You can deploy this model and start earning money today!

st-karlos-efood/
setfit-multilabel-one-vs-rest-feb-2024

Name: setfit-multilabel-one-vs-rest-feb-2024
Availability: InStock

setfit

safetensors

xlm-roberta

sentence-transformers

text-classification

generated_from_setfit_trainer

2209.11055

lighteternal/stsb-xlm-r-greek-transfer

finetune

model-index

Views

No views yet

SetFit with lighteternal/stsb-xlm-r-greek-transfer

This is a SetFit model that can be used for Text Classification. This SetFit model uses lighteternal/stsb-xlm-r-greek-transfer 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: lighteternal/stsb-xlm-r-greek-transfer
Classification head: a OneVsRestClassifier instance
Maximum Sequence Length: 400 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	0.1589

Uses

Direct Use for Inference

First install the SetFit library:

pip install setfit

Then you can load this model and run inference.

python

from setfit import SetFitModel

# Download from the 🤗 Hub
model = SetFitModel.from_pretrained("st-karlos-efood/setfit-multilabel-one-vs-rest-feb-2024")
# Run inference
preds = model("παστα ατομικη")

Training Details

Training Set Metrics

Training set	Min	Median	Max
Word count	1	8.6048	116

Training Hyperparameters

batch_size: (48, 48)
num_epochs: (5, 5)
max_steps: -1
sampling_strategy: oversampling
num_iterations: 10
body_learning_rate: (2e-05, 2e-05)
head_learning_rate: 2e-05
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: False

Training Results

Epoch	Step	Training Loss	Validation Loss
0.0008	1	0.2009	-
0.0377	50	0.1674	-
0.0754	100	0.1593	-
0.1131	150	0.1793	-
0.1508	200	0.176	-
0.1885	250	0.1818	-
0.2262	300	0.1209	-
0.2640	350	0.1546	-
0.3017	400	0.0996	-
0.3394	450	0.1108	-
0.3771	500	0.1163	-
0.4148	550	0.1102	-
0.4525	600	0.1477	-
0.4902	650	0.0973	-
0.5279	700	0.1324	-
0.5656	750	0.1792	-
0.6033	800	0.1026	-
0.6410	850	0.1461	-
0.6787	900	0.117	-
0.7164	950	0.0907	-
0.7541	1000	0.0904	-
0.7919	1050	0.1168	-
0.8296	1100	0.0831	-
0.8673	1150	0.0623	-
0.9050	1200	0.0802	-
0.9427	1250	0.0802	-
0.9804	1300	0.1212	-
1.0181	1350	0.0872	-
1.0558	1400	0.1068	-
1.0935	1450	0.0975	-
1.1312	1500	0.096	-
1.1689	1550	0.0649	-
1.2066	1600	0.1004	-
1.2443	1650	0.0818	-
1.2821	1700	0.0714	-
1.3198	1750	0.0875	-
1.3575	1800	0.0893	-
1.3952	1850	0.1132	-
1.4329	1900	0.1127	-
1.4706	1950	0.0707	-
1.5083	2000	0.0819	-
1.5460	2050	0.0954	-
1.5837	2100	0.0948	-
1.6214	2150	0.0953	-
1.6591	2200	0.0813	-
1.6968	2250	0.0974	-
1.7345	2300	0.0785	-
1.7722	2350	0.086	-
1.8100	2400	0.0808	-
1.8477	2450	0.1014	-
1.8854	2500	0.112	-
1.9231	2550	0.0765	-
1.9608	2600	0.0694	-
1.9985	2650	0.0915	-
2.0362	2700	0.087	-
2.0739	2750	0.0831	-
2.1116	2800	0.1223	-
2.1493	2850	0.0897	-
2.1870	2900	0.0937	-
2.2247	2950	0.0862	-
2.2624	3000	0.0977	-
2.3002	3050	0.0563	-
2.3379	3100	0.1197	-
2.3756	3150	0.095	-
2.4133	3200	0.0702	-
2.4510	3250	0.0823	-
2.4887	3300	0.1309	-
2.5264	3350	0.0612	-
2.5641	3400	0.0994	-
2.6018	3450	0.0904	-
2.6395	3500	0.0678	-
2.6772	3550	0.0896	-
2.7149	3600	0.0753	-
2.7526	3650	0.0997	-
2.7903	3700	0.0956	-
2.8281	3750	0.1016	-
2.8658	3800	0.0784	-
2.9035	3850	0.0911	-
2.9412	3900	0.0485	-
2.9789	3950	0.1078	-
3.0166	4000	0.0659	-
3.0543	4050	0.0802	-
3.0920	4100	0.12	-
3.1297	4150	0.0519	-
3.1674	4200	0.047	-
3.2051	4250	0.0906	-
3.2428	4300	0.0999	-
3.2805	4350	0.059	-
3.3183	4400	0.0533	-
3.3560	4450	0.1033	-
3.3937	4500	0.0871	-
3.4314	4550	0.065	-
3.4691	4600	0.1487	-
3.5068	4650	0.0542	-
3.5445	4700	0.0846	-
3.5822	4750	0.0756	-
3.6199	4800	0.0518	-
3.6576	4850	0.1035	-
3.6953	4900	0.1129	-
3.7330	4950	0.1319	-
3.7707	5000	0.0804	-
3.8084	5050	0.108	-
3.8462	5100	0.1246	-
3.8839	5150	0.0923	-
3.9216	5200	0.1048	-
3.9593	5250	0.0951	-
3.9970	5300	0.1015	-
4.0347	5350	0.0888	-
4.0724	5400	0.0917	-
4.1101	5450	0.0823	-
4.1478	5500	0.0882	-
4.1855	5550	0.0807	-
4.2232	5600	0.0997	-
4.2609	5650	0.0782	-
4.2986	5700	0.1165	-
4.3363	5750	0.0837	-
4.3741	5800	0.1098	-
4.4118	5850	0.0564	-
4.4495	5900	0.0715	-
4.4872	5950	0.0858	-
4.5249	6000	0.0889	-
4.5626	6050	0.0719	-
4.6003	6100	0.1076	-
4.6380	6150	0.1044	-
4.6757	6200	0.0914	-
4.7134	6250	0.1078	-
4.7511	6300	0.1137	-
4.7888	6350	0.0666	-
4.8265	6400	0.1009	-
4.8643	6450	0.0537	-
4.9020	6500	0.0576	-
4.9397	6550	0.1366	-
4.9774	6600	0.1009	-

Framework Versions

Python: 3.10.12
SetFit: 1.0.3
Sentence Transformers: 2.3.1
Transformers: 4.35.2
PyTorch: 2.1.0+cu121
Datasets: 2.17.0
Tokenizers: 0.15.1

Citation

BibTeX

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}
}