metadata
license: apache-2.0
tags:
- feature-extraction
- text-classification
- sentence-transformers
- setfit
pipeline_tag: feature-extraction
datasets:
- gentilrenard/lmd_ukraine_comments
metrics:
- accuracy
base_model: sentence-transformers/paraphrase-multilingual-mpnet-base-v2
model-index:
- name: sentence-transformers/paraphrase-multilingual-mpnet-base-v2
results:
- task:
type: feature-extraction
name: Text Classification
dataset:
name: gentilrenard/lmd_ukraine_comments
type: gentilrenard/lmd_ukraine_comments
split: test
metrics:
- type: accuracy
value: 0.762589928057554
name: Accuracy
SetFit with sentence-transformers/paraphrase-multilingual-mpnet-base-v2
This is a SetFit model trained on the gentilrenard/lmd_ukraine_comments dataset that can be used for Text Classification. This SetFit model uses sentence-transformers/paraphrase-multilingual-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:
- 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/paraphrase-multilingual-mpnet-base-v2
- Classification head: a LogisticRegression instance
- Maximum Sequence Length: 128 tokens
- Number of Classes: 3 classes
- Training Dataset: gentilrenard/lmd_ukraine_comments
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 |
|
0 |
|
1 |
|
Evaluation
Metrics
Label | Accuracy |
---|---|
all | 0.7626 |
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("gentilrenard/paraphrase-multilingual-mpnet-base-v2_setfit-lemonde-french")
# Run inference
preds = model("Pour Yves Pozzo di Borgo, c'est une tradition familliale. Charles André Pozzo di Borgo fut ambassadeur de la Russie.")
Training Details
Training Set Metrics
Training set | Min | Median | Max |
---|---|---|---|
Word count | 1 | 63.1703 | 180 |
Label | Training Sample Count |
---|---|
0 | 115 |
1 | 82 |
2 | 126 |
Training Hyperparameters
- batch_size: (32, 32)
- num_epochs: (2, 2)
- max_steps: 2350
- sampling_strategy: oversampling
- body_learning_rate: (3e-07, 3e-07)
- 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
- run_name: setfit_optimized_v4
- eval_max_steps: -1
- load_best_model_at_end: True
Training Results
Epoch | Step | Training Loss | Validation Loss |
---|---|---|---|
0.0005 | 1 | 0.243 | - |
0.0234 | 50 | 0.2654 | 0.2636 |
0.0467 | 100 | 0.2942 | 0.2611 |
0.0701 | 150 | 0.2462 | 0.2572 |
0.0934 | 200 | 0.2562 | 0.2546 |
0.1168 | 250 | 0.2445 | 0.2505 |
0.1401 | 300 | 0.2206 | 0.2473 |
0.1635 | 350 | 0.2435 | 0.2453 |
0.1868 | 400 | 0.1985 | 0.2425 |
0.2102 | 450 | 0.265 | 0.2411 |
0.2335 | 500 | 0.2408 | 0.2387 |
0.2569 | 550 | 0.1986 | 0.2369 |
0.2802 | 600 | 0.2071 | 0.2351 |
0.3036 | 650 | 0.2119 | 0.2341 |
0.3270 | 700 | 0.2558 | 0.2314 |
0.3503 | 750 | 0.215 | 0.2292 |
0.3737 | 800 | 0.2286 | 0.2271 |
0.3970 | 850 | 0.2495 | 0.2256 |
0.4204 | 900 | 0.1844 | 0.2237 |
0.4437 | 950 | 0.2529 | 0.2216 |
0.4671 | 1000 | 0.2074 | 0.2202 |
0.4904 | 1050 | 0.1753 | 0.2188 |
0.5138 | 1100 | 0.2216 | 0.2169 |
0.5371 | 1150 | 0.1878 | 0.2153 |
0.5605 | 1200 | 0.1862 | 0.2142 |
0.5838 | 1250 | 0.1682 | 0.2129 |
0.6072 | 1300 | 0.2425 | 0.2116 |
0.6305 | 1350 | 0.174 | 0.211 |
0.6539 | 1400 | 0.1641 | 0.209 |
0.6773 | 1450 | 0.2014 | 0.2094 |
0.7006 | 1500 | 0.1423 | 0.2083 |
0.7240 | 1550 | 0.204 | 0.2078 |
0.7473 | 1600 | 0.2265 | 0.2075 |
0.7707 | 1650 | 0.1812 | 0.2063 |
0.7940 | 1700 | 0.1804 | 0.2058 |
0.8174 | 1750 | 0.1658 | 0.2055 |
0.8407 | 1800 | 0.1374 | 0.2064 |
0.8641 | 1850 | 0.1316 | 0.2057 |
0.8874 | 1900 | 0.1566 | 0.205 |
0.9108 | 1950 | 0.2053 | 0.2035 |
0.9341 | 2000 | 0.1436 | 0.2046 |
0.9575 | 2050 | 0.2436 | 0.2039 |
0.9809 | 2100 | 0.1999 | 0.2038 |
1.0042 | 2150 | 0.1459 | 0.2042 |
1.0276 | 2200 | 0.1669 | 0.2044 |
1.0509 | 2250 | 0.1705 | 0.2042 |
1.0743 | 2300 | 0.1509 | 0.2038 |
1.0976 | 2350 | 0.1382 | 0.2036 |
- The bold row denotes the saved checkpoint.
Framework Versions
- Python: 3.10.12
- SetFit: 1.0.3
- Sentence Transformers: 2.3.0
- Transformers: 4.36.0
- PyTorch: 2.0.0
- Datasets: 2.16.1
- 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}
}