metadata

tags:
  - setfit
  - sentence-transformers
  - text-classification
  - generated_from_setfit_trainer
widget:
  - text: Mujhe apne galtiyon ka ehsaas hai aur main unke liye maafi chahta hoon.
  - text: >-
      Mujhe yeh step samajhne mein dikkat ho rahi hai, kya aap madad kar sakte
      hain?
  - text: Mujhe abhi tak kuch update kyun nahi mila, yeh bahut frustrating hai.
  - text: Is app ka loading time mujhe thoda zyada lagta hai.
  - text: Kya aap mujhe is event ki timing bata sakte hain?
metrics:
  - accuracy
pipeline_tag: text-classification
library_name: setfit
inference: true
base_model: MoritzLaurer/mDeBERTa-v3-base-mnli-xnli
model-index:
  - name: SetFit with MoritzLaurer/mDeBERTa-v3-base-mnli-xnli
    results:
      - task:
          type: text-classification
          name: Text Classification
        dataset:
          name: Unknown
          type: unknown
          split: test
        metrics:
          - type: accuracy
            value: 0.32
            name: Accuracy

SetFit with MoritzLaurer/mDeBERTa-v3-base-mnli-xnli

This is a SetFit model that can be used for Text Classification. This SetFit model uses MoritzLaurer/mDeBERTa-v3-base-mnli-xnli 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:

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: MoritzLaurer/mDeBERTa-v3-base-mnli-xnli
Classification head: a LogisticRegression instance
Maximum Sequence Length: 512 tokens
Number of Classes: 19 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
4	'Yeh rahin wo steps jisse aap apni payment kar sakte hain.' 'Kya aap mujhe yeh batane ka tarika thoda aasan kar sakte hain?' 'Is option ke madhyam se aap apni queries kaise solve kar sakte hain, jaan lijiye.'
16	'Aapke feedback ko humne dhyan mein rakha hai.' 'Yeh galti humare systems ki wajah se hui hai.' 'Mujhe is samasya ko suljhane mein zyada samay lena nahi chahiye tha.'
8	'Main aapko pareshan karne ke liye maafi chahta hoon.' 'Humein is samasya ke liye maafi chahiye.' 'Mere kaam se agar aapko takleef hui ho, toh mujhe maaf kar dijiye.'
13	'Mujhe yeh clarify karne ki zarurat hai ki agla step kya hai?' 'Mujhe pata karna hai ki maine jo complaint ki thi uska kya hua.' 'Mujhe bataye ki pehle kitne payments honge iss plan ke liye.'
15	'Yeh features sahi hai, lekin kuch aur additional functionalities honi chahiye.' 'Product ke size ki jankari hamesha saaf honi chahiye.' 'Main chahunga ki online form aur simple ho.'
12	'Mujhe product ke sath kuch samasya hai.' 'Mera phone charging nahi ho raha.' 'Mujhe courier service mein dikkat hai, report karna hai.'
11	'Mujhe samajh nahi aa raha, is offer mein koi chhupi shartein toh nahi hai?' 'Kis tarah se main feedback de sakta hoon?' 'Kya koi referral program hai jo mujhe join karna chahiye?'
2	'Item ke sath saathi accessories nahi mil rahe hain.' 'Aap logon ne jo samay liya, wo bilkul zyada tha.' 'Meri order delivery mein bahut der ho gayi hai.'
18	'Mujhe yeh bilkul pasand nahi hai ki meri baat ignore ki gayi.' 'Kam ke liye mera dosto ka support bahut sukhdayak hai.' 'Aaj ka din kaafi udaas beete raha hai.'
14	'Kya main kal ki delivery ko agle hafte reschedule kar sakta/sakti hoon?' 'Mujhe refund ke liye kya documents chahiye?' 'Kya main appointment ko dobara set kar sakta/sakti hoon?'
7	'Main aapko dhanyavad dena chahta hoon, aapne meri madad ki.' 'Aapne jo kiya, uske liye aapko sabse pehle prashansha milni chahiye.' 'Aapka samay dene ke liye abhaar.'
3	'Mujhe kisi event ke tickets ka status check karna hai.' 'Kya aap mujhe customer support number de sakte hain?' 'Main apne account ka balance kaise check kar sakta/sakti hoon?'
5	'Alvida, tumhara din acha rahe!' 'Hello! Aaj aap kaise hain?' 'Swagat hai! Kya main aapki kuch madad kar sakta hoon?'
0	'Mujhe kuch samajh nahi aa raha hai, kya mujhe thoda aur samjha sakte hain?' 'Agar main aisa karoon, to kya kuch badal jaayega? Main sure nahi hoon.' 'Yeh product ki warranty ki details clear nahi hain.'
6	'Chalo, alvida bolte hain!' 'Phir se baat karte hain!' 'Adieu, aapka din shubh ho!'
17	'Mere account mein login karne mein dikkat aa rahi hai, madad karein.' 'Mujhe apne account mein login karne mein madad chahiye.' 'Kya aap mujhe terms and conditions ke details de sakte hain?'
10	'Main aapki baat se sehmat hoon.' 'Mujhe yeh batayein ki meri booking sahi hai na?'
9	'Kya aap mujhe yeh concept aur clear kar sakte hain?' 'Mujhe yeh samajhne mein dikkat ho rahi hai, kya aap vyakhya de sakte hain?'
1	'Aaj dosto ke sath waqt bitana bahut acha laga.' 'Aaj baarish me bheegna bahut refreshing tha, mujhe yeh moment pasand aaya.' 'Aapka support bahut madadgar raha.'

Evaluation

Metrics

Label	Accuracy
all	0.32

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("rbojja/FT-mDeBERTa-v3-base-mnli-xnli")
# Run inference
preds = model("Kya aap mujhe is event ki timing bata sakte hain?")

Training Details

Training Set Metrics

Training set	Min	Median	Max
Word count	3	9.76	15

Label	Training Sample Count
0	6
1	3
2	3
3	5
4	7
5	3
6	6
7	8
8	6
9	2
10	2
11	5
12	6
13	5
14	9
15	9
16	9
17	3
18	3

Training Hyperparameters

batch_size: (16, 2)
num_epochs: (1, 16)
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
l2_weight: 0.01
seed: 42
eval_max_steps: -1
load_best_model_at_end: False

Training Results

Epoch	Step	Training Loss	Validation Loss
0.0017	1	0.2335	-
0.0853	50	0.2514	-
0.1706	100	0.1619	-
0.2560	150	0.1124	-
0.3413	200	0.078	-
0.4266	250	0.0623	-
0.5119	300	0.0576	-
0.5973	350	0.0421	-
0.6826	400	0.0391	-
0.7679	450	0.0386	-
0.8532	500	0.0302	-
0.9386	550	0.0245	-

Framework Versions

Python: 3.10.16
SetFit: 1.1.1
Sentence Transformers: 3.3.1
Transformers: 4.46.3
PyTorch: 2.5.1+cpu
Datasets: 3.2.0
Tokenizers: 0.20.3

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}
}