Hugging Face's logo

cbpuschmann
/
paraphrase-multilingual-mpnet-klimacoder_v0.8

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

SetFit with sentence-transformers/paraphrase-multilingual-mpnet-base-v2

This is a SetFit model that can be used for Text Classification. This SetFit model uses sentence-transformers/paraphrase-multilingual-mpnet-base-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 Sources

Model Labels

Label Examples
neutral
  • 'Die Bundesregierung plant, bis 2024 ein sogenanntes Heizungsgesetz vorzulegen, das unter anderem eine flächendeckende Nutzung von Wärmepumpen als Teil eines umfassenden Plans zur Reduzierung der Treibhausgasemissionen im Gebäudesektor vorsehen soll.'
  • '"Die Bundesregierung plant, die Einführung von Wärmepumpen für Neubauten und den Austausch alter Heizungsanlagen in Bestandsgebäuden durch ein Gesetz zu forcieren, während Kritiker warnen, dass die Maßnahmen die Belastung für private Haushalte und Unternehmen erhöhen könnten."'
  • ' Die Diskussion über ein nationales Tempolimit auf Autobahnen spaltet die Gemüter, während Experten die potenziellen Vorteile und Nachteile abwägen.'
opposed
  • '"Millionen von Hausbesitzern sollen zu unfreiwilligen Versuchskaninchen für die teuren und unzuverlässlichen Wärmepumpen werden, ohne dass es auch nur einen Hauch von echter Wahlmöglichkeit gibt."'
  • '"Die von den Grünen und Linken geträumte Tempodiktatur auf unseren Autobahnen ist nichts als ein weiterer Schritt in Richtung einer überbürokratisierten, unfreien Gesellschaft."'
  • '"Die geplanten Vorschriften würden vielen Familien den Traum vom Eigenheim in weite Ferne rücken, da die Kosten für die Installation einer Wärmepumpe oft ein Vielfaches dessen betragen, was ein durchschnittlicher Haushalt in einem Jahr für Heizkosten ausgibt."'
supportive
  • 'Die Bundesregierung hat mit dem Heizungsgesetz einen wichtigen Schritt in Richtung Klimaneutralität gemacht, indem sie die Verpflichtung zur Nutzung erneuerbarer Wärmequellen bei Neubauten festlegt.'
  • '"Ein Tempolimit auf Autobahnen könnte nicht nur die Umweltbelastung verringern, sondern auch die Zahl der Verkehrsunfälle reduzieren und somit Menschenleben retten."'
  • ' Eine nationale Geschwindigkeitsbegrenzung auf Autobahnen könnte nicht nur die Unfallzahlen senken, sondern auch einen wichtigen Beitrag zum Klimaschutz leisten.'

Evaluation

Metrics

Label Accuracy
all 0.9534

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("cbpuschmann/paraphrase-multilingual-mpnet-klimacoder_v0.8")
# Run inference
preds = model(" \"Das Heizungsgesetz ist nichts weiter als ein weiterer Schritt in Richtung eines grünen Diktats, das die Bürger in die Kälte schickt.\"")

Training Details

Training Set Metrics

Training set Min Median Max
Word count 10 25.6541 57
Label Training Sample Count
neutral 321
opposed 391
supportive 404

Training Hyperparameters

  • batch_size: (32, 32)
  • num_epochs: (1, 1)
  • 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: False

