Hugging Face's logo

kasparas12
/
crypto_individual_infer_model_setfit

Text Classification
setfit
PyTorch
sentence-transformers
bert
generated_from_setfit_trainer
Eval Results
Model card Files Community
Edit model card

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
UNDETERMINED
  • 'Professor Emeritus of Cognitive Sciences at the University of California Irvine Research Visual perception evolutionary psychology consciousness AI Irvine CA'
  • 'Emeritus Professor of War Studies Kings College London just published Command The Politics of Military Operations from Korea to Ukraine UK Penguin US OUP '
  • 'XML apologist Erlang enthusiast Currently JVMs Performance stuff at Netflix Previously JVMs performative stuff at Twitter Hehim San Francisco California'
NFT_ARTIST
  • 'Artist Web3 Marketing Advisor Educator Making history everyday Trapped in the blockchain'
  • 'OwnYourAssets TokenGatedFile Access For CrossPlatformInteroperableGaming C5isComing CYBΞRVΞRSΞ'
  • 'Pronounced Akossya artist Zurich'
ONCHAIN_ANALYST
  • 'I write about onchain stuff fixer AleoHQ prev rabbithole_gg and plenty of DAOs youve heard of '
  • 'cofounder 3pochLabs onchain'
  • 'onchain data farcer building mosaicdrops media CryptoSapiens_ OntologyNetwork OrangeProtocol banklessDAO s0 _buildspace s4 Mosaicverse'
BUSINESS_DEVELOPER
  • 'Prev opensea TheBlock__ amazon '
  • 'Building HxroNetwork variable'
  • 'Building something old CoFounder alongsidefi '
NFT_COLLECTOR
  • 'Building glitchmarfa Collecting brightopps prev brtmoments '
  • 'My soul is a cat My two children rpcnftclub ChainFeedsxyz Bangkok'
  • 'prev OpenSea NYC'
DEVELOPER
  • 'Architect DoraHacks DoraFactory The everlasting hacker movement Menlo Park'
  • 'Engineer at Inria scikitlearn developer supported by Python and Machine Learning Between Vannes Paris France'
  • 'Working paritytech on substrate Views are my own I working mostly with rustlang nowadays '
TRADER
  • 'Applied game theorist blog occasionally at formerly not a very serious person Scott Alexander '
  • 'Crypto Trading Bitcoin class of 2013 insilicotrading COO Banana Cabana'
  • 'token maxi '
COMMUNITY_MANAGER
  • 'chutzpah controlled chaos connoisseur arbitrum chinshilling chinchillin thoughts are my own Rio de Janeiro Brazil'
  • 'commonsstack CoFounder tecmns Founding Steward KERNEL0x KB5 trustedseed tamaralens '
  • 'Community Admin at The Arbitrum Foundation Helping to scale Ethereum at Arbitrum Feed KOL Binance WEB3'
SECURITY_AUDITOR
  • 'founder adjacentfi cofounder former auditor osec_io MEV on solana '
  • 'Security Researcher Googles Threat Analysis Group 0days all day Love all things bytes assembly and glitter sheher '
  • '採用マーケ得意仮想通貨エンジニア4社1社ホワイトハッカーとして月110万達成現在歯科衛生士の妻と事業開始 実績年商1億超えのマーケ担当 開始5ヶ月で6名見学開始2年で累計DH11名見学6名採用 ハイライト要チェック ブログに今までの有益投稿をまとめました 岩手長野福岡ドバイ沖縄'
VENTURE_CAPITALIST
  • 'Liquid Crypto Brevan Howard Prev dragonfly_xyz consensys Arena'
  • 'maverick LA'
  • 'Founder of SavvyBooks Degen dcv_capital Summoner ElasticDAO metafam Judge code4rena Contributor CantoPublic Nomadic'
INVESTOR
  • 'Crypto Investor at Tephra Digital Ex Head of Research Grayscale DCGco FMR Head of Digital Asset Strategy Fundstrat New York NY'
  • 'Capital Allocators New York NY'
  • 'Director of Research Autonomous Technology Robotics ARKinvest Automation robotics energy storage alternative energy and space Disclosure New York NY'
ANGEL_INVESTOR
  • 'larp LawliettesLab angel uvocapital '
  • 'Initiator inverternetwork I Angel Investor I ex Gitcoin '
  • 'VP Head of BD AleoHQ Mainnet Launch Soon Strategic Advisor VoxiesNFT Angel Investor rcsdao ExOP ExCoinbase Professionally CuriousOpinions My Own Manhattan NY'
EXECUTIVE
  • 'Chief Strategy Marketing Officer of Liquidity Group Im also the cofounder of Hudson Rock RockHudsonRock a cybercrime intelligence company TelAviv'
  • 'CEO Polymarket Ethereum since 14 I love music and collect art new york'
  • 'CEO StartaleHQ Founder AstarNetwork All things for Web3 for billions Japanese Sota_Web3 Earth'
MARKETER
  • 'Director General en Kayum comparador de seguros insurance PPC tech crypto f1 Mexico City Mexico'
