README.md

metadata

language:
  - fr
tags:
  - classification
license: apache-2.0
datasets:
  - >-
    the model was trained on a small corpus of 140 sentences for training, 100
    for validation and 40 sentences for the test.
metrics:
  - accuracy
  - f1-score

camembert-fr-covid-tweet-classification

This model is a fine-tune checkpoint of Yanzhu/bertweetfr-base, fine-tuned on SST-2. This model reaches an accuracy of 66.0 on the dev set (for comparison, Bert bert-base-uncased version reaches an accuracy of 92.7).

Fine-tuning hyper-parameters

• learning_rate = 1e-5
• batch_size = 32
• warmup = 600
• max_seq_length = 128
• num_train_epochs = 3.0