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	0.8151

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/doubt_repetition_with_noPropaganda_SetFit")
# Run inference
preds = model("At some point, the officer fired her weapon striking the victim.")

Training Details

Training Set Metrics

Training set	Min	Median	Max
Word count	1	20.8138	129

Training Hyperparameters

batch_size: (16, 16)
num_epochs: (2, 2)
max_steps: -1
sampling_strategy: oversampling
num_iterations: 5
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
0.0004	1	0.3567	-
0.0209	50	0.3286	-
0.0419	100	0.2663	-
0.0628	150	0.2378	-
0.0838	200	0.1935	-
0.1047	250	0.2549	-
0.1257	300	0.2654	-
0.1466	350	0.1668	-
0.1676	400	0.1811	-
0.1885	450	0.1884	-
0.2095	500	0.157	-
0.2304	550	0.1237	-
0.2514	600	0.1318	-
0.2723	650	0.1334	-
0.2933	700	0.1067	-
0.3142	750	0.1189	-
0.3351	800	0.135	-
0.3561	850	0.0782	-
0.3770	900	0.0214	-
0.3980	950	0.0511	-
0.4189	1000	0.0924	-
0.4399	1050	0.1418	-
0.4608	1100	0.0132	-
0.4818	1150	0.0018	-
0.5027	1200	0.0706	-
0.5237	1250	0.1502	-
0.5446	1300	0.133	-
0.5656	1350	0.0207	-
0.5865	1400	0.0589	-
0.6075	1450	0.0771	-
0.6284	1500	0.0241	-
0.6494	1550	0.0905	-
0.6703	1600	0.0106	-
0.6912	1650	0.0451	-
0.7122	1700	0.0011	-
0.7331	1750	0.0075	-
0.7541	1800	0.0259	-
0.7750	1850	0.0052	-
0.7960	1900	0.0464	-
0.8169	1950	0.0039	-
0.8379	2000	0.0112	-
0.8588	2050	0.0061	-
0.8798	2100	0.0143	-
0.9007	2150	0.0886	-
0.9217	2200	0.2225	-
0.9426	2250	0.0022	-
0.9636	2300	0.0035	-
0.9845	2350	0.002	-
1.0	2387	-	0.2827
1.0054	2400	0.0315	-
1.0264	2450	0.0049	-
1.0473	2500	0.0305	-
1.0683	2550	0.0334	-
1.0892	2600	0.0493	-
1.1102	2650	0.0424	-
1.1311	2700	0.0011	-
1.1521	2750	0.0109	-
1.1730	2800	0.0009	-
1.1940	2850	0.0005	-
1.2149	2900	0.0171	-
1.2359	2950	0.0004	-
1.2568	3000	0.0717	-
1.2778	3050	0.0019	-
1.2987	3100	0.062	-
1.3196	3150	0.0003	-
1.3406	3200	0.0018	-
1.3615	3250	0.0011	-
1.3825	3300	0.0005	-
1.4034	3350	0.0208	-
1.4244	3400	0.0004	-
1.4453	3450	0.001	-
1.4663	3500	0.0003	-
1.4872	3550	0.0015	-
1.5082	3600	0.0004	-
1.5291	3650	0.0473	-
1.5501	3700	0.0092	-
1.5710	3750	0.032	-
1.5920	3800	0.0016	-
1.6129	3850	0.0623	-
1.6339	3900	0.0291	-
1.6548	3950	0.0386	-
1.6757	4000	0.002	-
1.6967	4050	0.0006	-
1.7176	4100	0.0005	-
1.7386	4150	0.0004	-
1.7595	4200	0.0004	-
1.7805	4250	0.0007	-
1.8014	4300	0.033	-
1.8224	4350	0.0001	-
1.8433	4400	0.0489	-
1.8643	4450	0.0754	-
1.8852	4500	0.0086	-
1.9062	4550	0.0092	-
1.9271	4600	0.0591	-
1.9481	4650	0.0013	-
1.9690	4700	0.0043	-
1.9899	4750	0.0338	-
2.0	4774	-	0.3304

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
/

doubt_repetition_with_noPropaganda_SetFit