mann2107/BCMPIIRAB_MiniLM_ALL

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: 14 classes

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
0	'Please send me quotation for a flight for Lindelani Mkhize - East London/ Durban 31 August @ 12:00' "I need to go to Fort Smith AR via XNA for PD days. I'd like to take AA 4064 at 10:00 am arriving 11:58 am on Monday, May 11 returning on AA 4064 at 12:26 pm arriving 2:16 pm on Saturday May 16. I will need a Hertz rental. I d like to stay at the Courtyard Marriott in Fort Smith on Monday through Thursday nights checking out on Friday morning." 'Can you please send me flight quotations for Mr Mthetho Sovara for travel to Bologna, Italy as per details below: 7 Oct: JHB to Bologna, Italy 14 Oct: Bologna, Italy to JHB'
1	'I need to cancel my flight booking from London Heathrow to JFK, New York, scheduled for August 15th, 2024. The booking reference is XJ12345.' 'Please cancel my flight for late March to Chicago and DC. Meetings have been cancelled. I am not available by phone.' 'I need to cancel the below trip due to illness in family. Could you please assist with this?'
2	'I need to change the departure time for my one-way flight from SFO to LAX on October 15th. Could you please reschedule it to a later flight around 6:00 PM on the same day?' 'Can you please extend my hotel reservation at the Marriott in Denver from November 19th to November 23rd, 2024? Originally, I was scheduled to check out on the 19th.' "Lerato I checked Selbourne B/B, its not a nice place. Your colleague Stella booked Lindelani Mkhize in Hempston it's a beautiful place next to Garden Court, please change the accommodation from Selbourne to Hempston. This Selbourne is on the outskirt and my colleagues are not familiar with East London"
3	'Please add the below employee to our Concur system. In addition, make sure the Ghost Card is added into their profile. Lindsay Griffin lgriffin@arlingtonroe.com' "Good afternoon - CAEP has 4 new staff members that we'd like to set - up new user profiles for. Please see the below information and let me know should anything additional be required. Last First Middle Travel Class Email Gender DOB Graham Rose - Helen Xiuqing Staff rose - helen.graham@caepnet.org Female 6/14/1995 Gumbs Mary - Frances Akua Staff mary.gumbs@caepnet.org Female 10/18/1995 Lee Elizabeth Andie Staff liz.lee@caepnet.org Female 4/23/1991 Gilchrist Gabriel Jake Staff gabriel.gilchrist@caepnet.org Male" 'Good Morning, Please create a profile for Amelia West: Name: Amelia Jean - Danielle West DOB: 05/21/1987 PH: 202 - 997 - 6592 Email: asuermann@facs.org'
4	'Hi, My name is Lucia De Las Heras property accountant at Trion Properties. I am missing a few receipts to allocate the following charges. Would you please be able to provide a detailed invoice? 10/10/2019 FROSCH/GANT TRAVEL MBLOOMINGTON IN - 21' 'I would like to request an invoice/s for the above-mentioned employee who stayed at your establishment.' "Hello, Looking for an invoice for the below charge to Ryan Schulke's card - could you please assist? Vendor: United Airlines Transaction Date: 02/04/2020 Amount: $2,132.07 Ticket Number: 0167515692834"
5	'This is the second email with this trip, but I still need an itinerary for trip scheduled for January 27. Derek' 'Please send us all the flights used by G4S Kenya in the year 2022. Sorry for the short notice but we need the information by 12:00 noon today.' 'Jen Holt Can you please send me the itinerary for Jen Holt for this trip this week to Jackson Mississippi?'
6	"I've had to call off my vacation. What are my options for getting refunded?" "Looks like I won't be traveling due to some health issues. Is getting a refund for my booking possible?" "I've fallen ill and can't travel as planned. Can you process a refund for me?"
7	'The arrangements as stated are acceptable. Please go ahead and confirm all bookings accordingly.' "I've reviewed the details and everything seems in order. Please proceed with the booking." 'This travel plan is satisfactory. Please secure the necessary reservations.'
8	'I need some clarification on charges for a rebooked flight. It seems higher than anticipated. Who can provide more details?' 'Wishing you and your family a very Merry Christmas and a Happy and Healthy New Year. I have one unidentified item this month, hope you can help, and as always thanks in advance. Very limited information on this. 11/21/2019 #N/A #N/A #N/A 142.45 Rail Europe North Amer' "We've identified a mismatch between our booking records and credit card statement. Who can assist with this issue?"
9	'I booked a hotel in Berlin for next month, but the confirmation email I received has the wrong dates. Can you please correct this and resend the confirmation?' "I need to arrange a shuttle for our team from the airport to the conference venue, but I haven't received any confirmation yet. Can someone check on this for me?" "When trying to book a flight for our CEO, the system shows an error stating 'payment not processed.' Can you assist in resolving this issue quickly?"
10	'Please assist with payment for the conference room booking at Hilton last week.' 'Kindly process the invoice for the catering services provided during the annual company meeting.' "Supplier, please find a statement with all invoices listed due for the IT maintenance services. If you've already paid, please forward proof and date of payment. Thank you for your support."
11	"Congratulations! You've been selected to win a brand new iPhone 14. Click here to claim your prize now!" 'Get rich quick! Invest in our exclusive cryptocurrency and watch your money grow 10x in just a month. Limited time offer!' 'Your PayPal account has been compromised. Please click here to verify your information and secure your account.'
12	'Your flight booking has been confirmed. Flight details: Flight #BA283 from LHR to LAX on November 10th, departure at 12:30 PM.' 'We regret to inform you that your hotel reservation at The Plaza, New York, was unsuccessful due to unavailability. Please try booking another date.' 'Your car rental reservation with Hertz has been confirmed. Pickup location: JFK Airport, Date: October 20th, Time: 10:00 AM.'
13	'We have received a request to charge the attached invoice to the corporate credit card on file for Jane Doe. Please confirm the payment details at your earliest convenience.' 'Dear Travel Agency, we regret to inform you that the room booked for Mr. John Smith is unavailable due to overbooking. We have arranged an alternative accommodation at a nearby hotel. Please advise if this is acceptable.' 'Regarding the recent stay of Mr. Alan Harper, we noticed a discrepancy in the billing. The minibar charges were not included in the initial invoice. Kindly review the attached revised bill.'

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("mann2107/BCMPIIRAB_MiniLM_ALL")
# Run inference
preds = model("Thank you for your email. Please go ahead and issue. Please invoice in KES")

Training Details

Training Set Metrics

Training set	Min	Median	Max
Word count	1	25.6577	136

Label	Training Sample Count
0	24
1	24
2	24
3	24
4	24
5	24
6	24
7	24
8	24
9	24
10	24
11	24
12	24
13	24

Training Hyperparameters

• batch_size: (16, 16)
• num_epochs: (2, 2)
• max_steps: -1
• sampling_strategy: oversampling
• num_iterations: 2
• body_learning_rate: (0.0002853593106019776, 0.0002853593106019776)
• head_learning_rate: 0.0002853593106019776
• 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
0.0119	1	0.2534	-
0.5952	50	0.0849	-
1.1905	100	0.0077	-
1.7857	150	0.0099	-

Framework Versions

• Python: 3.10.12
• SetFit: 1.1.0.dev0
• Sentence Transformers: 3.0.1
• Transformers: 4.42.4
• PyTorch: 2.3.1+cu121
• Datasets: 2.20.0
• 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}
}