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:

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-mpnet-base-v2
Classification head: a OneVsRestClassifier instance
Maximum Sequence Length: 512 tokens

Model Sources

Repository: SetFit on GitHub
Paper: Efficient Few-Shot Learning Without Prompts
Blogpost: SetFit: Efficient Few-Shot Learning Without Prompts

Evaluation

Metrics

Label	Accuracy
all	0.5849

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-multilabel-setfit-model")
# Run inference
preds = model("It was a jihad training camp.
")

Training Details

Training Set Metrics

Training set	Min	Median	Max
Word count	1	26.6518	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
0.0006	1	0.3905	-
0.0275	50	0.2239	-
0.0550	100	0.2359	-
0.0826	150	0.2443	-
0.1101	200	0.2495	-
0.1376	250	0.2498	-
0.1651	300	0.116	-
0.1926	350	0.1672	-
0.2201	400	0.1281	-
0.2477	450	0.139	-
0.2752	500	0.0615	-
0.3027	550	0.0972	-
0.3302	600	0.0851	-
0.3577	650	0.1769	-
0.3853	700	0.1673	-
0.4128	750	0.0615	-
0.4403	800	0.1232	-
0.4678	850	0.0094	-
0.4953	900	0.0135	-
0.5228	950	0.0107	-
0.5504	1000	0.1137	-
0.5779	1050	0.0173	-
0.6054	1100	0.0573	-
0.6329	1150	0.0115	-
0.6604	1200	0.0374	-
0.6879	1250	0.0231	-
0.7155	1300	0.0392	-
0.7430	1350	0.0754	-
0.7705	1400	0.007	-
0.7980	1450	0.0138	-
0.8255	1500	0.0569	-
0.8531	1550	0.0971	-
0.8806	1600	0.1052	-
0.9081	1650	0.0084	-
0.9356	1700	0.0859	-
0.9631	1750	0.0081	-
0.9906	1800	0.0362	-
1.0	1817	-	0.2354
1.0182	1850	0.0429	-
1.0457	1900	0.056	-
1.0732	1950	0.0098	-
1.1007	2000	0.002	-
1.1282	2050	0.0892	-
1.1558	2100	0.0557	-
1.1833	2150	0.001	-
1.2108	2200	0.0125	-
1.2383	2250	0.0152	-
1.2658	2300	0.0202	-
1.2933	2350	0.0593	-
1.3209	2400	0.007	-
1.3484	2450	0.014	-
1.3759	2500	0.003	-
1.4034	2550	0.0012	-
1.4309	2600	0.0139	-
1.4584	2650	0.0149	-
1.4860	2700	0.002	-
1.5135	2750	0.009	-
1.5410	2800	0.0066	-
1.5685	2850	0.0173	-
1.5960	2900	0.0052	-
1.6236	2950	0.0039	-
1.6511	3000	0.0042	-
1.6786	3050	0.0339	-
1.7061	3100	0.001	-
1.7336	3150	0.0005	-
1.7611	3200	0.0049	-
1.7887	3250	0.01	-
1.8162	3300	0.0815	-
1.8437	3350	0.0227	-
1.8712	3400	0.005	-
1.8987	3450	0.0053	-
1.9263	3500	0.0152	-
1.9538	3550	0.0155	-
1.9813	3600	0.0182	-
2.0	3634	-	0.2266

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}
}

anismahmahi
/

G2-multilabel-setfit-model