 | 0.1002 | - |
2.0767 | 650 | 0.0834 | - |
2.2364 | 700 | 0.0828 | - |
2.3962 | 750 | 0.0698 | - |
2.5559 | 800 | 0.0604 | - |
2.7157 | 850 | 0.0281 | - |
2.8754 | 900 | 0.0148 | - |
3.0351 | 950 | 0.0129 | - |
3.1949 | 1000 | 0.0102 | - |
3.3546 | 1050 | 0.0083 | - |
3.5144 | 1100 | 0.007 | - |
3.6741 | 1150 | 0.0042 | - |
3.8339 | 1200 | 0.0021 | - |
3.9936 | 1250 | 0.0002 | - |
4.1534 | 1300 | 0.0001 | - |
4.3131 | 1350 | 0.0003 | - |
4.4728 | 1400 | 0.0001 | - |
4.6326 | 1450 | 0.0001 | - |
4.7923 | 1500 | 0.0 | - |
4.9521 | 1550 | 0.0 | - |
5.1118 | 1600 | 0.0 | - |
5.2716 | 1650 | 0.0 | - |
5.4313 | 1700 | 0.0 | - |
5.5911 | 1750 | 0.0003 | - |
5.7508 | 1800 | 0.0 | - |
5.9105 | 1850 | 0.0004 | - |
6.0703 | 1900 | 0.0005 | - |
6.2300 | 1950 | 0.0026 | - |
6.3898 | 2000 | 0.0006 | - |
6.5495 | 2050 | 0.0002 | - |
6.7093 | 2100 | 0.0 | - |
6.8690 | 2150 | 0.0002 | - |
7.0288 | 2200 | 0.0002 | - |
7.1885 | 2250 | 0.0005 | - |
7.3482 | 2300 | 0.0006 | - |
7.5080 | 2350 | 0.0002 | - |
7.6677 | 2400 | 0.0002 | - |
7.8275 | 2450 | 0.0001 | - |
7.9872 | 2500 | 0.0014 | - |
8.1470 | 2550 | 0.0001 | - |
8.3067 | 2600 | 0.0 | - |
8.4665 | 2650 | 0.0 | - |
8.6262 | 2700 | 0.0 | - |
8.7859 | 2750 | 0.0 | - |
8.9457 | 2800 | 0.0 | - |
9.1054 | 2850 | 0.0 | - |
9.2652 | 2900 | 0.0004 | - |
9.4249 | 2950 | 0.0 | - |
9.5847 | 3000 | 0.0 | - |
9.7444 | 3050 | 0.0 | - |
9.9042 | 3100 | 0.0 | - |
10.0639 | 3150 | 0.0 | - |
10.2236 | 3200 | 0.0 | - |
10.3834 | 3250 | 0.0 | - |
10.5431 | 3300 | 0.0 | - |
10.7029 | 3350 | 0.0021 | - |
10.8626 | 3400 | 0.0002 | - |
11.0224 | 3450 | 0.0 | - |
11.1821 | 3500 | 0.0001 | - |
11.3419 | 3550 | 0.0 | - |
11.5016 | 3600 | 0.0 | - |
11.6613 | 3650 | 0.0 | - |
11.8211 | 3700 | 0.0 | - |
11.9808 | 3750 | 0.0 | - |
12.1406 | 3800 | 0.0 | - |
12.3003 | 3850 | 0.0002 | - |
12.4601 | 3900 | 0.0 | - |
12.6198 | 3950 | 0.0008 | - |
12.7796 | 4000 | 0.0002 | - |
12.9393 | 4050 | 0.0002 | - |
13.0990 | 4100 | 0.0002 | - |
13.2588 | 4150 | 0.0 | - |
13.4185 | 4200 | 0.0 | - |
13.5783 | 4250 | 0.0 | - |
13.7380 | 4300 | 0.0 | - |
13.8978 | 4350 | 0.0 | - |
14.0575 | 4400 | 0.0 | - |
14.2173 | 4450 | 0.0 | - |
14.3770 | 4500 | 0.0 | - |
14.5367 | 4550 | 0.0 | - |
14.6965 | 4600 | 0.0003 | - |
14.8562 | 4650 | 0.0 | - |
15.0160 | 4700 | 0.0 | - |
15.1757 | 4750 | 0.0 | - |
15.3355 | 4800 | 0.0 | - |
15.4952 | 4850 | 0.0 | - |
15.6550 | 4900 | 0.0 | - |
15.8147 | 4950 | 0.0 | - |
15.9744 | 5000 | 0.0 | - |
16.1342 | 5050 | 0.0 | - |
16.2939 | 5100 | 0.0 | - |
16.4537 | 5150 | 0.0001 | - |
16.6134 | 5200 | 0.0002 | - |
16.7732 | 5250 | 0.0 | - |
16.9329 | 5300 | 0.0002 | - |
