vipinbansal179/SetFit_sms_Analyzer5c95292

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

This is a SetFit model that can be used for Text Classification. This SetFit model uses sentence-transformers/all-mpnet-base-v2 as the Sentence Transformer embedding model. A LogisticRegression 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/all-mpnet-base-v2
• Classification head: a LogisticRegression instance
• Maximum Sequence Length: 384 tokens
• Number of Classes: 3 classes

Model Sources

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

Model Labels

Label	Examples
2	'validity airtel xstream fiber id 20001896982 expire 04-sep-23 . please recharge rs 589 enjoy uninterrupted service . recharge , click www.airtel.in/5/c_summary ? n=021710937343_dsl . please ignore already pay .' 'initiate process add a/c . xxxx59 upi app - axis bank' 'google-pay registration initiate icici bank . do , report bank . card details/otp/cvv secret . disclose anyone .'
0	'rs 260.00 debit a/c xxxxxx7783 credit krjngm @ oksbi upi ref:325154274303. ? call 18005700 -bob' 'send rs.400.00 kotak bank ac x4524 7800600122 @ ybl 15-10-23.upi ref 328855774953. , kotak.com/fraud' 'send rs.400.00 kotak bank ac x4524 7800600122 @ ybl 15-10-23.upi ref 328855774953. , kotak.com/fraud'
1	'dear bob upi user , account credit inr 50.00 date 2023-08-27 11:41:09 upi ref 360562629741 - bob' 'receive rs.10000.00 kotak bank ac x4524 mahimagyamlani08 @ okaxis 21-08-23.bal:197,838.14.upi ref:323334598750' 'update ! inr5.66 credit federal bank account xxxx9374 jupiter app . happy bank !'

Evaluation

Metrics

Label	Accuracy
all	0.9722

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("vipinbansal179/SetFit_sms_Analyzer5c95292")
# Run inference
preds = model("< # > use otp : 8233 login turtlemintpro zck+rfoaqnm")

Training Details

Training Set Metrics

Training set	Min	Median	Max
Word count	5	20.5357	35

Label	Training Sample Count
0	31
1	28
2	81

Training Hyperparameters

• batch_size: (16, 16)
• num_epochs: (4, 4)
• max_steps: -1
• sampling_strategy: oversampling
• 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
0.0014	1	0.2939	-
0.0708	50	0.1698	-
0.1416	100	0.0557	-
0.2125	150	0.0614	-
0.2833	200	0.0099	-
0.3541	250	0.0005	-
0.4249	300	0.0002	-
0.4958	350	0.0001	-
0.5666	400	0.0001	-
0.6374	450	0.0001	-
0.7082	500	0.0001	-
0.7790	550	0.0001	-
0.8499	600	0.0002	-
0.9207	650	0.0001	-
0.9915	700	0.0001	-
1.0	706	-	0.0312
1.0623	750	0.0001	-
1.1331	800	0.0001	-
1.2040	850	0.0001	-
1.2748	900	0.0	-
1.3456	950	0.0001	-
1.4164	1000	0.0	-
1.4873	1050	0.0	-
1.5581	1100	0.0	-
1.6289	1150	0.0	-
1.6997	1200	0.0	-
1.7705	1250	0.0	-
1.8414	1300	0.0001	-
1.9122	1350	0.0	-
1.9830	1400	0.0001	-
2.0	1412	-	0.0366
2.0538	1450	0.0	-
2.1246	1500	0.0001	-
2.1955	1550	0.0	-
2.2663	1600	0.0	-
2.3371	1650	0.0	-
2.4079	1700	0.0	-
2.4788	1750	0.0	-
2.5496	1800	0.0	-
2.6204	1850	0.0	-
2.6912	1900	0.0	-
2.7620	1950	0.0	-
2.8329	2000	0.0	-
2.9037	2050	0.0	-
2.9745	2100	0.0	-
3.0	2118	-	0.0414
3.0453	2150	0.0	-
3.1161	2200	0.0	-
3.1870	2250	0.0	-
3.2578	2300	0.0	-
3.3286	2350	0.0	-
3.3994	2400	0.0	-
3.4703	2450	0.0	-
3.5411	2500	0.0	-
3.6119	2550	0.0	-
3.6827	2600	0.0	-
3.7535	2650	0.0	-
3.8244	2700	0.0	-
3.8952	2750	0.0	-
3.9660	2800	0.0	-
4.0	2824	-	0.0366

• 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: 2.16.0
• 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}
}