Hugging Face's logo

chandanzeon
/
setfit_finetuned_iaf_98

Text Classification
setfit
Safetensors
sentence-transformers
bert
generated_from_setfit_trainer
Eval Results
Model card Files Community

You need to agree to share your contact information to access this model

This repository is publicly accessible, but you have to accept the conditions to access its files and content.

Log in or Sign Up to review the conditions and access this model content.

SetFit with BAAI/bge-small-en-v1.5

This is a SetFit model that can be used for Text Classification. This SetFit model uses BAAI/bge-small-en-v1.5 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 Sources

Model Labels

Label Examples
3
  • 'Closing Ceremony and Awards Distribution\nThe event concluded with the closing ceremony on August 18, 2024. The ceremony was attended by senior military officials, including the Chief of Naval Staff and Chief of Air Staff. The athletes were recognized for their outstanding performances with awards presented in various categories, such as Best Athlete, Best Team, and Best Sportsmanship. The Indian Air Force won the Best Team Performance trophy, while Captain Aaryan Verma from the Army was named the Best Athlete for his exceptional performance in athletics. The Indian Army was declared the overall winner of the competition, having secured the most points across all events. A highlight of the ceremony was the traditional military drill performed by the three services, showcasing the discipline and precision that is characteristic of the Indian Armed Forces.'
  • 'Community Engagement Activities\nIn addition to the aerial demonstrations, the IAF organized several community outreach activities. A special education booth was set up for schoolchildren, focusing on the history of the Indian Air Force, aviation careers, and the importance of air defense. The booth also displayed several educational films and interactive content about the Air Forceâ\x80\x99s role in peacekeeping, disaster relief, and national defense. Additionally, a blood donation camp was established near the main entrance, in collaboration with Bangaloreâ\x80\x99s city hospital, to encourage voluntary blood donation. Visitors were encouraged to participate, with the goal of increasing awareness about health and wellness in the community. The camp ran smoothly and successfully collected over 500 units of blood, which would be distributed to regional hospitals.'
  • 'Environmental Considerations\nWith the ongoing infrastructure developments, the Indian Air Force has taken steps to minimize the environmental impact of the construction. Measures are being implemented to reduce waste and promote sustainability during the projectâ\x80\x99s execution. Additionally, the base has set up an environmental monitoring system to track air and water quality in the vicinity, ensuring that the baseâ\x80\x99s operations do not adversely affect local ecosystems. Moreover, the wastewater treatment facility at the station is being upgraded to ensure that all waste generated from daily operations is properly treated and does not affect the surrounding areas. This initiative is part of the Air Forceâ\x80\x99s broader commitment to environmental responsibility and sustainability.'
1
  • 'Radio Frequency Allocation Updates\nThe communications division recently conducted a comprehensive review of radio frequency allocations across the northern and northeastern sectors. Adjustments were made to avoid overlaps that could interfere with civilian and military operations. A new allocation plan has been implemented for units stationed at Bagdogra and Dimapur, ensuring seamless communication during both routine and emergency operations. Periodic audits of frequency usage continue to safeguard against potential breaches or overlaps.'
