Hugging Face's logo

allenai
/
led-base-16384-ms2

Text2Text Generation Transformers PyTorch Safetensors
English led generated_from_trainer Inference Endpoints
Model card Files Community
14
led-base-16384-ms2 / README.md
johngiorgi's picture
johngiorgi
Update README.md
6ac0e49
preview code
|
raw history blame
No virus
1.83 kB
metadata
tags:
  - generated_from_trainer
datasets:
  - allenai/mslr2022
model-index:
  - name: baseline
    results: []

Overview

This model is a fine-tuned version of allenai/led-base-16384 on the allenai/mslr2022 ms2 dataset. It achieves the following results on the evaluation set:

  • eval_loss: 3.7527
  • eval_rouge1_fmeasure_mean: 27.9314
  • eval_rouge2_fmeasure_mean: 9.4000
  • eval_rougeL_fmeasure_mean: 20.9302
  • eval_rougeLsum_fmeasure_mean: 23.6179
  • eval_bertscore_hashcode: microsoft/deberta-xlarge-mnli_L40_no-idf_version=0.3.11(hug_trans=4.21.0.dev0)-rescaled_fast-tokenizer
  • eval_bertscore_f1_mean: 23.5092
  • eval_seed: 42
  • eval_model_name_or_path: output/ms2/led-base/baseline
  • eval_doc_sep_token:
  • eval_runtime: 820.6405
  • eval_samples_per_second: 2.463
  • eval_steps_per_second: 0.617
  • step: 0

Model description

More information needed

Intended uses & limitations

More information needed

Training and evaluation data

More information needed

Training procedure

Training hyperparameters

The following hyperparameters were used during training:

  • learning_rate: 2e-05
  • train_batch_size: 4
  • eval_batch_size: 4
  • seed: 42
  • gradient_accumulation_steps: 8
  • total_train_batch_size: 32
  • optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
  • lr_scheduler_type: linear
  • lr_scheduler_warmup_ratio: 0.1
  • num_epochs: 10
  • mixed_precision_training: Native AMP
  • label_smoothing_factor: 0.1

Framework versions

  • Transformers 4.21.0.dev0
  • Pytorch 1.10.0
  • Datasets 2.3.3.dev0
  • Tokenizers 0.12.1