Training Results

Epoch Step Training Loss Validation Loss
0.0000 1 0.1985 -
0.0019 50 0.2445 -
0.0039 100 0.2321 -
0.0058 150 0.2012 -
0.0077 200 0.1614 -
0.0097 250 0.1188 -
0.0116 300 0.0849 -
0.0136 350 0.0563 -
0.0155 400 0.0374 -
0.0174 450 0.0216 -
0.0194 500 0.0144 -
0.0213 550 0.0099 -
0.0232 600 0.0061 -
0.0252 650 0.007 -
0.0271 700 0.0026 -
0.0290 750 0.0017 -
0.0310 800 0.0012 -
0.0329 850 0.0014 -
0.0349 900 0.002 -
0.0368 950 0.0008 -
0.0387 1000 0.0009 -
0.0407 1050 0.0003 -
0.0426 1100 0.0007 -
0.0445 1150 0.0008 -
0.0465 1200 0.0006 -
0.0484 1250 0.0002 -
0.0503 1300 0.0001 -
0.0523 1350 0.0001 -
0.0542 1400 0.0001 -
0.0562 1450 0.0001 -
0.0581 1500 0.0007 -
0.0600 1550 0.0005 -
0.0620 1600 0.0007 -
0.0639 1650 0.0012 -
0.0658 1700 0.0007 -
0.0678 1750 0.0038 -
0.0697 1800 0.0018 -
0.0716 1850 0.0049 -
0.0736 1900 0.0061 -
0.0755 1950 0.0038 -
0.0775 2000 0.0037 -
0.0794 2050 0.0006 -
0.0813 2100 0.0001 -
0.0833 2150 0.0 -
0.0852 2200 0.0 -
0.0871 2250 0.0 -
0.0891 2300 0.0 -
0.0910 2350 0.0 -
0.0929 2400 0.0 -
0.0949 2450 0.0 -
0.0968 2500 0.0 -
0.0987 2550 0.0 -
0.1007 2600 0.0 -
0.1026 2650 0.0 -
0.1046 2700 0.0 -
0.1065 2750 0.0 -
0.1084 2800 0.0 -
0.1104 2850 0.0 -
0.1123 2900 0.0 -
0.1142 2950 0.0 -
0.1162 3000 0.0 -
0.1181 3050 0.0 -
0.1200 3100 0.0 -
0.1220 3150 0.0 -
0.1239 3200 0.0 -
0.1259 3250 0.0 -
0.1278 3300 0.0 -
0.1297 3350 0.0 -
0.1317 3400 0.0 -
0.1336 3450 0.0 -
0.1355 3500 0.0 -
0.1375 3550 0.0 -
0.1394 3600 0.0 -
0.1413 3650 0.0 -
0.1433 3700 0.0 -
0.1452 3750 0.0 -
0.1472 3800 0.0 -
0.1491 3850 0.0 -
0.1510 3900 0.0 -
0.1530 3950 0.0 -
0.1549 4000 0.0 -
0.1568 4050 0.0 -
0.1588 4100 0.0 -
0.1607 4150 0.0 -
0.1626 4200 0.0 -
0.1646 4250 0.0 -
0.1665 4300 0.0 -
0.1685 4350 0.0 -
0.1704 4400 0.0 -
0.1723 4450 0.0 -
0.1743 4500 0.0 -
0.1762 4550 0.0 -
0.1781 4600 0.0 -
0.1801 4650 0.0 -
0.1820 4700 0.0 -
0.1839 4750 0.0 -
0.1859 4800 0.0 -
0.1878 4850 0.0 -
0.1898 4900 0.0 -
0.1917 4950 0.0 -
0.1936 5000 0.0 -
0.1956 5050 0.0 -
0.1975 5100 0.0 -
0.1994 5150 0.0 -
0.2014 5200 0.0 -
0.2033 5250 0.0 -
0.2052 5300 0.0 -
0.2072 5350 0.0 -
0.2091 5400 0.0 -
0.2111 5450 0.0 -
0.2130 5500 0.0 -
0.2149 5550 0.0 -
0.2169 5600 0.0 -
0.2188 5650 0.0 -
0.2207 5700 0.0 -
0.2227 5750 0.0 -
0.2246 5800 0.0 -
0.2265 5850 0.0 -
0.2285 5900 0.0 -
0.2304 5950 0.0 -
0.2324 6000 0.0 -
0.2343 6050 0.0 -
0.2362 6100 0.0 -
0.2382 6150 0.0 -
0.2401 6200 0.0 -
0.2420 6250 0.0 -
0.2440 6300 0.0 -
0.2459 6350 0.0 -
0.2478 6400 0.0 -
0.2498 6450 0.0 -
0.2517 6500 0.0 -
0.2536 6550 0.0 -
0.2556 6600 0.0 -
0.2575 6650 0.0 -