  • 'Signal Frequency Interference Monitoring\nSignal frequency interference has become a growing concern as electronic warfare threats evolve. The monitoring units, based in key strategic areas such as Karnal, Leh, and Udhampur, have observed unauthorized intrusions into radio communication patterns. Advanced detection technologies have been deployed to analyze this data, with initial results highlighting the need for improved counter-electronic warfare capabilities. The Signal Corps has expanded its focus on electronic jamming threats near key tactical airstrips and operational centers. Units in these regions are conducting surveillance with advanced signal detection systems, ensuring they can identify and neutralize attempts at electronic disruption. Military radar units in Jaisalmer and Pathankot are receiving new upgrades to improve signal detection in operationally sensitive regions. Commanders have emphasized the importance of coordinating electronic warfare drills with these signal monitoring operations to enhance response mechanisms. Coordination between signal analysis teams and field operations ensures timely detection and neutralization of electronic threats.'
  • 'Naval Assets and Maritime Patrolling Operations\nNaval deployments along key trade routes and strategic maritime chokepoints have seen increased patrols and strategic upgrades. Units have been repositioned in response to recent developments in regional waters, focusing on both counter-terror operations and maintaining freedom of navigation. Surveillance assets, such as Indian Navy frigates and long-range maritime reconnaissance aircraft, are actively monitoring the Malabar Sea and Arabian Sea for any irregular ship movements or unauthorized military deployments. Naval units stationed at key operational ports like Visakhapatnam and Karwar are equipped with advanced sonar and radar systems. Recent deployments emphasize anti-submarine warfare capabilities, leveraging advanced underwater detection technology to identify potential threats from hostile assets or insurgent activity. Coordination with air assets, including the Sea King and P-8I aircraft, has improved naval surveillance effectiveness, with regular joint operations enhancing strategic interoperability.'
2
  • "Training Manuals for Official Use Only\nThe following manuals were distributed among units for use during December 2024 training sessions: 1. Guidelines for Advanced Vehicle Maintenance: ï\x82· This manual provides detailed procedures for troubleshooting and repairing light utility vehicles commonly used in supply operations. Emphasis is placed on maintaining vehicle efficiency in cold-weather conditions. ï\x82· A new section outlines methods for diagnosing electronic systems, a critical aspect as newer models are introduced into service. 2. Basic Communication Protocols: ï\x82· Designed for new recruits, this guide introduces secure communication techniques, including encryption basics and signal relay procedures. ï\x82· The document also includes practical exercises to simulate field scenarios, enhancing recruits' readiness for real-world applications."
  • 'Routine Procurement Documents\nThe procurement department at Ambala Air Force Station has finalized contracts for the supply of spare parts for MiG-21 aircraft. The document details the scheduled delivery of parts such as hydraulic actuators, brake systems, and navigation units over the next quarter. These supplies are essential for routine maintenance and ensuring that the aircraft remains in operational condition for non-combat purposes. The report also includes internal memos on supplier performance and cost negotiations, which are classified as Restricted to prevent unauthorized access and ensure smooth contract execution. Highlights of the Competition\nThe athletics events were among the most anticipated, with the fastest runners from each branch competing for medals. The 100m sprint final featured a thrilling race between the top sprinters from the Army, Navy, and Air Force, with Captain Aaryan Verma of the Army securing the gold medal with a time of 10.87 seconds, followed by Lieutenant Neha Mehra of the Navy, who claimed the silver with 11.03 seconds. In the football tournament, the Indian Army emerged as the champions after a tense final match against the Indian Air Force. The game ended with a score of 2-1 in favor of the Army, with Subedar Major Vikram Singh scoring the winning goal in the final minutes. The Army team displayed exceptional teamwork and strategic play, which ultimately led them to victory. The cricket matches were highly competitive, with the Indian Navy defeating the Air Force team in a closely contested T20 match. The final was a nail-biting affair, with Navyâ\x80\x99s Lieutenant Commander Rahul Mehta hitting the winning six in the last over of the game.'
