README.md

---
library_name: setfit
tags:
- setfit
- sentence-transformers
- text-classification
- generated_from_setfit_trainer
datasets:
- Kevinger/hub-report-dataset
metrics:
- accuracy
widget:
- text: 'FOXBOROUGH — With Bill Belichick gone and no clear heir to his personnel
    throne in New England, it remains murky who will have final say on the roster
    as the offseason gets rolling.


    At Jerod Mayo’s introductory press conference, Robert Kraft said it’d be collaborative
    approach for now, but sought to debunk the idea that ownership will be more involved.
    He said his family will continue to delegate to the football operations staff
    as they have since purchasing the team in 1994.


    BET ANYTHING GET $250 BONUS ESPN BET CLAIM OFFER MASS 21+ and present in MA, NJ,
    PA, VA, MD, WV, TN, LA, KS, KY, CO, AZ, IL, IA, IN, OH, MI. Gambling problem?
    Call 1-800-Gambler.


    “It will be the same input that we’ve had for the last three decades: We try to
    hire the best people we can find and let them do their job and hold them accountable,”
    Kraft said. “If you get involved and tell them what to do or try to influence
    them, you can’t hold them responsible and have them accountable. It’ll be within
    the people’s discretion who are the decision makers to do it, and if we’ve hired
    the wrong people, then we’ll have to make a change. But we’re going to try to
    enjoy it as fans.”


    Kraft said there’s only one situation where ownership will get involved in football
    ops, and that’s when it comes off-the-field issues.


    “The only area that we have really weighed in is when it comes to bringing in
    people that we might think are not the right character to be here and they have
    done things in their past,” Kraft said. “That’s the only time we’ve really weighed
    in.”'
- text: 'My mom loved Christmas so much, she would sometimes leave the tree up until
    April.


    She dyed a sheet blue for the sky behind the crèche and made a star of tin foil.
    The cradle would stay empty until Christmas morning; when we tumbled downstairs,
    the baby would be in his place, and the house would smell of roasting turkey.


    Mom always took it personally if you didn’t wear red or green on Christmas, and
    she signed all the presents “Love, Baby Jesus,” “Love, Virgin Mary” or “Love,
    St. Joseph.”


    (My brother Kevin was always upset that Joseph got short shrift, disappearing
    from the Bible; why wasn’t he around to boast about Jesus turning water into wine?)


    We went to midnight Mass back then, and it was magical, despite some boys wearing
    Washington Redskins bathrobes as they carried presents down the aisle for Baby
    Jesus.'
- text: 'It is the first time that this food-dunking behavior has been documented
    in parrots — it has also been observed in grackles and crows. And it was a serendipitous
    discovery for the lab, which typically relies on meticulously planned experiments
    to test the cockatoos’ renowned problem-solving skills. “But sometimes we get
    gifted with accidental things that just happen,” Dr. Auersperg said.


    Goffin’s cockatoos are known for their ability to use and manipulate objects.
    In earlier studies, Dr. Auersperg and her colleagues found, for instance, that
    the birds could open locked puzzle boxes and make their own tools to obtain out-of-reach
    food.


    But the researchers at the Goffin Lab did not typically pay close attention to
    the birds’ behavior at lunch, said Jeroen Zewald, a doctoral student in the lab
    and another author of the study. Until, one day last summer, they noticed something
    curious. An affectionate male bird named Pipin — “the gentleman of the group,”
    Mr. Zewald said — was dunking his food into the tub of water typically used for
    drinking and bathing. Two other birds in the lab, Kiwi and Muki, turned out to
    be dunkers, too, the researchers noticed.


    To study the behavior more systematically, Mr. Zewald and Dr. Auersperg spent
    12 days observing the birds’ lunchtime behaviors. In total, seven of the 18 birds
    were observed dunking food at least once, they found. (Still, Pipin, Kiwi and
    Muki were the undisputed dunkmasters, racking up many more “dunking events” than
    the other birds.)'
- text: 'SAN FRANCISCO — Celtics fans held their breath midway through Tuesday’s game
    against the Warriors. C’s star Jayson Tatum appeared to roll his left ankle as
    he hobbled back to the locker room with 7:45 left in the first quarter.


