English Input Classifier

This is a SetFit model that can be used for Text Classification. This SetFit model uses sentence-transformers/all-MiniLM-L6-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:

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/all-MiniLM-L6-v2
Classification head: a LogisticRegression instance
Maximum Sequence Length: 256 tokens
Number of Classes: 4 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
request	'necessary lingerie' 'necessary material for today' 'I finished the room 234'
conversation	"What's up, uncle, all good?" 'Good, how is the thing going?!' 'Hello how are you'
help	'Please help' "Help I don't know what I can do" 'Hello, what can I do'
censorship	'You are a useless complete, you are useless' 'Always saying stupidities, better shut up' 'Your single existence is a shame'

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("monentiadev/en-input-classifier")
# Run inference
preds = model("Hello")

Training Details

Training Set Metrics

Training set	Min	Median	Max
Word count	1	5.1483	40

Label	Training Sample Count
censorship	576
conversation	123
help	204
request	520

Training Hyperparameters

batch_size: (128, 128)
num_epochs: (3, 3)
max_steps: -1
sampling_strategy: oversampling
num_iterations: 20
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
0.0022	1	0.3104	-
0.1124	50	0.3267	-
0.2247	100	0.2008	-
0.3371	150	0.0842	-
0.4494	200	0.0218	-
0.5618	250	0.0103	-
0.6742	300	0.0052	-
0.7865	350	0.0034	-
0.8989	400	0.0025	-
1.0112	450	0.0019	-
1.1236	500	0.0019	-
1.2360	550	0.0017	-
1.3483	600	0.001	-
1.4607	650	0.001	-
1.5730	700	0.0011	-
1.6854	750	0.0009	-
1.7978	800	0.001	-
1.9101	850	0.0007	-
2.0225	900	0.0008	-
2.1348	950	0.0007	-
2.2472	1000	0.0007	-
2.3596	1050	0.0006	-
2.4719	1100	0.0006	-
2.5843	1150	0.0006	-
2.6966	1200	0.0006	-
2.8090	1250	0.0006	-
2.9213	1300	0.0006	-

Framework Versions

Python: 3.10.0
SetFit: 1.1.2
Sentence Transformers: 5.0.0
Transformers: 4.53.1
PyTorch: 2.7.1+cu126
Datasets: 2.19.2
Tokenizers: 0.21.2

monentiadev
/

en-input-classifier