faodl/model_inclusive_credit_and_financial_services

SetFit with sentence-transformers/paraphrase-multilingual-MiniLM-L12-v2

This is a SetFit model that can be used for Text Classification. This SetFit model uses sentence-transformers/paraphrase-multilingual-MiniLM-L12-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 Type: SetFit
• Sentence Transformer body: sentence-transformers/paraphrase-multilingual-MiniLM-L12-v2
• Classification head: a OneVsRestClassifier instance
• Maximum Sequence Length: 128 tokens

Model Sources

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

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("faodl/model_inclusive_credit_and_financial_services")
# Run inference
preds = model("Training infrastructure will be adapted to accommodate 
new	programmes.")

Training Details

Training Set Metrics

Training set	Min	Median	Max
Word count	1	47.7118	1181

Training Hyperparameters

• batch_size: (16, 16)
• num_epochs: (1, 1)
• max_steps: -1
• sampling_strategy: oversampling
• num_iterations: 20
• 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
• l2_weight: 0.01
• seed: 42
• eval_max_steps: -1
• load_best_model_at_end: False

Training Results

Epoch	Step	Training Loss	Validation Loss
0.0003	1	0.265	-
0.0142	50	0.1787	-
0.0283	100	0.1639	-
0.0425	150	0.1446	-
0.0567	200	0.1216	-
0.0708	250	0.1237	-
0.0850	300	0.1108	-
0.0992	350	0.1177	-
0.1133	400	0.1069	-
0.1275	450	0.1032	-
0.1416	500	0.1064	-
0.1558	550	0.1048	-
0.1700	600	0.0913	-
0.1841	650	0.0868	-
0.1983	700	0.098	-
0.2125	750	0.0788	-
0.2266	800	0.0699	-
0.2408	850	0.0839	-
0.2550	900	0.0841	-
0.2691	950	0.0823	-
0.2833	1000	0.0732	-
0.2975	1050	0.0674	-
0.3116	1100	0.0731	-
0.3258	1150	0.0744	-
0.3399	1200	0.0748	-
0.3541	1250	0.0735	-
0.3683	1300	0.067	-
0.3824	1350	0.0712	-
0.3966	1400	0.0731	-
0.4108	1450	0.0582	-
0.4249	1500	0.0577	-
0.4391	1550	0.0681	-
0.4533	1600	0.0572	-
0.4674	1650	0.058	-
0.4816	1700	0.0611	-
0.4958	1750	0.063	-
0.5099	1800	0.0565	-
0.5241	1850	0.0592	-
0.5382	1900	0.0576	-
0.5524	1950	0.0624	-
0.5666	2000	0.0627	-
0.5807	2050	0.0695	-
0.5949	2100	0.0485	-
0.6091	2150	0.0546	-
0.6232	2200	0.0453	-
0.6374	2250	0.0569	-
0.6516	2300	0.0528	-
0.6657	2350	0.047	-
0.6799	2400	0.0503	-
0.6941	2450	0.0456	-
0.7082	2500	0.0563	-
0.7224	2550	0.0512	-
0.7365	2600	0.059	-
0.7507	2650	0.0398	-
0.7649	2700	0.0489	-
0.7790	2750	0.0477	-
0.7932	2800	0.0467	-
0.8074	2850	0.0479	-
0.8215	2900	0.049	-
0.8357	2950	0.0544	-
0.8499	3000	0.0504	-
0.8640	3050	0.0429	-
0.8782	3100	0.0479	-
0.8924	3150	0.0465	-
0.9065	3200	0.0478	-
0.9207	3250	0.056	-
0.9348	3300	0.0522	-
0.9490	3350	0.0459	-
0.9632	3400	0.0428	-
0.9773	3450	0.0418	-
0.9915	3500	0.0439	-

Framework Versions

• Python: 3.11.13
• SetFit: 1.1.2
• Sentence Transformers: 4.1.0
• Transformers: 4.52.4
• PyTorch: 2.6.0+cu124
• Datasets: 3.6.0
• Tokenizers: 0.21.1

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}
}