17.0927 | 5350 | 0.0 | - |
17.2524 | 5400 | 0.0 | - |
17.4121 | 5450 | 0.0 | - |
17.5719 | 5500 | 0.0006 | - |
17.7316 | 5550 | 0.0001 | - |
17.8914 | 5600 | 0.0001 | - |
18.0511 | 5650 | 0.0 | - |
18.2109 | 5700 | 0.0 | - |
18.3706 | 5750 | 0.0002 | - |
18.5304 | 5800 | 0.0 | - |
18.6901 | 5850 | 0.0 | - |
18.8498 | 5900 | 0.0 | - |
19.0096 | 5950 | 0.0 | - |
19.1693 | 6000 | 0.0 | - |
19.3291 | 6050 | 0.0 | - |
19.4888 | 6100 | 0.0 | - |
19.6486 | 6150 | 0.0 | - |
19.8083 | 6200 | 0.0 | - |
19.9681 | 6250 | 0.0 | - |
20.1278 | 6300 | 0.0 | - |
20.2875 | 6350 | 0.0 | - |
20.4473 | 6400 | 0.0 | - |
20.6070 | 6450 | 0.0 | - |
20.7668 | 6500 | 0.0 | - |
20.9265 | 6550 | 0.0 | - |
21.0863 | 6600 | 0.0 | - |
21.2460 | 6650 | 0.0 | - |
21.4058 | 6700 | 0.0 | - |
21.5655 | 6750 | 0.0 | - |
21.7252 | 6800 | 0.0 | - |
21.8850 | 6850 | 0.0 | - |
22.0447 | 6900 | 0.0 | - |
22.2045 | 6950 | 0.0 | - |
22.3642 | 7000 | 0.0 | - |
22.5240 | 7050 | 0.0 | - |
22.6837 | 7100 | 0.0 | - |
22.8435 | 7150 | 0.0 | - |
23.0032 | 7200 | 0.0 | - |
23.1629 | 7250 | 0.0 | - |
23.3227 | 7300 | 0.0 | - |
23.4824 | 7350 | 0.0 | - |
23.6422 | 7400 | 0.0 | - |
23.8019 | 7450 | 0.0 | - |
23.9617 | 7500 | 0.0 | - |
24.1214 | 7550 | 0.0 | - |
24.2812 | 7600 | 0.0 | - |
24.4409 | 7650 | 0.0002 | - |
24.6006 | 7700 | 0.0 | - |
24.7604 | 7750 | 0.0 | - |
24.9201 | 7800 | 0.0 | - |
25.0799 | 7850 | 0.0 | - |
25.2396 | 7900 | 0.0 | - |
25.3994 | 7950 | 0.0 | - |
25.5591 | 8000 | 0.0 | - |
25.7188 | 8050 | 0.0 | - |
25.8786 | 8100 | 0.0 | - |
26.0383 | 8150 | 0.0 | - |
26.1981 | 8200 | 0.0 | - |
26.3578 | 8250 | 0.0 | - |
26.5176 | 8300 | 0.0 | - |
26.6773 | 8350 | 0.0 | - |
26.8371 | 8400 | 0.0 | - |
26.9968 | 8450 | 0.0 | - |
27.1565 | 8500 | 0.0 | - |
27.3163 | 8550 | 0.0 | - |
27.4760 | 8600 | 0.0 | - |
27.6358 | 8650 | 0.0 | - |
27.7955 | 8700 | 0.0 | - |
27.9553 | 8750 | 0.0 | - |
28.1150 | 8800 | 0.0001 | - |
28.2748 | 8850 | 0.0 | - |
28.4345 | 8900 | 0.0 | - |
28.5942 | 8950 | 0.0 | - |
28.7540 | 9000 | 0.0 | - |
28.9137 | 9050 | 0.0 | - |
29.0735 | 9100 | 0.0 | - |
29.2332 | 9150 | 0.0 | - |
29.3930 | 9200 | 0.0 | - |
29.5527 | 9250 | 0.0 | - |
29.7125 | 9300 | 0.0 | - |
29.8722 | 9350 | 0.0 | - |
Framework Versions
- Python: 3.10.12
- SetFit: 1.1.0
- Sentence Transformers: 3.3.1
- Transformers: 4.44.2
- PyTorch: 2.2.0a0+81ea7a4
- Datasets: 3.2.0
- Tokenizers: 0.19.1
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
- 0
This model does not have enough activity to be deployed to Inference API (serverless) yet. Increase its social
visibility and check back later, or deploy to Inference Endpoints (dedicated)
instead.