  • 'Supply Chain and Procurement Documents\nRoutine procurement activities continue to fuel military preparedness. The most recent batch of documents contains procurement orders for various operational materials needed in peripheral zones. These orders range from vehicles used in reconnaissance missions to tactical gear for military units that are not directly involved in combat but are still crucial for maintaining defense capabilities. For example, a recent procurement request was made for a series of high-powered satellite phones that will be issued to units deployed in isolated locations. These phones are essential for ensuring that communication lines remain open in areas where traditional communication infrastructure is unavailable. Similarly, there are ongoing negotiations for acquiring medical supplies, such as portable surgical kits and trauma care equipment, specifically for units working in non-conflict zones where medical infrastructure might be limited. The documentation detailing these procurements includes specifics on supplier agreements, delivery schedules, and operational requirements. This is sensitive data, as it could potentially reveal gaps in military supply chains if accessed by unauthorized individuals. Suppliers are carefully vetted, and any leak of information regarding these supply chains could jeopardize the mission's success in certain strategic areas. Cipher Message: Cipher Text: "NQ5P7 QXZ8T 7J6B2 P1M9Y." â\x80\x93 Encrypted procurement details, listing authorized suppliers and material quantities for internal distribution only.'
0
  • 'Enhancement of Aerial Surveillance\nUnmanned Aerial Vehicles (UAVs) have been deployed from the Jorhat Air Force Station to maintain constant surveillance over disputed areas. These UAVs, equipped with high-resolution cameras and thermal sensors, provide real-time imagery of adversarial activities. Regular patrol missions conducted over regions like Kibithu and Walong have been instrumental in identifying unauthorized constructions. The data gathered is relayed to command centers in Shillong for detailed analysis. AI-powered algorithms help in detecting anomalies, ensuring swift decision-making in case of any potential threats. These proactive measures have significantly improved situational awareness. Implementation of AI-Based Border Surveillance\nThe recent deployment of artificial intelligence-driven surveillance mechanisms has introduced cutting-edge technology into border operations. Surveillance drones and AI-powered detection sensors have been positioned along key border regions, including the North Eastern States and the Indian-Pakistani border. These assets are leveraging machine learning algorithms to identify patterns of unusual activity, unauthorized crossings, and changes in terrain anomalies. The AI systems are capable of processing vast quantities of real-time data collected from UAVs, thermal imaging cameras, and ground-based radar installations. Machine learning analysis identifies trends that may go unnoticed by conventional monitoring, such as small troop movements or unauthorized infiltration attempts across the porous Indo-Bangladesh border. These capabilities have already proven effective in detecting early signs of infiltration and cross-border activity. Additionally, intelligence teams are collaborating with AI experts to fine-tune these tools for real- time decision-making support. Advanced signal detection and image recognition capabilities are improving response times and ensuring that border patrols can intercept threats with enhanced accuracy and minimal delay.'
  • 'Conclusion\nThe integration of advanced technology with strategic realignments across operational zones highlights the dynamic and robust approach adopted by the armed forces. These measures not only bolster defensive capabilities but also reinforce the nationâ\x80\x99s readiness to respond to evolving threats.'
  • 'New Munitions Deployment\nTo enhance combat effectiveness, advanced munitions tailored for specific operational conditions have been introduced. The recent deployment of guided mortar systems to units stationed in the Siachen Glacier highlights this focus. These munitions, tested under extreme conditions, provide unmatched accuracy and reliability. Additionally, countermeasure systems designed to neutralize enemy drones have been distributed across critical sectors. These systems employ directed energy technology, effectively disrupting the electronic controls of hostile UAVs.'