  • 'Insights about Web3 data economy and AI by oceanprotocol Currently in Marcom oceanprotocol ocean Ocean '
  • 'f加速 ethereum China internet culture history podcast growth marketing realmasknetwork prev newsbreakapp smartnews Zuzalu human Palo Alto USA'
DATA_SCIENTIST
  • 'data uniswap prev theTIEIO go bears New York NY'
  • 'engineering data science a16zcrypto '
  • 'LangChainAI previously robusthq kensho MLOps Generative AI sports analytics '
EDUCATOR
  • ' London'
  • 'MSc Immunology student Past cofounder prof director USF Center Applied Data Ethics math PhD math_rachelmastodonsocial sheher Brisbane Australia'
  • 'Here to build shared intelligence listen learn share via community tokenengineering KERNEL0x OptimismGov publicgoods education valuesmatter CyberDyn0x tauranga teikaamaui'
INFLUENCER
  • 'the destroyer Titan'
  • 'Healthy life style healthier bags Cape Town South Africa'
  • 'Beauty Brains Bitcoin Beauty in an anonymous world'
ADVISOR
  • 'A decentralized onchain governance consultant Health Wealth RunItUp The only Alpha discord youll ever need to joingametheoryweb3 squanchland Profit Land'
  • 'Design director Startup Advisor Midjourney Sharing learnings and prompts In my free time working on offscreenai Vancouver Canada'
  • 'I help fix and grow crypto portfolios through premium research and strategies 1000 members Founder cshift_io Podcast benandbergs Join 10k Crypto Investors '
BLOGGER
  • 'NOW Editor Forbes Writer Stripe HarvardBiz Back on Twitter after ignoring it for a decade I will try my best London'
  • 'larp coindesk '
  • ' '
RESEARCHER
  • 'Roblox Chief Scientist UWaterloo McGill Prof morgan3dbsky Known for NVIDIA Unity Graphics Codex Markdeep G3D Skylanders E Ink Titan Quest Williams Ontario Canada'
  • 'Simple human Simple life I am trying to do good around me Empathy creativity inspiration ArigatōMerci For ever apprenti researcher Nulle part ailleurs Nowhere'
  • 'Research community And we have our own NFT collection Telegram'
METAVERSE_ENTHUSIAST
  • 'fluent speaker of http and color virtual world evangelist game developer painter writer cj5 driver San Diego'
  • 'Blockchain Gaming Evangelist CritTheory Gaming CoFounder Earth'
  • 'We are a peeple obsessed recruiting service collective Treating everyone like a DMs checked infrequently Metaverse'
NODE_OPERATOR
  • 'into protocools and shitposting at nodeguardians '
  • ' CoFounder of onivalidator Filmmaker People Maxi Los Angeles CA'
  • 'I attest to block 247 Hobby involves the occasional block proposal Have commercial agreements with the MEV trade association Members of Sync Committees Los Angeles'
LAWYER
  • 'Law professor at Cal BerkeleyLaw Berkeley California'
  • 'IP litigator first sale doctrine respecter schedule a disrespecter wife mom to the tiny boss likes design patents needlework yarn new hampshire'
  • 'Lawyer FINTConsulting TechPolicy E4EProject upcoming GRC CybersecurityAnalyst ex InstituteGC Tweet law tech policy GRC Cybersecurity Decentralized'
DATA_ANALYST
  • 'Llama pilot at and '
  • 'blockchain data opensea kqian on Dune my views are my own dyor nfa data only wagmi open sea'
  • 'Blockchain analyst Cat and dog dad Taylor Swift fan Army veteran Pittsburgh PA'
MINER
  • 'Blockchain bitcoin mining since 2011 analyst 35 years in IT UnixNetwork engineer fpgachip design exCIO Bitfury BitfuryGroup LNSegWit taproot California USA'
  • 'Founder and CEO of Austin TX'
  • '在币圈捡矿泉水瓶子的人 0xb38544ccf295d78b7ae7b2bae5dbebdb1f09910dcrossbell Member of 33daoweb3 Metaverse'
SHITCOINER
  • 'Degen ETH and SOL lover '
  • 'VMPX mrjacklevin Draculaborg'
  • 'gripto alt notapornfolder_ '
FINANCIAL_ANALYST
  • 'Enrolled Agent Crypto Enthusiast Tax EXPERT StackingSats Chopping Tax Since 2016 NoSatoshiLeftBehind hodlmore payless crypto taxes Longmont CO'
  • 'Politico financial services editor zwarmbrodtpoliticocom zacharywarmbrodtprotonmailcom Washington DC'
  • 'Im just lookin for clues at the scene of the crime Sedona Arizona'
BUSINESS_ANALYST
  • 'Biz Analyst by day web3crypto learner by nightweekend Optimistic about Crypto FanVajpayeeji NaMo M Andreessen E Musk C Dixon Balaji S web3SF Bay Area'