0.2595 6700 0.0 -
0.2614 6750 0.0 -
0.2633 6800 0.0 -
0.2653 6850 0.0 -
0.2672 6900 0.0 -
0.2691 6950 0.0 -
0.2711 7000 0.0 -
0.2730 7050 0.0 -
0.2749 7100 0.0 -
0.2769 7150 0.0 -
0.2788 7200 0.0 -
0.2808 7250 0.0 -
0.2827 7300 0.0 -
0.2846 7350 0.0 -
0.2866 7400 0.0 -
0.2885 7450 0.0 -
0.2904 7500 0.0 -
0.2924 7550 0.0 -
0.2943 7600 0.0 -
0.2962 7650 0.0 -
0.2982 7700 0.0 -
0.3001 7750 0.0 -
0.3021 7800 0.0 -
0.3040 7850 0.0 -
0.3059 7900 0.0 -
0.3079 7950 0.0 -
0.3098 8000 0.0 -
0.3117 8050 0.0 -
0.3137 8100 0.0 -
0.3156 8150 0.0 -
0.3175 8200 0.0 -
0.3195 8250 0.0 -
0.3214 8300 0.0 -
0.3234 8350 0.0 -
0.3253 8400 0.0 -
0.3272 8450 0.0 -
0.3292 8500 0.0 -
0.3311 8550 0.0 -
0.3330 8600 0.0 -
0.3350 8650 0.0 -
0.3369 8700 0.0 -
0.3388 8750 0.0 -
0.3408 8800 0.0 -
0.3427 8850 0.0 -
0.3447 8900 0.0 -
0.3466 8950 0.0 -
0.3485 9000 0.0 -
0.3505 9050 0.0 -
0.3524 9100 0.0 -
0.3543 9150 0.0 -
0.3563 9200 0.0 -
0.3582 9250 0.0 -
0.3601 9300 0.0 -
0.3621 9350 0.0 -
0.3640 9400 0.0 -
0.3660 9450 0.0 -
0.3679 9500 0.0 -
0.3698 9550 0.0 -
0.3718 9600 0.0 -
0.3737 9650 0.0 -
0.3756 9700 0.0 -
0.3776 9750 0.0 -
0.3795 9800 0.0 -
0.3814 9850 0.0 -
0.3834 9900 0.0 -
0.3853 9950 0.0 -
0.3873 10000 0.0 -
0.3892 10050 0.0 -
0.3911 10100 0.0 -
0.3931 10150 0.0 -
0.3950 10200 0.0 -
0.3969 10250 0.0 -
0.3989 10300 0.0 -
0.4008 10350 0.0 -
0.4027 10400 0.0 -
0.4047 10450 0.0 -
0.4066 10500 0.0 -
0.4086 10550 0.0 -
0.4105 10600 0.0 -
0.4124 10650 0.0 -
0.4144 10700 0.0 -
0.4163 10750 0.0 -
0.4182 10800 0.0 -
0.4202 10850 0.0 -
0.4221 10900 0.0 -
0.4240 10950 0.0 -
0.4260 11000 0.0 -
0.4279 11050 0.0 -
0.4298 11100 0.0 -
0.4318 11150 0.0 -
0.4337 11200 0.0 -
0.4357 11250 0.0 -
0.4376 11300 0.0 -
0.4395 11350 0.0 -
0.4415 11400 0.0 -
0.4434 11450 0.0 -
0.4453 11500 0.0 -
0.4473 11550 0.0 -
0.4492 11600 0.0 -
0.4511 11650 0.0 -
0.4531 11700 0.0 -
0.4550 11750 0.0 -
0.4570 11800 0.0 -
0.4589 11850 0.0109 -
0.4608 11900 0.0218 -
0.4628 11950 0.0073 -
0.4647 12000 0.0056 -
0.4666 12050 0.0037 -
0.4686 12100 0.0011 -
0.4705 12150 0.0002 -
0.4724 12200 0.0014 -
0.4744 12250 0.0031 -
0.4763 12300 0.0013 -
0.4783 12350 0.0012 -
0.4802 12400 0.0022 -
0.4821 12450 0.0003 -
0.4841 12500 0.0 -
0.4860 12550 0.0 -
0.4879 12600 0.0 -
0.4899 12650 0.0 -
0.4918 12700 0.0 -
0.4937 12750 0.0 -
0.4957 12800 0.0 -
0.4976 12850 0.0 -
0.4996 12900 0.0 -
0.5015 12950 0.0 -
0.5034 13000 0.0 -
0.5054 13050 0.0 -
0.5073 13100 0.0 -
0.5092 13150 0.0 -
0.5112 13200 0.0 -
0.5131 13250 0.0 -
0.5150 13300 0.0 -
0.5170 13350 0.0 -
0.5189 13400 0.0 -
0.5209 13450 0.0 -
