Bert Base Cased Finetuned Sst2

🚀 bert-base-cased-finetuned-sst2

This model is a fine - tuned version of [bert - base - cased](https://huggingface.co/bert - base - cased) on the GLUE SST2 dataset, aiming to compare with [google/fnet - base](https://huggingface.co/google/fnet - base) and achieve high accuracy in text classification.

🚀 Quick Start

This model is a fine - tuned version of [bert - base - cased](https://huggingface.co/bert - base - cased) on the GLUE SST2 dataset. It achieves the following results on the evaluation set:

• Loss: 0.3649
• Accuracy: 0.9232

The model was fine - tuned to compare [google/fnet - base](https://huggingface.co/google/fnet - base) as introduced in this paper against [bert - base - cased](https://huggingface.co/bert - base - cased).

📦 Installation

This model is trained using the [run_glue](https://github.com/huggingface/transformers/blob/master/examples/pytorch/text - classification/run_glue.py) script. The following command was used:

#!/usr/bin/bash

python ../run_glue.py \
  --model_name_or_path bert - base - cased \
  --task_name sst2 \
  --do_train \
  --do_eval \
  --max_seq_length 512 \
  --per_device_train_batch_size 16 \
  --learning_rate 2e - 5 \
  --num_train_epochs 3 \
  --output_dir bert - base - cased - finetuned - sst2 \
  --push_to_hub \
  --hub_strategy all_checkpoints \
  --logging_strategy epoch \
  --save_strategy epoch \
  --evaluation_strategy epoch

Training hyperparameters

The following hyperparameters were used during training:

• learning_rate: 2e - 05
• train_batch_size: 16
• eval_batch_size: 8
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon = 1e - 08
• lr_scheduler_type: linear
• num_epochs: 3.0

Training results

Framework versions

• Transformers 4.11.0.dev0
• Pytorch 1.9.0
• Datasets 1.12.1
• Tokenizers 0.10.3

📄 License

This model is released under the Apache 2.0 license.