Evaluation

Metrics

Label Accuracy
all 0.5565

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("kasparas12/crypto_individual_infer_model_setfit")
# Run inference
preds = model("producer business and elsewhere  on leave  views my own la gran manzana")

Training Details

Training Set Metrics

Training set Min Median Max
Word count 1 13.3415 65
Label Training Sample Count
DEVELOPER 2111
DATA_SCIENTIST 93
DATA_ANALYST 25
NODE_OPERATOR 71
MINER 47
SECURITY_AUDITOR 352
INVESTOR 484
ANGEL_INVESTOR 160
VENTURE_CAPITALIST 941
TRADER 270
SHITCOINER 88
BUSINESS_DEVELOPER 917
BUSINESS_ANALYST 1
COMMUNITY_MANAGER 401
MARKETER 190
FINANCIAL_ANALYST 72
ADVISOR 150
RESEARCHER 691
ONCHAIN_ANALYST 45
EXECUTIVE 741
INFLUENCER 834
LAWYER 137
BLOGGER 198
NFT_COLLECTOR 335
NFT_ARTIST 598
EDUCATOR 281
METAVERSE_ENTHUSIAST 132
UNDETERMINED 2216

Training Hyperparameters

  • batch_size: (64, 64)
  • num_epochs: (1, 1)
  • max_steps: -1
  • sampling_strategy: oversampling
  • num_iterations: 20
  • 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
  • seed: 42
  • eval_max_steps: -1
  • load_best_model_at_end: False