0.5228 13500 0.0 -
0.5247 13550 0.0 -
0.5267 13600 0.0 -
0.5286 13650 0.0 -
0.5305 13700 0.0 -
0.5325 13750 0.0 -
0.5344 13800 0.0 -
0.5363 13850 0.0 -
0.5383 13900 0.0 -
0.5402 13950 0.0 -
0.5422 14000 0.0 -
0.5441 14050 0.0 -
0.5460 14100 0.0 -
0.5480 14150 0.0 -
0.5499 14200 0.0 -
0.5518 14250 0.0 -
0.5538 14300 0.0 -
0.5557 14350 0.0 -
0.5576 14400 0.0 -
0.5596 14450 0.0 -
0.5615 14500 0.0 -
0.5635 14550 0.0 -
0.5654 14600 0.0 -
0.5673 14650 0.0 -
0.5693 14700 0.0 -
0.5712 14750 0.0 -
0.5731 14800 0.0 -
0.5751 14850 0.0 -
0.5770 14900 0.0 -
0.5789 14950 0.0 -
0.5809 15000 0.0 -
0.5828 15050 0.0 -
0.5848 15100 0.0 -
0.5867 15150 0.0 -
0.5886 15200 0.0 -
0.5906 15250 0.0 -
0.5925 15300 0.0 -
0.5944 15350 0.0 -
0.5964 15400 0.0 -
0.5983 15450 0.0 -
0.6002 15500 0.0 -
0.6022 15550 0.0 -
0.6041 15600 0.0 -
0.6060 15650 0.0 -
0.6080 15700 0.0 -
0.6099 15750 0.0 -
0.6119 15800 0.0 -
0.6138 15850 0.0 -
0.6157 15900 0.0 -
0.6177 15950 0.0 -
0.6196 16000 0.0 -
0.6215 16050 0.0 -
0.6235 16100 0.0 -
0.6254 16150 0.0002 -
0.6273 16200 0.0 -
0.6293 16250 0.0002 -
0.6312 16300 0.0034 -
0.6332 16350 0.0062 -
0.6351 16400 0.0034 -
0.6370 16450 0.0001 -
0.6390 16500 0.0004 -
0.6409 16550 0.0 -
0.6428 16600 0.0 -
0.6448 16650 0.0 -
0.6467 16700 0.0 -
0.6486 16750 0.0 -
0.6506 16800 0.0 -
0.6525 16850 0.0 -
0.6545 16900 0.0 -
0.6564 16950 0.0 -
0.6583 17000 0.0 -
0.6603 17050 0.0 -
0.6622 17100 0.0 -
0.6641 17150 0.0 -
0.6661 17200 0.0 -
0.6680 17250 0.0 -
0.6699 17300 0.0 -
0.6719 17350 0.0 -
0.6738 17400 0.0 -
0.6758 17450 0.0 -
0.6777 17500 0.0 -
0.6796 17550 0.0 -
0.6816 17600 0.0 -
0.6835 17650 0.0 -
0.6854 17700 0.0 -
0.6874 17750 0.0 -
0.6893 17800 0.0 -
0.6912 17850 0.0 -
0.6932 17900 0.0 -
0.6951 17950 0.0 -
0.6971 18000 0.0 -
0.6990 18050 0.0 -
0.7009 18100 0.0 -
0.7029 18150 0.0 -
0.7048 18200 0.0 -
0.7067 18250 0.0 -
0.7087 18300 0.0 -
0.7106 18350 0.0 -
0.7125 18400 0.0 -
0.7145 18450 0.0 -
0.7164 18500 0.0 -
0.7184 18550 0.0 -
0.7203 18600 0.0 -
0.7222 18650 0.0 -
0.7242 18700 0.0 -
0.7261 18750 0.0 -
0.7280 18800 0.0 -
0.7300 18850 0.0 -
0.7319 18900 0.0 -
0.7338 18950 0.0 -
0.7358 19000 0.0 -
0.7377 19050 0.0 -
0.7397 19100 0.0 -
0.7416 19150 0.0 -
0.7435 19200 0.0 -
0.7455 19250 0.0 -
0.7474 19300 0.0 -
0.7493 19350 0.0 -
0.7513 19400 0.0 -
0.7532 19450 0.0 -
0.7551 19500 0.0 -
0.7571 19550 0.0 -
0.7590 19600 0.0 -
0.7609 19650 0.0 -
0.7629 19700 0.0 -
0.7648 19750 0.0 -
0.7668 19800 0.0 -
0.7687 19850 0.0 -
0.7706 19900 0.0 -
0.7726 19950 0.0 -
0.7745 20000 0.0 -
0.7764 20050 0.0 -
0.7784 20100 0.0 -
0.7803 20150 0.0 -
0.7822 20200 0.0 -
