badBART_T7_sampled-H100-110K

This model is a fine-tuned version of facebook/mbart-large-50 on the None dataset. It achieves the following results on the evaluation set:

  • Loss: 0.6117
  • Rouge1: 80.75
  • Rouge2: 28.37
  • Rougel: 78.38
  • Rougelsum: 78.38

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: 1e-05
  • train_batch_size: 16
  • eval_batch_size: 16
  • seed: 42
  • gradient_accumulation_steps: 2
  • total_train_batch_size: 32
  • optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
  • lr_scheduler_type: linear
  • lr_scheduler_warmup_steps: 73
  • num_epochs: 6
  • mixed_precision_training: Native AMP

Training results

Training Loss Epoch Step Validation Loss Rouge1 Rouge2 Rougel Rougelsum
No log 1.0 370 0.7570 75.32 27.03 73.18 73.19
1.7411 2.0 740 0.6397 79.34 27.41 77.12 77.08
0.6184 3.0 1110 0.6107 79.34 28.46 77.09 77.05
0.6184 4.0 1480 0.6073 82.13 28.8 79.86 79.83
0.4851 5.0 1850 0.6078 79.6 28.69 77.24 77.24
0.4154 6.0 2220 0.6117 80.75 28.37 78.38 78.38

Framework versions

  • Transformers 4.45.2
  • Pytorch 2.4.0+cu124
  • Datasets 3.2.0
  • Tokenizers 0.20.3
Downloads last month
14
Safetensors
Model size
611M params
Tensor type
F32
·
Inference Examples
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 damienliccia/badBART_T7_sampled-H100-110K

Finetuned
(134)
this model