    On the bright side, Tatum retreated to the Celtics locker room under his own power.
    Tatum appeared to step on the Warriors’ Brandin Podziemski shoe midway through
    play. Fortunately, it didn’t look like there was too much force on the play as
    Tatum went to the locker room after twisting his ankle. Here’s a look at the play.


    BET ANYTHING GET $250 BONUS ESPN BET CLAIM OFFER MASS 21+ and present in MA, NJ,
    PA, VA, MD, WV, TN, LA, KS, KY, CO, AZ, IL, IA, IN, OH, MI. Gambling problem?
    Call 1-800-Gambler.


    Up to that point, Tatum had put up four points, two rebounds and one assist on
    2-for-3 shooting in four minutes.


    The injury didn’t appear serious initially, and that was indeed the case. Tatum
    returned to the Celtics bench with 2:19 left in the first quarter, walking under
    his own power and without a limp. Tatum made his return to the game to start the
    second quarter.


    This story will be updated.'
- text: 'A new episode of “Love After Lockup” will air on Friday, Jan. 12 on WE Tv
    at 9 p.m. ET.


    The new episode can also be streamed live on Philo, DirecTV Stream and fuboTV.
    All platforms offer a free trial for those interested in signing up for an account.


    “Love After Lockup” is said to be a spin off from WE Tv’s “Love During Lockup”
    as couples navigate their love lives through prison. The show will show inmates
    struggle to keep their love through video dates, letters and phone calls. But
    there’s no telling who can and can’t handle the cell wall that separates the couples.


    In the new episode, “Tayler confronts Chance; Melissa reveals secret surgery plans.
    Tensions flare as Kerok seeks Bri’s family’s acceptance. Shavel’s shower explodes
    as the mothers-in-law face off again. Mike comes clean; Blaine’s confession sends
    Lindsay spiraling.”


    How can I watch if I don’t have cable?


    If you don’t have access to cable television, you can stream “Love After Lockup”
    on streaming platforms Philo, DirecTV Stream and fuboTV.


    What is Philo?


    Philo is an over-the-top internet live TV streaming service that offers 60+ entertainment
    and lifestyle channels for the budget-friendly price of $25/month.


    If you purchase a product or register for an account through one of the links
    on our site, we may receive compensation.


    What is FuboTV?


    FuboTV is an over-the-top internet live TV streaming service that offers more
    than 100 channels, such as sports, news, entertainment and local channels.


    What is DirecTV Stream?


    The streaming platform offers a plethora of content including streaming the best
    of live and On Demand, starting with more than 75 live TV channels.'
pipeline_tag: text-classification
inference: false
base_model: sentence-transformers/paraphrase-mpnet-base-v2
model-index:
- name: SetFit with sentence-transformers/paraphrase-mpnet-base-v2
  results:
  - task:
      type: text-classification
      name: Text Classification
    dataset:
      name: Kevinger/hub-report-dataset
      type: Kevinger/hub-report-dataset
      split: test
    metrics:
    - type: accuracy
      value: 0.5086206896551724
      name: Accuracy
---

# SetFit with sentence-transformers/paraphrase-mpnet-base-v2

This is a [SetFit](https://github.com/huggingface/setfit) model trained on the [Kevinger/hub-report-dataset](https://huggingface.co/datasets/Kevinger/hub-report-dataset) dataset that can be used for Text Classification. This SetFit model uses [sentence-transformers/paraphrase-mpnet-base-v2](https://huggingface.co/sentence-transformers/paraphrase-mpnet-base-v2) as the Sentence Transformer embedding model. A OneVsRestClassifier 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](https://www.sbert.net) 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/paraphrase-mpnet-base-v2](https://huggingface.co/sentence-transformers/paraphrase-mpnet-base-v2)
- **Classification head:** a OneVsRestClassifier instance
- **Maximum Sequence Length:** 512 tokens
<!-- - **Number of Classes:** Unknown -->
- **Training Dataset:** [Kevinger/hub-report-dataset](https://huggingface.co/datasets/Kevinger/hub-report-dataset)
<!-- - **Language:** Unknown -->
<!-- - **License:** Unknown -->