Evaluation

Metrics

Label Accuracy
all 0.9194

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("chandanzeon/setfit_finetuned_iaf_98")
# Run inference
preds = model("Tactical Coordination and Training
Joint training exercises involving armored and artillery units have been conducted to refine battlefield tactics. These exercises, held in the Thar Desert, simulated multi-front conflict scenarios, emphasizing coordination between various branches of the armed forces. Feedback from these exercises has led to the adoption of new operational guidelines, such as optimized deployment patterns for tanks and artillery systems. Post-exercise debriefings at Jodhpur Cantonment highlighted the importance of synchronized maneuvers in achieving tactical superiority.")

Training Details

Training Set Metrics

Training set Min Median Max
Word count 39 130.3317 475
Label Training Sample Count
0 49
1 56
2 49
3 51

Training Hyperparameters

  • batch_size: (32, 32)
  • num_epochs: (5, 5)
  • max_steps: -1
  • sampling_strategy: oversampling
  • body_learning_rate: (2e-05, 1e-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: True

Training Results

Epoch Step Training Loss Validation Loss
0.0010 1 0.267 -
0.0508 50 0.2533 -
0.1016 100 0.2342 -
0.1524 150 0.2272 -
0.2033 200 0.2065 -
0.2541 250 0.1573 -
0.3049 300 0.1051 -
0.3557 350 0.0546 -
0.4065 400 0.011 -
0.4573 450 0.004 -
0.5081 500 0.0028 -
0.5589 550 0.0023 -
0.6098 600 0.0019 -
0.6606 650 0.0015 -
0.7114 700 0.0014 -
0.7622 750 0.0014 -
0.8130 800 0.0013 -
0.8638 850 0.0012 -
0.9146 900 0.0011 -
0.9654 950 0.001 -
1.0 984 - 0.0731
1.0163 1000 0.001 -
1.0671 1050 0.0009 -
1.1179 1100 0.0009 -
1.1687 1150 0.0008 -
1.2195 1200 0.0008 -
1.2703 1250 0.0008 -
1.3211 1300 0.0008 -
1.3720 1350 0.0007 -
1.4228 1400 0.0007 -
1.4736 1450 0.0007 -
1.5244 1500 0.0007 -
1.5752 1550 0.0006 -
1.6260 1600 0.0006 -
1.6768 1650 0.0006 -
1.7276 1700 0.0006 -
1.7785 1750 0.0006 -
1.8293 1800 0.0006 -
1.8801 1850 0.0006 -
1.9309 1900 0.0006 -
1.9817 1950 0.0005 -
2.0 1968 - 0.0762
2.0325 2000 0.0005 -
2.0833 2050 0.0005 -
2.1341 2100 0.0005 -
2.1850 2150 0.0005 -
2.2358 2200 0.0005 -
2.2866 2250 0.0005 -
2.3374 2300 0.0005 -
2.3882 2350 0.0005 -
2.4390 2400 0.0005 -
2.4898 2450 0.0005 -
2.5407 2500 0.0005 -
2.5915 2550 0.0004 -
2.6423 2600 0.0004 -
2.6931 2650 0.0004 -
2.7439 2700 0.0004 -
2.7947 2750 0.0004 -
2.8455 2800 0.0004 -
2.8963 2850 0.0004 -
2.9472 2900 0.0004 -
2.9980 2950 0.0004 -
3.0 2952 - 0.0786
3.0488 3000 0.0004 -
3.0996 3050 0.0004 -
3.1504 3100 0.0004 -
3.2012 3150 0.0004 -
3.2520 3200 0.0004 -
3.3028 3250 0.0004 -
3.3537 3300 0.0004 -
3.4045 3350 0.0004 -
3.4553 3400 0.0004 -
3.5061 3450 0.0004 -
3.5569 3500 0.0003 -
3.6077 3550 0.0004 -
3.6585 3600 0.0004 -
3.7093 3650 0.0004 -
3.7602 3700 0.0003 -
3.8110 3750 0.0003 -
3.8618 3800 0.0004 -
3.9126 3850 0.0003 -
3.9634 3900 0.0003 -
4.0 3936 - 0.0813
4.0142 3950 0.0003 -
4.0650 4000 0.0003 -
4.1159 4050 0.0003 -
4.1667 4100 0.0003 -
4.2175 4150 0.0003 -
4.2683 4200 0.0003 -
4.3191 4250 0.0003 -
4.3699 4300 0.0003 -
4.4207 4350 0.0003 -
4.4715 4400 0.0003 -
4.5224 4450 0.0003 -
4.5732 4500 0.0003 -
4.6240 4550 0.0003 -
4.6748 4600 0.0003 -
4.7256 4650 0.0003 -
4.7764 4700 0.0003 -
4.8272 4750 0.0003 -
4.8780 4800 0.0003 -
4.9289 4850 0.0003 -
4.9797 4900 0.0003 -
5.0 4920 - 0.0804

Framework Versions

  • Python: 3.10.12
  • SetFit: 1.1.0
  • Sentence Transformers: 3.2.1
  • Transformers: 4.42.2
  • PyTorch: 2.5.1+cu121
  • Datasets: 3.2.0
  • 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}
}
Downloads last month
19
Safetensors
Model size
33.4M params
Tensor type
F32
·
Inference Examples
Text Classification
This model does not have enough activity to be deployed to Inference API (serverless) yet. Increase its social visibility and check back later, or deploy to Inference Endpoints (dedicated) instead.

Model tree for chandanzeon/setfit_finetuned_iaf_98

Base model

BAAI/bge-small-en-v1.5
Finetuned
(139)
this model

Evaluation results

View on Papers With Code