metadata

library_name: setfit
tags:
  - setfit
  - sentence-transformers
  - text-classification
  - generated_from_setfit_trainer
datasets:
  - hojzas/proj4-all-labs
metrics:
  - accuracy
widget:
  - text: return list(dict.fromkeys(sorted(it)))
  - text: '    perms = all_permutations_substrings(string)\n    result = perms & set(words)\n    return set(i for i in words if i in perms)'
  - text: return [l for i, l in enumerate(it) if i == it.index(l)]
  - text: |2-
          unique_items = set(it)
          return sorted(list(unique_items))
  - text: |2-
          seen = set()
          result = []
          for word in it:
              if word not in seen:
                  result.append(word)
                  seen.add(word)
          return result
pipeline_tag: text-classification
inference: true
co2_eq_emissions:
  emissions: 6.0133985248367114
  source: codecarbon
  training_type: fine-tuning
  on_cloud: false
  cpu_model: Intel(R) Xeon(R) Silver 4314 CPU @ 2.40GHz
  ram_total_size: 251.49161911010742
  hours_used: 0.019
  hardware_used: 4 x NVIDIA RTX A5000
base_model: sentence-transformers/all-mpnet-base-v2

SetFit with sentence-transformers/all-mpnet-base-v2

This is a SetFit model trained on the hojzas/proj4-all-labs dataset that can be used for Text Classification. This SetFit model uses sentence-transformers/all-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:

Fine-tuning a Sentence Transformer with contrastive learning.
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-mpnet-base-v2
Classification head: a LogisticRegression instance
Maximum Sequence Length: 384 tokens
Number of Classes: 7 classes
Training Dataset: hojzas/proj4-all-labs

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	" perms = all_permutations_substrings(string)\n return set(''.join(perm) for word in words for perm in perms if word == perm)" ' perms = all_permutations_substrings(string)\n out = set()\n for w in words:\n for s in perms:\n if w == s:\n out.add(w)\n return out' ' perms = all_permutations_substrings(string)\n return set(word for word in words if word in perms)'
1	' perms = all_permutations_substrings(string)\n return perms.intersection(words)' ' perms = all_permutations_substrings(string)\n return set.intersection(perms,words)' ' perms = all_permutations_substrings(string)\n return set(perms).intersection(words)'
3	' it = list(dict.fromkeys(it))\n it.sort()\n return it' ' sequence = []\n for i in it:\n if i in sequence:\n pass\n else:\n sequence.append(i)\n sequence.sort()\n return sequence' ' unique = list(set(it))\n unique.sort()\n return unique'
2	'return sorted(list({word : it.count(word) for (word) in set(it)}.keys())) ' 'return list(dict.fromkeys(sorted(it)))' 'return sorted((list(dict.fromkeys(it)))) '
4	' unique_items = set(it)\n return sorted(list(unique_items))' ' letters = set(it)\n sorted_letters = sorted(letters)\n return sorted_letters' 'return list(sorted(set(it)))'
5	' outputSequence = []\n for input in it:\n found = 0\n for output in outputSequence:\n if output == input:\n found = 1\n break\n if not found:\n outputSequence.append(input)\n return outputSequence' ' uniq = []\n for char in it:\n if not char in uniq:\n uniq.append(char)\n return uniq' 'return sorted(set(it), key=lambda y: it.index(y)) '
6	'return [tmp for tmp in dict.fromkeys(it).keys()]' 'return [i for i in dict.fromkeys(it)]' 'return list(dict.fromkeys(it))'

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("hojzas/proj4-all-labs")
# Run inference
preds = model("return list(dict.fromkeys(sorted(it)))")

Training Details

Training Set Metrics

Training set	Min	Median	Max
Word count	2	25.0515	140

Label	Training Sample Count
0	35
1	14
2	8
3	10
4	9
5	13
6	8

Training Hyperparameters

batch_size: (16, 16)
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
0.0041	1	0.1745	-
0.2058	50	0.0355	-
0.4115	100	0.0168	-
0.6173	150	0.0042	-
0.8230	200	0.0075	-

Environmental Impact

Carbon emissions were measured using CodeCarbon.

Carbon Emitted: 0.006 kg of CO2
Hours Used: 0.019 hours

Training Hardware

On Cloud: No
GPU Model: 4 x NVIDIA RTX A5000
CPU Model: Intel(R) Xeon(R) Silver 4314 CPU @ 2.40GHz
RAM Size: 251.49 GB

Framework Versions

Python: 3.10.12
SetFit: 1.0.3
Sentence Transformers: 2.2.2
Transformers: 4.36.1
PyTorch: 2.1.2+cu121
Datasets: 2.14.7
Tokenizers: 0.15.1

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}
}