### Model Sources

- **Repository:** [SetFit on GitHub](https://github.com/huggingface/setfit)
- **Paper:** [Efficient Few-Shot Learning Without Prompts](https://arxiv.org/abs/2209.11055)
- **Blogpost:** [SetFit: Efficient Few-Shot Learning Without Prompts](https://huggingface.co/blog/setfit)

## Evaluation

### Metrics
| Label   | Accuracy |
|:--------|:---------|
| **all** | 0.5086   |

## Uses

### Direct Use for Inference

First install the SetFit library:

```bash
pip install setfit
```

Then you can load this model and run inference.

```python
from setfit import SetFitModel

# Download from the 🤗 Hub
model = SetFitModel.from_pretrained("Kevinger/setfit-hub-multilabel-example")
# Run inference
preds = model("My mom loved Christmas so much, she would sometimes leave the tree up until April.

She dyed a sheet blue for the sky behind the crèche and made a star of tin foil. The cradle would stay empty until Christmas morning; when we tumbled downstairs, the baby would be in his place, and the house would smell of roasting turkey.

Mom always took it personally if you didn’t wear red or green on Christmas, and she signed all the presents “Love, Baby Jesus,” “Love, Virgin Mary” or “Love, St. Joseph.”

(My brother Kevin was always upset that Joseph got short shrift, disappearing from the Bible; why wasn’t he around to boast about Jesus turning water into wine?)

We went to midnight Mass back then, and it was magical, despite some boys wearing Washington Redskins bathrobes as they carried presents down the aisle for Baby Jesus.")
```

<!--
### Downstream Use

*List how someone could finetune this model on their own dataset.*
-->

<!--
### Out-of-Scope Use

*List how the model may foreseeably be misused and address what users ought not to do with the model.*
-->

<!--
## Bias, Risks and Limitations

*What are the known or foreseeable issues stemming from this model? You could also flag here known failure cases or weaknesses of the model.*
-->

<!--
### Recommendations

*What are recommendations with respect to the foreseeable issues? For example, filtering explicit content.*
-->

## Training Details

### Training Set Metrics
| Training set | Min | Median   | Max  |
|:-------------|:----|:---------|:-----|
| Word count   | 12  | 526.0625 | 6633 |

### Training Hyperparameters
- batch_size: (8, 8)
- num_epochs: (1, 1)
- max_steps: -1
- sampling_strategy: oversampling
- num_iterations: 20
- body_learning_rate: (2e-05, 2e-05)
- head_learning_rate: 2e-05
- 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: False

### Training Results
| Epoch  | Step | Training Loss | Validation Loss |
|:------:|:----:|:-------------:|:---------------:|
| 0.0031 | 1    | 0.1691        | -               |
| 0.1562 | 50   | 0.0678        | -               |
| 0.3125 | 100  | 0.0949        | -               |
| 0.4688 | 150  | 0.0083        | -               |
| 0.625  | 200  | 0.0048        | -               |
| 0.7812 | 250  | 0.0011        | -               |
| 0.9375 | 300  | 0.0005        | -               |

### Framework Versions
- Python: 3.10.12
- SetFit: 1.0.3
- Sentence Transformers: 2.3.1
- Transformers: 4.35.2
- PyTorch: 2.1.0+cu121
- Datasets: 2.16.1
- Tokenizers: 0.15.1

## Citation

### BibTeX
```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}
}
```

<!--
## Glossary

*Clearly define terms in order to be accessible across audiences.*
-->

<!--
## Model Card Authors

*Lists the people who create the model card, providing recognition and accountability for the detailed work that goes into its construction.*
-->

<!--
## Model Card Contact

*Provides a way for people who have updates to the Model Card, suggestions, or questions, to contact the Model Card authors.*
-->