--- library_name: setfit tags: - setfit - sentence-transformers - text-classification - generated_from_setfit_trainer metrics: - accuracy widget: - text: 'versace art portfolio up for sale the art collection of murdered fashion designer gianni versace could fetch up to £9m ($17m) when it is auctioned in new york and london later this year. among the pictures for sale are works by roy lichtenstein andy warhol and henri matisse. the collection was housed at versace s six-storey new york townhouse. the 51-year-old designer was shot outside his florida home in 1997 by suspected serial killer andrew cunanan who later killed himself. the auction at sotheby s will feature 45 contemporary impressionist and 19th century paintings. one of the highlights of the sale is roy lichtenstein s blue nude which has been given an estimate of £1.8m ($3.4m). tobias meyer sotheby s worldwide head of contemporary art said: this collection reflects mr versace s wide-ranging taste and impeccable eye and many of the works were commissioned directly from the artists. outstanding later examples from champions of the pop movement such as roy lichtenstein are juxtaposed with masterpieces from the most visible artists of the 1980 s including jean-michel basquiat and the collaborative genius of basquiat and warhol as well as francesco clemente. much of the collection will be offered for sale at three auctions in new york in june with smaller contemporary paintings going under the hammer in london on 22 and 23 june. a sale of versace s furniture and artworks sold in 2001fetched £5.5m ($10.3m).' - text: 'councils prepare to set tax rises council tax in scotland is set to rise by an average of about 4% in the coming year bbc scotland has learned. authorities will decide final figures on thursday when projected increases will be more than twice the rate of inflation which is currently 1.6%. the finance minister has urged councils to limit increases but they have warned that they will struggle to maintain services unless funding is increased. they say much additional government money is for new initiatives. scottish finance minister tom mccabe msp said: last week in parliament i announced an additional £419m for core expenditure to local government in scotland. that s a 5.5% increase and sits against an inflation rate of 1.6% so i think we have quite rightly said to councils this year that we would at the very least ask them to exercise restraint. mr mccabe is also looking for local authorities to become more efficient and save money in coming years. he told bbc radio scotland s sunday live programme: here in scotland we have 32 councils who all have their own individual collection systems for council tax they have their own payroll systems and their own human resource systems. we think there has to be opportunities there for rationalisation and using the money saved to reinvest in frontline services. the councils umbrella organisation cosla which provided bbc scotland with the indicative figures for next year warned that councils would face a continuous struggle to maintain services. mr mccabe has promised them about £8.1bn next year. however most of the increase is targeted to new initiatives and councils will experience difficulties in maintaining core services a cosla spokesman said. cosla says that it is willing to work with the executive on finding efficiency savings but that these will not be enough to maintain services. they say the funding plans for the next three years will see councils lose more of the share of public spending. the conservatives accuse the scottish executive of using the council tax to raise funds because it is too afraid to raise income tax. the tory finance spokesman brian monteith msp said: its a form of disguise... yet again we see that council tax is being used as a way of passing on costs. scared of actually using its three pence income tax that it could put up what we ve seen over the years is more and more burdens being put onto local authorities and the council tax payer having to pick up the bill. there are also warnings that unless funding to councils is increased in the next few years then services may have to be reduced. linda knox director of the scottish local authority management centre at strathclyde university said: with this current settlement the increase is slowing. at the same time the burdens on councils are greater than they were. the settlement figures don t include pay increases and the executive is also requiring a substantial figure - in the area of £325m - in efficiency savings across the settlement period. education will be protected from any cuts but linda knox says this will mean other services will suffer. she said: in practice that will mean a 4-5% cut for other services. on the face of it the settlement looks like an increase of about 9.7% but by the time you take into account other factors its probably only about 1% in real terms.' - text: gadget show heralds mp3 christmas