Training Results

Epoch Step Training Loss Validation Loss
0.0001 1 0.2625 -
0.0064 50 0.2677 -
0.0127 100 0.2515 -
0.0191 150 0.2413 -
0.0254 200 0.2374 -
0.0318 250 0.2383 -
0.0381 300 0.222 -
0.0445 350 0.1972 -
0.0509 400 0.2268 -
0.0572 450 0.2333 -
0.0636 500 0.199 -
0.0699 550 0.2035 -
0.0763 600 0.1676 -
0.0827 650 0.1566 -
0.0890 700 0.1909 -
0.0954 750 0.189 -
0.1017 800 0.1872 -
0.1081 850 0.1576 -
0.1144 900 0.1382 -
0.1208 950 0.1603 -
0.1272 1000 0.155 -
0.1335 1050 0.1764 -
0.1399 1100 0.1506 -
0.1462 1150 0.1439 -
0.1526 1200 0.1581 -
0.1590 1250 0.1494 -
0.1653 1300 0.1622 -
0.1717 1350 0.1503 -
0.1780 1400 0.1094 -
0.1844 1450 0.1576 -
0.1907 1500 0.1194 -
0.1971 1550 0.1515 -
0.2035 1600 0.1662 -
0.2098 1650 0.1642 -
0.2162 1700 0.0943 -
0.2225 1750 0.1472 -
0.2289 1800 0.1622 -
0.2352 1850 0.0809 -
0.2416 1900 0.1623 -
0.2480 1950 0.1444 -
0.2543 2000 0.1304 -
0.2607 2050 0.1175 -
0.2670 2100 0.078 -
0.2734 2150 0.1189 -
0.2798 2200 0.141 -
0.2861 2250 0.1233 -
0.2925 2300 0.1446 -
0.2988 2350 0.1076 -
0.3052 2400 0.1016 -
0.3115 2450 0.0818 -
0.3179 2500 0.1384 -
0.3243 2550 0.1065 -
0.3306 2600 0.1029 -
0.3370 2650 0.1227 -
0.3433 2700 0.0982 -
0.3497 2750 0.0959 -
0.3561 2800 0.0851 -
0.3624 2850 0.1028 -
0.3688 2900 0.1136 -
0.3751 2950 0.1111 -
0.3815 3000 0.115 -
0.3878 3050 0.1183 -
0.3942 3100 0.0689 -
0.4006 3150 0.1004 -
0.4069 3200 0.1079 -
0.4133 3250 0.112 -
0.4196 3300 0.0758 -
0.4260 3350 0.09 -
0.4323 3400 0.1267 -
0.4387 3450 0.1024 -
0.4451 3500 0.1352 -
0.4514 3550 0.0681 -
0.4578 3600 0.0483 -
0.4641 3650 0.0937 -
0.4705 3700 0.0744 -
0.4769 3750 0.0926 -
0.4832 3800 0.0764 -
0.4896 3850 0.0814 -
0.4959 3900 0.108 -
0.5023 3950 0.0936 -
0.5086 4000 0.0687 -
0.5150 4050 0.0607 -
0.5214 4100 0.0829 -
0.5277 4150 0.0772 -
0.5341 4200 0.0309 -
0.5404 4250 0.0797 -
0.5468 4300 0.063 -
0.5532 4350 0.071 -
0.5595 4400 0.0667 -
0.5659 4450 0.121 -
0.5722 4500 0.0565 -
0.5786 4550 0.0915 -
0.5849 4600 0.0613 -
0.5913 4650 0.0479 -
0.5977 4700 0.0622 -
0.6040 4750 0.0687 -
0.6104 4800 0.0635 -
0.6167 4850 0.1233 -
0.6231 4900 0.0351 -
0.6295 4950 0.0717 -
0.6358 5000 0.0906 -
0.6422 5050 0.0712 -
0.6485 5100 0.1133 -
0.6549 5150 0.0757 -
0.6612 5200 0.0809 -
0.6676 5250 0.112 -
0.6740 5300 0.0893 -
0.6803 5350 0.0591 -
0.6867 5400 0.0872 -
0.6930 5450 0.0937 -
0.6994 5500 0.038 -
0.7057 5550 0.0793 -
0.7121 5600 0.0569 -
0.7185 5650 0.0861 -
0.7248 5700 0.1022 -
0.7312 5750 0.0759 -
0.7375 5800 0.0451 -
0.7439 5850 0.08 -
0.7503 5900 0.058 -
0.7566 5950 0.0423 -
0.7630 6000 0.043 -
0.7693 6050 0.109 -
0.7757 6100 0.072 -
0.7820 6150 0.0342 -
0.7884 6200 0.0833 -
0.7948 6250 0.0643 -
0.8011 6300 0.1069 -
0.8075 6350 0.0713 -
0.8138 6400 0.0807 -
0.8202 6450 0.0518 -
0.8266 6500 0.0796 -
0.8329 6550 0.0954 -
0.8393 6600 0.0709 -
0.8456 6650 0.0541 -
0.8520 6700 0.0503 -
0.8583 6750 0.0737 -
0.8647 6800 0.0931 -
0.8711 6850 0.0636 -
0.8774 6900 0.0579 -
0.8838 6950 0.1168 -
0.8901 7000 0.0751 -
0.8965 7050 0.0945 -
0.9028 7100 0.0396 -
0.9092 7150 0.0623 -
0.9156 7200 0.0641 -
0.9219 7250 0.0697 -
0.9283 7300 0.0675 -
0.9346 7350 0.0544 -
0.9410 7400 0.0803 -
0.9474 7450 0.0549 -
0.9537 7500 0.0612 -
0.9601 7550 0.0721 -
0.9664 7600 0.0692 -
0.9728 7650 0.07 -
0.9791 7700 0.0476 -
0.9855 7750 0.0673 -
0.9919 7800 0.0606 -
0.9982 7850 0.1001 -

Framework Versions

  • Python: 3.9.16
  • SetFit: 1.0.3
  • Sentence Transformers: 2.2.2
  • Transformers: 4.21.3
  • PyTorch: 1.12.1+cu116
  • Datasets: 2.4.0
  • Tokenizers: 0.12.1

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
1,065

Finetuned from

Evaluation results

View on Papers With Code