metadata
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 model trained on the Kevinger/hub-report-dataset dataset that can be used for Text Classification. This SetFit model uses 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:
- 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-mpnet-base-v2
- Classification head: a OneVsRestClassifier instance
- Maximum Sequence Length: 512 tokens
- Training Dataset: Kevinger/hub-report-dataset
Model Sources
- Repository: SetFit on GitHub
- Paper: Efficient Few-Shot Learning Without Prompts
- Blogpost: SetFit: Efficient Few-Shot Learning Without Prompts
Evaluation
Metrics
| Label | Accuracy |
|---|---|
| all | 0.5086 |
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("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.")
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
@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}
}