partners of those who love their hi-tech gear may want to get their presents in early as experts predict a gadget shortage this christmas. with apple s ipod topping wish lists again there may not be enough ipod minis to go round predicts oliver irish editor of gadget magazine stuff. the ipod mini is likely to be this year s tracey island said mr irish. stuff has compiled a list of the top 10 gadgets for 2004 and the ipod is at number one. for anyone bewildered by the choice of gadgets on the market stuff and what hi-fi are hosting a best-of gadget show in london this weekend. star of the show will be sony s qrio robot an all-singing all-dancing football-playing man-machine who can even hold intelligent conversations. but he is not for sale and sony has no commercial plans for the robot. he will greet visitors and is flying in from japan. he probably has his own airplane seat that is how highly sony prize him said mr irish. also on display will be a virtual keyboard which projects itself onto any flat surface. the event will play host to a large collection of digital music players from companies such as creative sony and philips as well as the ubiquitously fashionable ipod from apple. suggestions that it could be a gaming or wireless christmas are unlikely to come true as mp3 players remain the most popular stocking filler said mr irish. demand is huge and apple has promised that it can supply enough but people might struggle to get their hands on ipod minis said mr irish. for those who like their gadgets to be multi-talented the gizmondo a powerful gaming console with gps and gprs that also doubles up as an mp3 player movie player and camera could be a must-have. what is impressive is how much it can do and how well it can do them said mr irish. this christmas gadgets will not be an all-male preserve. women will be getting gadgets from husbands and boyfriends as well as buying them for themselves said mr irish. gadgets nowadays are lifestyle products rather than just for geeks. - text: 'virus poses as christmas e-mail security firms are warning about a windows virus disguising itself as an electronic christmas card. the zafi.d virus translates the christmas greeting on its subject line into the language of the person receiving infected e-mail. anti-virus firms speculate that this multilingual ability is helping the malicious program spread widely online. anti-virus firm sophos said that 10% of the e-mail currently on the net was infected with the zafi virus. like many other windows viruses zafi-d plunders microsoft outlook for e-mail addresses and then uses mail-sending software to despatch itself across the web to new victims. to be infected users must open up the attachment travelling with the message which bears the code for the malicious bug. the attachment on the e-mail poses as an electronic christmas card but anyone opening it will simply get a crude image of two smiley faces. the virus subject line says merry christmas and translates this into one of 15 languages depending of the final suffix of the e-mail address the infected message has been sent to. the message in the body of the e-mail reads: happy holidays and this too is translated. on infected machines the virus tries to disable anti-virus and firewall software and opens up a backdoor on the pc to hand over control to the writer of the virus. the virus is thought to have spread most widely in south america italy spain bulgaria and hungary. the original zafi virus appeared in april this year. we have seen these hoaxes for several christmases already and personally i prefer traditional pen and paper cards and we recommend this to all our clients too said mikko hypponen who heads f-secure s anti-virus team.' - text: desailly backs blues revenge trip marcel desailly insists there is no chance of history repeating itself when chelsea take on barcelona on wednesday. the french star was part of the chelsea side crushed 5-1 at the nou camp in the champions league quarter-final second leg in 2000. things will be totally different this time he told bbc sport. now everyone knows about chelsea and is a little bit afraid of them. they are one of the major clubs in europe and the pressure will be on barcelona. chelsea have not played barcelona since that quarter-final tie five years ago. the blues had looked destined to progress after winning the first leg at stamford bridge 3-1 courtesy of two goals from tore andre flo and one by gianfranco zola. but they collapsed in the second leg going down to strikes from rivaldo (2) luis figo dani and patrick kluivert. former chelsea captain desailly who is now playing for al-gharafa in qatar says there is no comparison between that side and the current blues team who are top of the premiership. mentally they are much stronger even though a lot of their players are young the 36-year-old said. we made some mistakes at the nou camp in 2000 - a lot of them were individual mistakes. it would not happen now. this