SetFit with sentence-transformers/paraphrase-mpnet-base-v2

This is a SetFit model that can be used for Text Classification. This SetFit model uses sentence-transformers/paraphrase-mpnet-base-v2 as the Sentence Transformer embedding model. A OneVsRestClassifier instance is used for classification.

The model has been trained using an efficient few-shot learning technique that involves:

  1. Fine-tuning a Sentence Transformer with contrastive learning.
  2. Training a classification head with features from the fine-tuned Sentence Transformer.

Model Details

Model Description

Model Sources

Evaluation

Metrics

Label Accuracy
all 0.6021

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("anismahmahi/G2_replace_Whata_repetition_with_noPropaganda_SetFit")
# Run inference
preds = model("Columbus police are investigating the shootings.")

Training Details

Training Set Metrics

Training set Min Median Max
Word count 1 23.1093 129

Training Hyperparameters

  • batch_size: (16, 16)
  • num_epochs: (2, 2)
  • max_steps: -1
  • sampling_strategy: oversampling
  • num_iterations: 10
  • 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
0.0002 1 0.3592 -
0.0121 50 0.2852 -
0.0243 100 0.2694 -
0.0364 150 0.2182 -
0.0486 200 0.2224 -
0.0607 250 0.2634 -
0.0729 300 0.2431 -
0.0850 350 0.2286 -
0.0971 400 0.197 -
0.1093 450 0.2466 -
0.1214 500 0.2374 -
0.1336 550 0.2134 -
0.1457 600 0.2092 -
0.1578 650 0.1987 -
0.1700 700 0.2288 -
0.1821 750 0.1562 -
0.1943 800 0.27 -
0.2064 850 0.1314 -
0.2186 900 0.2144 -
0.2307 950 0.184 -
0.2428 1000 0.2069 -
0.2550 1050 0.1773 -
0.2671 1100 0.0704 -
0.2793 1150 0.1139 -
0.2914 1200 0.2398 -
0.3035 1250 0.0672 -
0.3157 1300 0.1321 -
0.3278 1350 0.0803 -
0.3400 1400 0.0589 -
0.3521 1450 0.0428 -
0.3643 1500 0.0886 -
0.3764 1550 0.0839 -
0.3885 1600 0.1843 -
0.4007 1650 0.0375 -
0.4128 1700 0.114 -
0.4250 1750 0.1264 -
0.4371 1800 0.0585 -
0.4492 1850 0.0586 -
0.4614 1900 0.0805 -
0.4735 1950 0.0686 -
0.4857 2000 0.0684 -
0.4978 2050 0.0803 -
0.5100 2100 0.076 -
0.5221 2150 0.0888 -
0.5342 2200 0.1091 -
0.5464 2250 0.038 -
0.5585 2300 0.0674 -
0.5707 2350 0.0562 -
0.5828 2400 0.0603 -
0.5949 2450 0.0669 -
0.6071 2500 0.0829 -
0.6192 2550 0.1442 -
0.6314 2600 0.0914 -
0.6435 2650 0.0357 -
0.6557 2700 0.0546 -
0.6678 2750 0.0748 -
0.6799 2800 0.0149 -
0.6921 2850 0.1067 -
0.7042 2900 0.0054 -
0.7164 2950 0.0878 -
0.7285 3000 0.0385 -
0.7407 3050 0.036 -
0.7528 3100 0.0902 -
0.7649 3150 0.0734 -
0.7771 3200 0.0369 -
0.7892 3250 0.0031 -
0.8014 3300 0.0113 -
0.8135 3350 0.0862 -
0.8256 3400 0.0549 -
0.8378 3450 0.0104 -
0.8499 3500 0.0072 -
0.8621 3550 0.0546 -
0.8742 3600 0.0579 -
0.8864 3650 0.0789 -
0.8985 3700 0.0711 -
0.9106 3750 0.0361 -
0.9228 3800 0.0292 -
0.9349 3850 0.0121 -
0.9471 3900 0.0066 -
0.9592 3950 0.0091 -
0.9713 4000 0.0027 -
0.9835 4050 0.0891 -
0.9956 4100 0.0186 -
1.0 4118 - 0.2746
1.0078 4150 0.0246 -
1.0199 4200 0.0154 -
1.0321 4250 0.0056 -
1.0442 4300 0.0343 -
1.0563 4350 0.0375 -
1.0685 4400 0.0106 -
1.0806 4450 0.0025 -
1.0928 4500 0.0425 -
1.1049 4550 0.0019 -
1.1170 4600 0.0014 -
1.1292 4650 0.0883 -
1.1413 4700 0.0176 -
1.1535 4750 0.0204 -
1.1656 4800 0.0011 -
1.1778 4850 0.005 -
1.1899 4900 0.0238 -
1.2020 4950 0.0362 -
1.2142 5000 0.0219 -
1.2263 5050 0.0487 -
1.2385 5100 0.0609 -
1.2506 5150 0.0464 -
1.2627 5200 0.0033 -
1.2749 5250 0.0087 -
1.2870 5300 0.0101 -
1.2992 5350 0.0529 -
1.3113 5400 0.0243 -
1.3235 5450 0.001 -
1.3356 5500 0.0102 -
1.3477 5550 0.0047 -
1.3599 5600 0.0034 -
1.3720 5650 0.0118 -
1.3842 5700 0.0742 -
1.3963 5750 0.0538 -
1.4085 5800 0.0162 -
1.4206 5850 0.0079 -
1.4327 5900 0.0027 -
1.4449 5950 0.0035 -
1.4570 6000 0.0581 -
1.4692 6050 0.0813 -
1.4813 6100 0.0339 -
1.4934 6150 0.0312 -
1.5056 6200 0.0323 -
1.5177 6250 0.0521 -
1.5299 6300 0.0016 -
1.5420 6350 0.0009 -
1.5542 6400 0.0967 -
1.5663 6450 0.0009 -
1.5784 6500 0.031 -
1.5906 6550 0.0114 -
1.6027 6600 0.0599 -
1.6149 6650 0.0416 -
1.6270 6700 0.0047 -
1.6391 6750 0.0234 -
1.6513 6800 0.0609 -
1.6634 6850 0.022 -
1.6756 6900 0.0042 -
1.6877 6950 0.0336 -
1.6999 7000 0.0592 -
1.7120 7050 0.0536 -
1.7241 7100 0.1198 -
1.7363 7150 0.1035 -
1.7484 7200 0.0549 -
1.7606 7250 0.027 -
1.7727 7300 0.0251 -
1.7848 7350 0.0225 -
1.7970 7400 0.0027 -
1.8091 7450 0.0309 -
1.8213 7500 0.024 -
1.8334 7550 0.0355 -
1.8456 7600 0.0239 -
1.8577 7650 0.0377 -
1.8698 7700 0.012 -
1.8820 7750 0.0233 -
1.8941 7800 0.0184 -
1.9063 7850 0.0022 -
1.9184 7900 0.0043 -
1.9305 7950 0.014 -
1.9427 8000 0.0083 -
1.9548 8050 0.0084 -
1.9670 8100 0.0009 -
1.9791 8150 0.002 -
1.9913 8200 0.0002 -
2.0 8236 - 0.2768
  • 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
21
Safetensors
Model size
109M params
Tensor type
F32
·
Inference Examples
Inference API (serverless) has been turned off for this model.

Model tree for anismahmahi/G2_replace_Whata_repetition_with_noPropaganda_SetFit

Finetuned
(256)
this model

Evaluation results