0.7842 20250 0.0 -
0.7861 20300 0.0 -
0.7881 20350 0.0 -
0.7900 20400 0.0 -
0.7919 20450 0.0 -
0.7939 20500 0.0 -
0.7958 20550 0.0 -
0.7977 20600 0.0 -
0.7997 20650 0.0 -
0.8016 20700 0.0 -
0.8035 20750 0.0 -
0.8055 20800 0.0 -
0.8074 20850 0.0 -
0.8094 20900 0.0 -
0.8113 20950 0.0 -
0.8132 21000 0.0 -
0.8152 21050 0.0 -
0.8171 21100 0.0 -
0.8190 21150 0.0 -
0.8210 21200 0.0 -
0.8229 21250 0.0 -
0.8248 21300 0.0 -
0.8268 21350 0.0 -
0.8287 21400 0.0 -
0.8307 21450 0.0 -
0.8326 21500 0.0 -
0.8345 21550 0.0 -
0.8365 21600 0.0 -
0.8384 21650 0.0 -
0.8403 21700 0.0 -
0.8423 21750 0.0 -
0.8442 21800 0.0 -
0.8461 21850 0.0 -
0.8481 21900 0.0 -
0.8500 21950 0.0 -
0.8520 22000 0.0 -
0.8539 22050 0.0 -
0.8558 22100 0.0 -
0.8578 22150 0.0 -
0.8597 22200 0.0 -
0.8616 22250 0.0 -
0.8636 22300 0.0 -
0.8655 22350 0.0 -
0.8674 22400 0.0 -
0.8694 22450 0.0 -
0.8713 22500 0.0 -
0.8733 22550 0.0 -
0.8752 22600 0.0 -
0.8771 22650 0.0 -
0.8791 22700 0.0 -
0.8810 22750 0.0 -
0.8829 22800 0.0 -
0.8849 22850 0.0 -
0.8868 22900 0.0 -
0.8887 22950 0.0 -
0.8907 23000 0.0 -
0.8926 23050 0.0 -
0.8946 23100 0.0 -
0.8965 23150 0.0 -
0.8984 23200 0.0 -
0.9004 23250 0.0 -
0.9023 23300 0.0 -
0.9042 23350 0.0 -
0.9062 23400 0.0 -
0.9081 23450 0.0 -
0.9100 23500 0.0 -
0.9120 23550 0.0 -
0.9139 23600 0.0 -
0.9159 23650 0.0 -
0.9178 23700 0.0 -
0.9197 23750 0.0 -
0.9217 23800 0.0 -
0.9236 23850 0.0 -
0.9255 23900 0.0 -
0.9275 23950 0.0 -
0.9294 24000 0.0 -
0.9313 24050 0.0 -
0.9333 24100 0.0 -
0.9352 24150 0.0 -
0.9371 24200 0.0 -
0.9391 24250 0.0 -
0.9410 24300 0.0 -
0.9430 24350 0.0 -
0.9449 24400 0.0 -
0.9468 24450 0.0 -
0.9488 24500 0.0 -
0.9507 24550 0.0 -
0.9526 24600 0.0 -
0.9546 24650 0.0 -
0.9565 24700 0.0 -
0.9584 24750 0.0 -
0.9604 24800 0.0 -
0.9623 24850 0.0 -
0.9643 24900 0.0 -
0.9662 24950 0.0 -
0.9681 25000 0.0 -
0.9701 25050 0.0 -
0.9720 25100 0.0 -
0.9739 25150 0.0 -
0.9759 25200 0.0 -
0.9778 25250 0.0 -
0.9797 25300 0.0 -
0.9817 25350 0.0 -
0.9836 25400 0.0 -
0.9856 25450 0.0 -
0.9875 25500 0.0 -
0.9894 25550 0.0 -
0.9914 25600 0.0 -
0.9933 25650 0.0 -
0.9952 25700 0.0 -
0.9972 25750 0.0 -
0.9991 25800 0.0 -

Framework Versions

  • Python: 3.10.12
  • SetFit: 1.1.0
  • Sentence Transformers: 3.3.1
  • 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
3
Safetensors
Model size
278M 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 cbpuschmann/paraphrase-multilingual-mpnet-klimacoder_v0.8

Base model

sentence-transformers/paraphrase-multilingual-mpnet-base-v2
Finetuned
(57)
this model

Evaluation results

View on Papers With Code