team has a new motivation and a different mentality. world cup winner desailly saw huge changes during his time at stamford bridge. he was signed for £4.6m from ac milan in 1998 by ruud gullit and went on to play under gianluca vialli and claudio ranieri. but the biggest change occurred when billionaire roman abramovich bought the club in 2003. desailly says the russian s arrival helped to instil a winning mentality at the club as well as a demand for success. the whole of chelsea is different now - the chairman the manager and all the players he said. everything is new and there is a huge determination to win. since that game in 2000 chelsea have gained more experience in europe and were very close to reaching the champions league final last season. desailly is one of the most decorated players in the history of football. he won the 1998 world cup and 2000 european championship with france the champions league in 1993 with marseilles and 1994 with ac milan two serie a titles and the fa cup in 2000 with chelsea. he is now winding down his career in qatar alongside the likes of frank lebeouf josep guardiola titi camara gabriel batistuta and christophe dugarry. so he is full of admiration for two of his colleagues from the great milan side of the mid-90s who are likely to line up against manchester united on wednesday - paolo maldini and alessandro costacurta. i m happy that they have managed to play so long at a high level he said. i made a vow to costacurta that as long as he plays i will continue to play. and it s amazing that paolo has managed to play at such a high level for such a long time. pipeline_tag: text-classification inference: true 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: Unknown type: unknown split: test metrics: - type: accuracy value: 0.953 name: Accuracy --- # SetFit with sentence-transformers/paraphrase-mpnet-base-v2 This is a [SetFit](https://github.com/huggingface/setfit) model 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 [LogisticRegression](https://scikit-learn.org/stable/modules/generated/sklearn.linear_model.LogisticRegression.html) 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 [LogisticRegression](https://scikit-learn.org/stable/modules/generated/sklearn.linear_model.LogisticRegression.html) instance - **Maximum Sequence Length:** 512 tokens - **Number of Classes:** 5 classes ### 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) ### Model Labels | Label | Examples | |:------|:-------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------| | 4 | | | 0 | | | 2 | | | 3 | | | 1 | | ## Evaluation ### Metrics | Label | Accuracy | |:--------|:---------| | **all** | 0.953 | ## 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("vidhi0206/setfit-paraphrase-mpnet-base-v2") # Run inference preds = model("versace art portfolio up for sale the art collection of murdered fashion designer gianni versace could fetch up to £9m ($17m) when it is auctioned in new york and london later this year. among the pictures for sale are works by roy lichtenstein andy warhol and henri matisse. the collection was housed at versace s six-storey new york townhouse. the 51-year-old designer was shot outside his florida home in 1997 by suspected serial killer andrew cunanan who later killed himself. the auction at sotheby s will feature 45 contemporary impressionist and 19th century paintings. one of the highlights of the sale is roy lichtenstein s blue nude which has been given an estimate of £1.8m ($3.4m). tobias meyer sotheby s worldwide head of contemporary art said: this collection reflects mr versace s wide-ranging taste and impeccable eye and many of the works were commissioned directly from the artists. outstanding later examples from champions of the pop movement such as roy lichtenstein are juxtaposed with masterpieces from the most visible artists of the 1980 s including jean-michel basquiat and the collaborative genius of basquiat and warhol as well as francesco clemente. much of the collection will be offered for sale at three auctions in new york in june with smaller contemporary paintings going under the hammer in london on 22 and 23 june. a sale of versace s furniture and artworks sold in 2001fetched £5.5m ($10.3m).") ``` ## Training Details ### Training Set Metrics | Training set | Min | Median | Max | |:-------------|:----|:--------|:-----| | Word count | 173 | 419.325 | 1121 | | Label | Training Sample Count | |:------|:----------------------| | 0 | 8 | | 1 | 8 | | 2 | 8 | | 3 | 8 | | 4 | 8 | ### 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 | |:-----:|:----:|:-------------:|:---------------:| | 0.005 | 1 | 0.245 | - | | 0.25 | 50 | 0.0174 | - | | 0.5 | 100 | 0.0008 | - | | 0.75 | 150 | 0.0005 | - | | 1.0 | 200 | 0.0002 | - | ### Framework Versions - Python: 3.8.10 - SetFit: 1.0.3 - Sentence Transformers: 2.3.1 - Transformers: 4.37.2 - PyTorch: 2.2.0+cu121 - Datasets: 2.17.0 - 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} } ```