Zlovoblachko/dimension3_setfit

SetFit with sentence-transformers/all-MiniLM-L6-v2

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:

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-MiniLM-L6-v2
• Classification head: a LogisticRegression instance
• Maximum Sequence Length: 256 tokens
• Number of Classes: 2 classes

Model Sources

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

Evaluation

Metrics

Label	F1
all	0.5495

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("Zlovoblachko/dimension3_setfit")
# Run inference
preds = model("I loved the spiderman movie!")

Training Details

Training Hyperparameters

• batch_size: (16, 16)
• num_epochs: (1, 1)
• max_steps: -1
• sampling_strategy: oversampling
• body_learning_rate: (2.260895905036282e-05, 2.260895905036282e-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
0.0004	1	0.3835	-
0.0177	50	0.3106	-
0.0353	100	0.3232	-
0.0530	150	0.319	-
0.0706	200	0.3146	-
0.0883	250	0.3194	-
0.1059	300	0.3166	-
0.1236	350	0.2941	-
0.1412	400	0.3289	-
0.1589	450	0.3108	-
0.1766	500	0.3099	-
0.1942	550	0.3072	-
0.2119	600	0.2994	-
0.2295	650	0.3062	-
0.2472	700	0.3046	-
0.2648	750	0.3086	-
0.2825	800	0.3039	-
0.3001	850	0.3096	-
0.3178	900	0.3134	-
0.3355	950	0.2965	-
0.3531	1000	0.3147	-
0.3708	1050	0.317	-
0.3884	1100	0.3123	-
0.4061	1150	0.3221	-
0.4237	1200	0.2971	-
0.4414	1250	0.2928	-
0.4590	1300	0.2977	-
0.4767	1350	0.3268	-
0.4944	1400	0.2785	-
0.5120	1450	0.3156	-
0.5297	1500	0.3148	-
0.5473	1550	0.2909	-
0.5650	1600	0.3225	-
0.5826	1650	0.3072	-
0.6003	1700	0.3099	-
0.6179	1750	0.311	-
0.6356	1800	0.3213	-
0.6532	1850	0.2937	-
0.6709	1900	0.3177	-
0.6886	1950	0.3088	-
0.7062	2000	0.3017	-
0.7239	2050	0.3076	-
0.7415	2100	0.3164	-
0.7592	2150	0.295	-
0.7768	2200	0.2957	-
0.7945	2250	0.3064	-
0.8121	2300	0.3146	-
0.8298	2350	0.3114	-
0.8475	2400	0.3151	-
0.8651	2450	0.3033	-
0.8828	2500	0.3039	-
0.9004	2550	0.3152	-
0.9181	2600	0.3185	-
0.9357	2650	0.2927	-
0.9534	2700	0.3174	-
0.9710	2750	0.3003	-
0.9887	2800	0.3157	-

Framework Versions

• Python: 3.10.12
• SetFit: 1.1.0
• Sentence Transformers: 3.2.1
• Transformers: 4.44.2
• PyTorch: 2.5.0+cu121
• Datasets: 3.0.2
• Tokenizers: 0.19.1

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}
}